Citation for published version (APA): Bouman, C. S. C. (2007). Continuous venovenous hemofiltration in the critically ill

Transcription

1 UvA-DARE (Digital Academic Repository) Continuous venovenous hemofiltration in the critically ill Bouman, C.S.C. Link to publication Citation for published version (APA): Bouman, C. S. C. (2007). Continuous venovenous hemofiltration in the critically ill General rights It is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons). Disclaimer/Complaints regulations If you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, stating your reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Ask the Library: or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam, The Netherlands. You will be contacted as soon as possible. UvA-DARE is a service provided by the library of the University of Amsterdam ( Download date: 28 Jun 2018

2 Continuous venovenous hemofiltration in the critically ill Catherine Sabine Chantal Bouman

3 Bouman, Catherine Sabine Chantal Continuous venovenous hemofiltration in the critically ill, University of Amsterdam Financial support: Cover: Lay-out: Printer: B.Braun Medical, Baxter, Dirinco, Eli Lilly, Arrow, Hamilton, Dutch Kidney Foundation (Nierstichting Nederland). Le phare s allume au port de Locquirec Anne Bouman-Roulier (Nanou) Legatron Electronic Publishing, Rotterdam PrintPartners IPSKAMP, Enschede ( ISBN-10: ISBN-13: Copyright 2006 C.S.C Bouman, Amsterdam, The Netherlands

4 Continuous venovenous hemofiltration in the critically ill ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad van doctor aan de Universiteit van Amsterdam op gezag van de Rector Magnificus prof. mr. P.F. van der Heijden ten overstaan van een door het college voor promoties ingestelde commissie, in het openbaar te verdedigen in de Aula der Universiteit op woensdag 10 januari 2007, te uur door Catherine Sabine Chantal Bouman geboren te Wenen

5 Promotiecommissie: Promotor: Prof. dr M.B. Vroom Co-promotores: Dr M.J. Schultz Dr H.M. Oudemans-van Straaten Overige leden: Prof. dr A.B.J. Groeneveld Prof. dr J. Kesecioglu Prof. dr R.J. de Haan Prof. dr R.T. Krediet Prof. dr M.M. Levi Prof. dr T. Van der Poll Faculteit der Geneeskunde

6 Un sourire ne coûte rien et produit beaucoup Il enrichit ceux qui le reçoivent Sans appauvrir ceux qui le donnent Il ne dure qu un instant Mais son souvenir est parfois éternel Pour Balou et Nanou

7

8 Contents Chapter 1 General introduction and outline of the thesis 9 Chapter 2 Guidelines for timing, dose, and mode of continuous renal 17 replacement therapy for acute renal failure in the critically ill Neth J Crit Care 2006;10(5); Chapter 3 Effects of early high-volume continuous venovenous 41 hemofiltration on survival and recovery of renal function in intensive care patients with acute renal failure a prospective randomized trial Crit Care Med 2002;30: Chapter 4 The effects of continuous venovenous hemofiltration on 59 coagulation activation Crit Care 2006;10:R150 Chapter 5 Predilution versus postdilution during continuous venovenous 75 hemofiltration: a comparison of circuit thrombogenesis ASAIO Journal 2006;52: Chapter 6 Discrepancies between observed and predicted CVVH removal 93 of antimicrobial agents in critically ill patients and the effects on dosing Intensive Care Med 2006; Oct 17 (Epub ahead of print) Chapter 7 Cystatin C in critically ill patients treated with continuous 109 venovenous hemofiltration Hemodialysis International 2006;10:S11-S15 Chapter 8 Cytokine filtration and adsorption during predilution and 119 postdilution hemofiltration in four different membranes Blood Purif 1998;16:

9 Chapter 9 Hemofiltration in sepsis and SIRS: 133 The role of dosing and timing Journal Crit Care (in press) Chapter 10 Summary 157 Chapter 11 Samenvatting 163 Dankwoord 169 Curriculum vitae 173 Abbreviations 175

10 Chapter 1 General introduction and outline of the thesis

11 Chapter 1 Renal replacement therapy for ARF in the critically ill Acute renal failure is a common complication of critical illness and carries a high mortality [1]. A recent worldwide survey showed that RRT therapy is required in 4% of the patients admitted to the ICU [1]. Mortality in ARF is often higher than predicted according to the commonly used scores [2]. Controversy exists about whether this excess mortality of critically ill patients with ARF is merely a reflection of the severity of the underlying disease or whether it can additionally be attributed to the acutely uremic state itself. ARF in ICU care rarely presents as an isolated organ dysfunction, but rather as a component of the multiple organ dysfunction syndrome, following a broad spectrum of diseases, such as severe sepsis, pancreatitis or cardiogenic shock. Thus, ARF is certainly an indicator of the severity of disease. Increasing evidence, however, suggests an independent effect of ARF on mortality in critically ill patients [2-5]. Unfortunately there is still no consensus on the optimal management of ARF. This is even more strongly the case for the application of RRT. The optimal mode, dose and timing of RRT are a matter of debate. Every effort should be made to prevent (further) injury to the kidney and some studies suggest that the type of RRT may contribute to the recovery of renal function [1,6]. Although the importance of adequacy of dialysis is widely recognized in patients with end stage renal failure, much less attention has been paid to the concept of adequacy of dialysis in critically ill patients with ARF. In the ICU, renal replacement therapies are primarily limited to IHD and CRRT, including CVVH. During IHD, intensive dialysis is performed for a few hours at variable intervals, whereas during CRRT, treatment at lower blood flow rates is applied for 24 h per day (or as near as possible) providing gradual removal of fluid and toxins. In the United States IHD is currently the predominant mode, while CRRT is more widely used in Australia and Europe [6-9]. The continuous techniques offer certain advantages over the more traditional IHD, including better hemodynamic stability, less restriction of fluid delivery, and better control of electrolytes and uremia [10-12]. CVVH and (anti)coagulation Clotting of the extracorporeal system can significantly impair the efficiency of CVVH. Partial clotting of the hemofilter decreases ultrafiltration rates and reduces solute clearance. Eventually, occlusion of the entire extracorporeal circuit requires its replacement, leading to significant loss of blood and treatment time. Changing the system also incurs significant costs related to supplies and nursing time. Anticoagulants are frequently used to prolong circuit life during CVVH. Critically ill patients with ARF, 10

12 Introduction however, often have complex coagulation disorders and an increased risk of bleeding. Therefore changes in hemostasis during CVVH may adversely influence the prognosis of the patient [13,14]. Understanding the mechanisms involved in premature clotting of the filtration circuit is useful to optimize anticoagulation and maintain filter patency. CVVH and antimicrobial treatment Many critically ill patients with ARF have serious infections and require treatment with one or more antimicrobial drugs. The importance of optimization of antimicrobial drug dose in these patients has been highlighted recently [15]. While underdosing may lead to treatment failures and the selection of resistant micro-organisms, overdosing may cause drug toxicities and higher costs. Some drugs are removed substantially by CVVH while others are not and an adjusted dose is required to prevent under or overdosing of the substance [16]. In comparison with data on antibiotic dosing in patients undergoing IHD there is little information on optimal antibiotic dosing during CVVH in critically ill patients. CVVH and estimation of residual renal function Other than diuresis, at present there is no measure of residual renal function in a patient on CVVH, because serum concentrations and urine outputs of urea and creatinine and other calculated renal function tests are influenced by the method itself. Cystatin C has been introduced as a new marker of glomerular filtration [17,18]. CysC is a 13 kda, non-glycosylated basic protein, produced at a constant rate by all nucleated cells. It is freely filtered by the renal glomeruli and catabolized in tubuli. The serum concentration is independent of age, gender and muscle mass. Cystatin C may be a reliable marker to estimate residual renal function during CVVH, however little is known about potential removal of CysC during CVVH. CVVH and sepsis The use of CVVH in critically ill patients with ARF is widely accepted. CVVH is also applied for some non-renal indications, including sepsis and SIRS, but these indications are less well established. The sepsis syndrome is associated with an overwhelming systemic overflow of both proinflammatory and antiinflammatory mediators; this leads 11

13 Chapter 1 to altered immune cellular responsiveness, generalized endothelial damage and multiple organ failure derived from a complete disruption of the immunological homeostasis [19,20]. Although our understanding of the complex pathophysiological alterations that occur in severe sepsis and septic shock has increased greatly as a result of clinical and preclinical studies, mortality associated with the disorder remains unacceptably high, ranging from 30 to 50%. Beneficial effects of hemofiltration in sepsis have been reported in animals and humans [21,22]. The group of Ronco et al. proposed the peak concentration hypothesis as the mechanism by which hemofiltration in sepsis could be beneficial [23]. This concept refers to the ability of hemofiltration to lower peaks of the proinflammatory and antiinflammatory mediators, reducing their toxic effects and creating a new equilibrium. The application of hemofiltration as adjunctive therapy in sepsis, is, however, still a subject of much debate [24]. Outline of the thesis This thesis addresses various aspects of CVVH in the critically ill with or without ARF. Chapter 2 reviews the available literature on CRRT and provides evidenced based recommendations for timing, treatment dose and mode of RRT in the critically ill patient with ARF. Chapter 3 presents the results of a randomized controlled two-center study on the effects of dose and initiation time of CVVH on clinical outcomes in ventilated patients with circulatory failure developing early ARF. Chapter 4 shows the results of a prospective clinical study on the effects of CVVH without therapeutic anticoagulation on activation of coagulation. This study investigates whether CVVH causes activation of the contact factor pathway or the tissue factor pathway of coagulation. Chapter 5 presents the results of a prospective clinical study comparing the effects of predilution CVVH with postdilution CVVH on circuit thrombogenesis. Chapter 6 presents the results of a prospective clinical study comparing the observed and predicted CVVH clearances of seven antimicrobial agents. In addition it investigates whether dose adjustment according to the predicted CVVH clearance provides an as reliable estimate than that accoridng to the observed CVVH removal. 12

14 Introduction Chapter 7 reports the results of a study on measuring the removal of CysC during CVVH and discusses whether CVVH is likely to affect serum CysC levels. Chapter 8 presents the results of an in vitro hemofiltration study on the convective and adsorptive removal of TNF, IL-6 and IL-8. Four different membranes are investigated both in the predilution and postdilution mode. Chapter 9 reviews the available literature on hemofiltration in sepsis and SIRS focusing on the role of initiation time of hemofiltration and ultrafiltrate dose. Chapter 10 summarizes the results of the above mentioned studies. 13

15 Chapter 1 References Uchino S, Kellum JA, Bellomo R, Doig GS, et al. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA 2005;294(7): Metnitz PG, Krenn CG, Steltzer H, et al. Effect of acute renal failure requiring renal replacement therapy on outcome in critically ill patients. Crit Care Med 2002;30(9): Chertow GM, Levy EM, Hammermeister KE, et al. Independent association between acute renal failure and mortality following cardiac surgery. Am J Med 1998;104(4): de Mendonca A, Vincent JL, Suter PM, et al. Acute renal failure in the ICU: risk factors and outcome evaluated by the SOFA score. Intensive Care Med 2000;26(7): Levy EM, Viscoli CM, Horwitz RI. The effect of acute renal failure on mortality. A cohort analysis. JAMA 1996;275(19): Mehta RL, Letteri JM. Current status of renal replacement therapy for acute renal failure. A survey of US nephrologists. The National Kidney Foundation Council on Dialysis. Am J Nephrol 1999;19(3): Cole L, Bellomo R, Silvester W, et al. A prospective, multicenter study of the epidemiology, management, and outcome of severe acute renal failure in a «closed» ICU system. Am J Respir Crit Care Med. 2000;162(1): Liano F, Pascual J. Epidemiology of acute renal failure: a prospective, multicenter, community-based study. Madrid Acute Renal Failure Study Group. Kidney Int 1996;50(3): Oudemans-van straaten HM, Wester JPJ. Resultaten van de enquete naar de praktijk van nierfunctievervangende behandeling op de intensive care in Nederland. Neth J Crit Care 2002;6:18-9. Augustine JJ, Sandy D, Seifert TH, et al. A randomized controlled trial comparing intermittent with continuous dialysis in patients with ARF. Am J Kidney Dis 2004;44(6): Davenport A, Will EJ, Davison AM, et al. Changes in intracranial pressure during machine and continuous haemofiltration. Int J Artif Organs 1989;12(7): John S, Griesbach D, Baumgartel M, et al. Effects of continuous haemofiltration vs intermittent haemodialysis on systemic haemodynamics and splanchnic regional perfusion in septic shock patients: a prospective, randomized clinical trial. Nephrol Dial Transplant 2001;16(2):

16 Introduction Garcia-Fernandez N, Montes R, Purroy A, et al Rocha E. Hemostatic disturbances in patients with systemic inflammatory response syndrome (SIRS) and associated acute renal failure (ARF). Thromb Res 2000;100(1): Schetz MR. Coagulation disorders in acute renal failure. Kidney Int. Suppl 1998;66: S Mehrotra R, De Gaudio R, Palazzo M. Antibiotic pharmacokinetic and pharmacodynamic considerations in critical illness. Intensive Care Med 2004;30(12): Shah M, Quigley R. Rapid removal of vancomycin by continuous veno-venous hemofiltration. Pediatr Nephrol 2000;14(10-11): Coll E, Botey A, Alvarez L, et al. Serum cystatin C as a new marker for noninvasive estimation of glomerular filtration rate and as a marker for early renal impairment. Am J Kidney Dis 2000;36(1): Hoek FJ, Kemperman FA, Krediet RT. A comparison between cystatin C, plasma creatinine and the Cockcroft and Gault formula for the estimation of glomerular filtration rate. Nephrol Dial Transplant 2003;18(10): Bone RC. Important new findings in sepsis. JAMA 1997;278(3):249. Hotchkiss RS, Karl IE. The pathophysiology and treatment of sepsis. N Engl J Med 2003;348(2): Grootendorst AF, van Bommel EF, van der Hoven B, et al. High volume hemofiltration improves right ventricular function in endotoxin-induced shock in the pig. Intensive Care Med 1992;18(4): Honore PM, Jamez J, Wauthier M, et al. Prospective evaluation of short-term, highvolume isovolemic hemofiltration on the hemodynamic course and outcome in patients with intractable circulatory failure resulting from septic shock. Crit Care Med 2000;28(11): Ronco C, Tetta C, Mariano F, et al. Interpreting the mechanisms of continuous renal replacement therapy in sepsis: the peak concentration hypothesis. Artif Organs 2003; 27(9): Reiter K, Bellomo R, Ronco C, et al. Pro/con clinical debate: is high-volume hemofiltration beneficial in the treatment of septic shock? Crit Care 2002;6(1):

17

18 Chapter 2 Guidelines for timing, dose, and mode of continuous renal replacement therapy for acute renal failure in the critically ill Catherine S.C. Bouman, Heleen M. Oudemans-van Straaten, On behalf of the committee of nephrology and intensive care of the NVIC and the committee of quality of the NFN (See appendix) Neth J Crit Care 2006;10(5);

19 Chapter 2 Abstract Objective To provide evidence-based recommendations for clinical practice on timing, dose, and mode of CRRT in critically ill patients with ARF admitted to the ICU. Methods Literature search was done in Pubmed database for human studies. Studies were rated at five levels to create recommendations grades from A to E, grade A being the highest. Conclusions In critically ill patients with ARF it is recommended: to define ARF according to the RIFLE classification system into ARF risk, ARF injury and ARF failure. to base the decision when to start RRT not only on the severity of ARF, but also on the severity of other organ failure (grade E). Initiation of RRT is to be considered in oliguric patients (RIFLE risk-oliguria or RIFLE injury-oliguria ), despite adequate fluid resuscitation, and/or a persisting steep rise in serum creatinine, in addition to persisting shock (grade E). RRT may be postponed when the underlying disease is improving, other organ failure recovering and the slope in the serum creatinine rise declines, in order to see if renal function is also recovering (grade E). to continue RRT as long as the criteria defining severe oliguric ARF (RIFLE failure-oliguria ) are present (grade E). If the clinical condition improves, it may be considered to wait before connecting a new circuit to see whether renal function recovers. RRT should be restarted in case of clinical or metabolic deterioration. to achieve a delivered (not prescribed) ultrafiltrate (dialysate) flow during CVVH(D) of 35 ml/kg/h in postdilution (grade A). A higher dose applied for a short period may be considered in sepsis/sirs (grade E). The dose needs to be adjusted for predilution using the dilution factor, and for filter down time. In non-shock patients, continuous and intermittent treatments are equivalent regarding survival (level I). However, CRRT is recommended over IHD for patients with ARF who have, or are at risk for, cerebral edema (grade C). CRRT is preferred in the management of patients with ARF and shock (grade E). CRRT should be applied in the venovenous mode (grade B). HF in patients with sepsis or SIRS without ARF is not supported by enough evidence to be recommended in daily clinical practice. 18

20 Guidelines CRRT Introduction Up to now, there are no standard guidelines for the application of CRRT in critically ill patients with ARF. Practice patterns vary widely between individual centers [1,2]. CRRT for the critically ill patient with ARF was introduced in 1977 by Kramer et al. [3]. Since then, many studies have reported on CRRT in the critically ill, but for several reasons comparison among studies is difficult: Various treatment modalities have been applied in heterogeneous populations that differ not only in co-morbidities, but also in the clinical setting and underlying molecular biological mechanisms that initiate and maintain ARF. Furthermore, there are more than 35 definitions of ARF [4]. Recently a process of international consensus and evidence-based statements in the definition and management of ARF was proposed by the ADQI [5,6]. Aim of the present contribution is to provide evidence based recommendations for clinical practice on the timing, treatment dose, and mode of CRRT in critically ill patients admitted to the ICU. Anticoagulation strategies, substitution fluids, membranes and non-renal indications are beyond the scope of the present paper. Consensus definition of acute renal failure Figure 1 summarizes the ADQI consensus criteria for ARF [6]. ARF is classified into three levels: ARF risk, ARF injury, and ARF failure, based on glomerular filtration rate or urine output criteria, whichever is more severe. Methods Studies were identified using the MeSH terms acute kidney failure OR acute renal failure OR shock combined with the words hemofiltration OR haemofiltration OR hemodialysis OR haemodialysis in PubMed, from 1984 until March 2006, and by scanning the lists of publications found by database searches and on the ADQI workgroup findings at www. ADQI.net. Searches were limited to adult human studies and English language. The identified studies were eligible if they fulfilled the following criteria: (a) critically ill patients with ARF or SIRS, and (b) RRT with specified treatment characteristics including at least mode, dose and/or timing. We excluded studies on CAVH, molecular adsorbent techniques, and plasmapheresis. We also excluded studies applying hemofiltration during cardiac surgery and in patients with cardiac failure, because these studies specifically focus on the beneficial effects of fluid removal. We classified evidence and formulated 19

21 Chapter 2 recommendations according to evidence based medicine methodology (Table 1) [7]. Criteria for the qualification level I and level II with respect to the size of the RCT are not well settled. In the present review, we defined level I studies to be those including at least 50 patients per randomized group. Table 1. The guidelines of evidence-based medicine s rating system for strength of recommendation and quality of evidence [7]. Rating system for references Level I Level II Level III Level IV Level V Large randomized trials with clear-cut results; low false positive ( ) or false negative ( ) error. Meta-analysis with low false positive ( ) or false negative ( ) error. Small, randomized trials with uncertain results; high false positive ( ) or negative ( ) error. Meta-analysis with high false positive ( ) or false negative ( ) error. Nonrandomized, contemporaneous control. Nonrandomized, historical controls. Case series, uncontrolled studies and expert opinion. Rating system for recommendations Grade A Grade B Grade C Grade D Grade E Supported by at least two level I investigations. Supported by only one level I investigation. Supported by level II investigations only. Supported by at least one level III investigation. Supported by level IV or level V investigations only. Timing There is significant variation in the timing of initiation of RRT, with up to twofold differences in the reported values of BUN, creatinine, or urine output at RRT initiation [8-11]. There are two RCTs [12,13], four non-randomized studies [14-17] and one observational study [18], investigating the effect of timing of RRT on mortality, and/ or recovery of renal function in critically ill patients with documented ARF or in patients with sepsis/sirs and imminent ARF (Table 2). 20

22 Guidelines CRRT GFR criteria Urine Output Criteria Risk Increased Screat x 1.5 or GFR decrease >25% UO < 0.5 ml/kg/h x 6 h Injury Increased Screat x2 or GFR decrease >25% UO < 0.5 ml/kg/h x 12 h High Sensitivity Failure Increased Screat x3 GFR decrease 75% OR Screat 350 μmol/l Acute rise 44 μmol/l UO < 0.3 ml/kg/h x 24 h or Anuria x 12 h O liguria Loss Persistent ARF = complete loss of kidney function > 4 weeks High Specificity ESKD End Stage Kidney Disease (> 3 months) Figure 1: The RIFLE (Risk of renal dysfunction, Injury to the kidney, Failure of kidney function, Loss of kidney function and End-stage kidney disease) classification for acute renal failure [6]. (With approval of ADQI). The classification system includes separate criteria for creatinine and urine output (UO). A patient can fulfill the criteria through changes in serum creatinine (SCreat) or changes in UO, or both. The criteria that lead to the worst possible classification should be used. Note that the F component of RIFLE is present even if the increase in SCreat is under threefold as long as the new SCreat is greater than 350 mol/l in the setting of an acute increase of at least 44 mol/l). The designation RIFLE-F C should be used in this case to denote acute-on-chronic disease. Similarly, when the RIFLE-F classification is achieved by UO criteria, a designation of RIFLE-FO should be used to denote oliguria. The shape of the figure denotes the fact that more patients (high sensitivity) will be included in the mild category, including some without actually having renal failure (less specificity). In contrast, at the bottom of the figure the criteria are strict and therefore specific, but some patients will be missed. GFR, Glomerular Filtration Rate; ARF, Acute Renal Failure 21

23 Chapter 2 Table 2. Clinical studies evaluating the timing of initiation of CRRT in critically ill patients. Study design [no. patients] Clinical setting Definition of timing Confounding CRRT factors Survival advantage early group Level of evidence Bouman [12] RCT (105) oliguric ARF and MOF Early: < 12 h after onset oliguria (< 180 ml in 6 h) Late: urea > 40 mmol/l or severe pulmonary edema a) no no II Jiang [13] RCT (37) Severe pancreatitis Early: < 48 h after onset abdominal pain. Late: > 96 h after onset abdominal pain no yes II Gettings [16] Retrospective (100) Post trauma Early: urea of < 60 mg/dl b) Late: urea of 60 mg/dl Dose not reported yes III Piccini [17] Retrospective (80) Sepsis with oliguric ARF and ALI Early: < 12 h after ICU admission. Late: urea > 35 mmol/l, scr > 600 μmol/l Dose early >> dose late yes IV Elahi [15] Retrospective (64) ARF after cardiac surgery Early: oliguria (<100 ml in 8 h) Late: urea > 30 mmol/l, scr > 250 μmol/l Dose not reported yes IV Demirkilic [14] Retrospective [61] ARF after cardiac surgery Early: oliguria (<100 ml in 8 h) Dose not reported yes IV Late: SCr > 5 mg/dl c) Honore [18] Cohort (20) Severe septic shock Early: refractory septic shock HF, 35 L in 4 h yes IV CRRT continuous renal replacement therapy; ARF, Acute renal failure; ALI, acute lung injury; scr, serum creatinine; HF, hemofiltration; a) po 2 /FiO 2 < 150 mm Hg and 10 PEEP cm H 2 O; b) 21 mmol/l; c) 420 μmol/l. 22

24 Guidelines CRRT Acute renal failure 1. In a two-center RCT (n = 106), in critically ill patients with oliguric ARF (diuresis of < 180 ml in 6 hours, despite fluid resuscitation, inotropic support and high-dose diuretics), 28-day survival and recovery of renal function were not increased when CVVH was started early (within 12 hours after the onset of oliguria) as compared to late (urea of > 40 mmol/l, and/or severe pulmonary edema with PO 2 /FiO 2 of < 150 mm Hg and 10 PEEP cm H 2 O) (level II) [12]. Of note, in this study, late was not as late as in earlier studies. Because of pulmonary reasons, nearly half of the patients in the late group started CVVH before serum urea reached 40 mmol/l. Median delay between the start of treatment and the development of oliguria was 42 hours. 2. A single-center, retrospective, non-randomized cohort study (n = 100) in trauma patients, used BUN as a surrogate of timing of intervention [16]. Survival was 39% in the early group (RRT started at a mean BUN of 42.6 mg/dl (15 mmol/l) compared with 20% in the late group (RRT started at a mean BUN of 94.5 mg/dl (34 mmol/l) (level III). However, this approach is likely to be seriously flawed, because BUN may reflect many factors other than time of initiation. 3. In a single-center retrospective cohort study in cardiac surgery patients, hospital mortality was higher in the late CVVH group (n = 28) compared with the early CVVH group (n = 36) (43% vs 22%, p <.05) (level IV) [15]. In the late group, CVVH was started on conventional reasons (urea of 30 mmol/l, creatinine of 250 μmol/l, or potassium of 6.0 mmol/l despite glucose-insulin infusion), regardless of urine output. In the early group, CVVH was started when urine output was < 100 ml within 8 hours, despite furosemide infusion. 4. In a single-center retrospective study in patients with ARF following cardiac surgery hospital mortality decreased after the introduction of early CVVHDF compared with a historical control group (23.5% vs 55%, p = 0.02) [14]. In the early group (n = 34), CVVHDF was started for oliguria (urine output of < 100 ml within 8 hours), and in the late group (n = 27), CVVHDF was started for conventional criteria (creatinine of > 444 μmol/l) (level IV). Sepsis or SIRS 5. In a small RCT (n = 37), in patients with severe pancreatitis without documented ARF, early CVVH (within 48 h after onset of abdominal pain) improved hemodynamics and short-term survival, compared with late CVVH (96 h after onset abdominal pain) (level II) [13]. 6. In a single-center retrospective study (n = 80) in patients with septic shock and oliguric ARF, the application of early CVVH improved hemodynamics, gas exchange, 23

25 Chapter 2 7. successful weaning, and 28-day survival compared with a historical control group receiving conventional therapy (level IV) [17]. However, only 75% of the patients in the conventional group received CVVH, despite overt renal failure, and the applied dose was lower (20 ml/kg/h) than in the early treatment group (mean daily dose ml/kg/h). In the cohort study by Honoré et al., in patients with refractory septic shock, post hoc analysis showed an association between increased survival and earlier start of hemofiltration (level V) [18]. Discontinuation RRT There are no clinical data on stopping criteria for RRT in critically ill patients with (recovering) ARF. Discussion In the above-mentioned studies there is a clear trend toward a better outcome with earlier timing of RRT. However, one small RCT did not confirm this trend. In the absence of large RCTs comparing early to late initiation of RRT, no firm overall recommendations for timing of RRT can be made. When initiation of RRT is considered, it is important to realize that the consequences of uremic toxicity, metabolic acidosis and/or fluid overload are likely to be more severe in the critically ill patient. Moreover, renal function is unlikely to recover within a short period during persistent and severe failure of other organs. Furthermore, various inflammatory mediators are cleared by the kidney. Treatment dose The importance of adequacy of dialysis is widely recognized in patients with ESRD; however, much less attention has been paid to the concept of adequacy of dialysis in critically ill patients with ARF [19]. In IHD, dose is generally expressed as Kt/V [20], where K = clearance, t = treatment duration and V = the volume of distribution. In ESRD a minimum Kt/V of 1.2 thrice weekly should be delivered, a lower dose is associated with higher mortality [19]. However, higher doses may be beneficial in critically ill patients with ARF. In CRRT, treatment dose is generally expressed as filtrate volume/kg per time, for pure convective transport with postfilter replacement, and as Kt/V for other modalities. To calculate the treatment dose for predilution HF, the recommended ultrafiltrate rate should be multiplied by the dilution [21]. It is to be emphasized that dose quantification in acute RRT is not thoroughly validated and associated with numerous problems [21;22]. Recenly, single pool Kt/V appeared to be a useful way 24

26 Guidelines CRRT to prescribe dose for different modalities of CRRT [23]. Moreover, dose estimates do not take into account differences between the pore size of membranes and mode. The middle molecular clearance is better when high cutoff membranes are compared with low cutoff membranes, and when hemofiltration is compared with hemodialysis. Furthermore, the removal of middle molecules declines when membranes are used for longer periods. There are at present six RCTs (one applying IHD) [12,24-27], and one retrospective study [28], on the effect of renal replacement dose on mortality and recovery of renal function and/or physiologic endpoints, in critically ill patients with ARF (Table 3). After the first observations of Gotloib et al. [29] on the beneficial effects of hemofiltration in the septic acute respiratory distress syndrome, four RCTs [13,30-32] and four observational cohort studies [18,33-36] evaluated the effects of dose of RRT in patients with SIRS without documented ARF. Acute renal failure 1. In a RCT in 146 critically ill patients with ARF, survival (14 days after the last IHD session) was significantly higher in the patients treated with daily IHD compared with alternate day IHD [27] (level I). Patients with hepatorenal syndrome or cardiogenic shock were excluded from the study and treated with CRRT. Patient characteristics were comparable between groups. Daily IHD resulted in a better control of uremia, fewer IHD related hypotension, and faster resolution of ARF, compared with alternate day IHD. In a multiple regression analysis, less frequent IHD was an independent risk factor for death. Unfortunately, although the prescribed dose of dialysis was 3.6 Kt/V per week in the alternate day group, the delivered dose was far less (about 3.0). All the surviving patients, except the two with Goodpasture s disease, had full recovery of renal function. 2. A positive association between survival time and ultrafiltrate dose was also described in a large (n = 425) RCT in patients with multiple organ failure and ARF treated with CVVH [26] (level 1). Small, but significant differences were present for age, APACHE II score, and BUN levels at baseline. Survival, 15 days after discontinuation of CVVH, was significantly lower in the group receiving 20 ml/kg/h (41%), compared with the higher volume groups receiving 35 ml/kg/h (57%) and group 45 ml/kg/h (58%). The difference in the duration of CVVH, and the rate of renal recovery were not significantly different among the survivors of the three groups. 3. In a RCT in 206 critically ill patients with ARF, 28-day survival was significantly increased in the group receiving a higher replacement dose by adding a dialysis dose to CVVH [32]. Renal recovery rate among survivors was comparable between the high dose CVVHDF group and the low dose CVVH group. 25

27 Chapter An association between survival time and ultrafiltrate dose was not found in a smaller RCT (n = 106) in critically ill, ventilated patients with shock and oliguric ARF (level II) [12]. The patients were randomized into three groups: early high-volume hemofiltration (EHV, L/24h), early low-volume hemofiltration (ELV, L/24h), and late low-volume hemofiltration (LLV, L/24h). Early treatment started within 12 hours after the onset of oliguria, and late when the patient fulfilled the conventional criteria for RRT (as in paragraph on timing). The 28-day survival was 74.3% in EHV, 68.8% in ELV and 75% in LLV (p = 0.80). All hospital survivors had recovery of renal function. In a RCT in 70 patients with ARF secondary to severe malaria or sepsis, the risk of death was higher in the group receiving peritoneal dialysis (70 L/day) compared with the group receiving CVVH (25 L/day) [25] (level II). The estimated Kt/V per week of 5.5 in the CVVH group was comparable to the low intensity groups in the studies of Ronco et al. [26] and Bouman et al. [12]. Unfortunately, the authors did not report the measurements necessary to calculate effective Kt/V in the peritoneal dialysis, but we can speculate that it was lower than in the CVVH group because the peritoneal dialysis group had a lower rate of resolution of acidosis and a slower rate of decline in plasma creatinine levels. In a crossover study that compared high-volume (6 L/h) with low-volume CVVH (1 L/h) in 11 septic shock patients with ARF, the dose of norepinephrine required for the maintenance of target MAP decreased more during high-volume CVVH than during low-volume CVVH (p = 0.02) (level II) [24]. In a non-randomized prospective interventional pilot study (n = 56), Brause et al. [28] compared very low-volume CVVH (1 L/h), with low-volume CVVH (1.5 L/h), and assessed the effect on the daily Kt/V. As expected the 1.5 L/h group achieved a higher Kt/V (0.80 per day versus 0.53 per day) and better control of uremia and acid base (level III). Mortality was high in both groups (73% and 69%, p = NS), but the study was not powered for survival as an endpoint. Sepsis or SIRS 1. In a small (n = 24) RCT in patients with early septic shock or organ dysfunction, CVVH at 2 L/h did not affect clinical outcome compared with no CVVH (level II) [30]. The study was not powered for survival as an endpoint. 2. In a small RCT in 37 patients with severe pancreatitis, hemodynamics and short term survival rate improved more during high-volume CVVH (4 L/h) compared to low-volume CVVH (1 L/h) (Level II) [13]. The study was not powered for survival as an endpoint. 26

28 Guidelines CRRT In a RCT in 61 patients after cardiac arrest, very high-volume HF (100 L in 8 hours) with, or without hypothermia significantly increased 6-months survival compared with standard care (level II) [31]. In a large observational study (n=306) in critically ill patients receiving CVVH (100 L/day) mortality was significantly lower (33%) than predicted by APACHE II (76%) and SAPS II (71%) illness severity scores [34]. Improved heamodynamics and increased survival were also reported in four smaller cohort studies (level IV) in: 5. Patients with intractable septic shock (n = 20) treated with short-term very highvolume HF (35 L in 4 hours) [18]. 6. Patients with septic shock (n = 24) treated with high-volume CVVH (40 60 ml/kg/h) [33]. 7. Patients with severe sepsis treated with pulse very high-volume HF (85 ml/kg/h for 6-8 hours) [35]. 8. Patients with severe septic shock treated with short-term very high-volume HF (100 ml/kg/h for 12 h) [36]. Discussion In some of the above-mentioned studies, Kt/V in the low-volume groups was extremely low, even lower than the Kt/V currently recommended for chronic dialysis [25,27,28] and nearly as low as the dose in earlier CAVH studies yielding a mortality rate of 80%. From the foregoing, it can be concluded that delivered RRT dose should not be too low. The highest evidence indicates a recommended dose of at least ml/kg/h for CVVH(D) [26,32] and daily sessions for IHD [27]. The 35 ml/kg/h dose corresponds to a single pool Kt/V of 1.4 per day [23]. In contrast, a smaller RCT, suggests that 1.5 L/h (~ 20 ml/kg/h) is as good as 4 L/h (~48 ml/kg/h) [12]. The differences in outcome of the randomized studies may result from differences in case mix, ICU format, membrane, substitution fluid or concomitant treatment [37]. There are three multicenter RCTs underway looking at dose of RRT in ARF: The Acute Renal Failure Trial Network (ATN) Study in the US run by Palevsky [38], The Renal Study in Australia and New Zealand run by Bellomo [39], and the IVOIRE study in Europe run by Joannes-Boyau [40]. It is to be emphasized that in daily clinical practice the prescribed ultrafiltrate flow should be adjusted, in order to achieve the intended delivered ultrafiltrate flow. Evidence for a beneficial effect of (short-term) high, or very-high volume in patients with SIRS/sepsis and imminent or no ARF is still low. The studies are not randomized or underpowered for survival [13,31]. Low-volume (2 L/h) CVVH [30] seems to have no positive effects in patients with sepsis/sirs and imminent ARF (level II). 27

29 Chapter 2 Table 3. Comparison of randomized controlled trials on the effect of renal replacement dose on mortality and recovery of renal function. Study [Ref] Randomization (number of patients) Mean Delivered dose Survival (%) p ARF in days ml/kg/h Kt/V per week (mean) Day 14 after end IHD p Evidence Level Schiffl [27] Alternate day IHD (72) Daily IHD (74) I Day 15 after end CVVH Ronco [26] CVVH 20 ml/kg/h (146) CVVH 35 ml/kg/h (139) CVVH 45 ml/kg/h (140) N.S. I Day 28 after inclusion Saudan [32] CVVH 25 ml/kg/h (102) CVVHDF 42 ml/kg/h (104) Not reported I Day 28 after inclusion Bouman [12] ELV 1,5 L/h (35) LLV 1,5 L/h (35) EHV 4 L/h (36) ,8 8,6 11,6 8,6.55 II ICU survival Phu [25] PD (36) CVVH 25 L/day (34) << 5,5 5, Not reported II Day 14 after start CVVH Jiang [13] ELV 1 L/h (9) LLV 1 L/h (10) EHV 4 L/h (9) LHV 4 L/h (9) Not reported HV 68 LV 89 E 84 L 74 < 0.01 < 0.05 Not reported II IHD, intermittent hemodialysis; CVVH, continuous venovenous hemofiltration; CVVHDF, continuous venovenous hemodiafiltration; ELV, early low-volume hemofiltration, LLV, late low-volume hemofiltration; EHV, early high volume hemofiltration; LHV, late high-volume hemofiltration; Kt/V, clearance times duration of treatment divided by volume of distribution; ARF acute renal failure; HV, high volume; LV, low volume; E. early; L, late. 28

30 Guidelines CRRT Modes of acute renal replacement therapy In the ICU, renal replacement therapies are primarily limited to conventional IHD and CRRT. During IHD, intensive dialysis is performed for 3 4 hours at variable intervals, whereas during CRRT, continuous and gradual removal of fluid and toxins is provided at lower blood flow. More recently several hybrid therapies [41] have been described, with a treatment duration between CRRT and conventional IHD, (ie extended dialysis [42], sustained low-efficiency dialysis [43], short-term HF [18] or pulse HF [35]. The nomenclature and definitions of the various CRRT techniques are based on their operational characteristics [44] (Table 4). Hemodialysis and hemofiltration are the two main principles of solute transport of CRRT. During h;emodialysis, removal of solutes is driven by diffusion (solute transport across a semi-permeable membrane generated by a concentration gradient). During hemofiltration, removal of solutes is based on convection (water and solute transport across a semi-permeable membrane generated by a pressure gradient). There are no data showing any given modality as superior with regard to clinical outcomes. Hemofiltration resembles most the principle of glomerular filtration and increases the middle molecule clearances [45]; however, whether this is beneficial in ARF is unknown. Factors that may affect current practice include local availability of equipment, fluids and costs. CRRT is applied either in the arteriovenous (driving force is patient s blood pressure and flow) or venovenous mode (driving force is external pump). Advantages of the arteriovenous therapies include ease of set-up and operation and low extracorporeal blood volumes. Disadvantages of arteriovenous therapies include the prolonged arterial cannulation, the requirement of a MAP of > 60 mm Hg to maintain circuit blood flow, and the low blood flows that can be achieved. Advantages of the venovenous therapies are the decreased risk of vascular damage as compared to the arteriovenous therapies, the ability to maintain blood flow independent of MAP, the ability to achieve higher blood flow rates and clearances (level III) [46,47]. The higher clearances associated with better survival [26] cannot be achieved without the introduction of a blood pump. The use of blood pumps has increased the complexity of CRRT systems, but in clinical practice this disadvantage does not counterbalance the advantages, and there is general consensus that venovenous systems are the modality of choice [46-49]. 29

31 Chapter 2 Table 4. Modes of CRRT. Solute transport Blood flow (ml/min) Ultrafiltrate flow (ml/min) Dialysate flow (ml/min) Clearance (L/24h) Slow continuous ultrafiltration Continuous arteriovenous or venovenous hemofiltration (CAVH or CVVH) Continuous arteriovenous or venovenous hemodialysis (CAVHD or CVVHD) No No No Convection No Diffusion Continuous arteriovenous hemodiafiltration or venovenous hemodiafiltration (CAVHDF or CVVHDF) Continuous arteriovenous or venovenous high flux dialysis (HDF) Convection and diffusion Convection and diffusion CRRT vs conventional IHD One of the most pressing clinical questions regarding the use of CRRT is whether CRRT offers an important advantage over IHD, regarding survival and/or recovery of renal function. The effects of IHD versus CRRT on survival and/or recovery of renal function were reported in five prospective RCTs [50-54] and two meta-analyses [55,56] In a large multicenter RCT (n = 160) CVVHDF showed no survival (ICU and hospital) advantage compared with alternate day IHD after adjustment for severity of illness (level I) [52]. However, CRRT was associated with a significantly higher rate of complete renal recovery in surviving patients who received an adequate trial of therapy, without crossover to IHD (CRRT 92.3% vs IHD 59.4%, p <.01). Of notice, in this study patients were excluded when MAP was < 70 mm Hg in the 8 hours preceding randomization. Furthermore, significant baseline differences in severity of illness existed between groups and the delivered dialysis dose per group was not reported, making comparison difficult. In a large multicenter RCT (n = 224) septic patients were randomized to receive either IHD or CVVHDF with the same polyacrilonitrile membrane and bicarbonate buffer [54]. The 60-day survival was 23,5% in the CVVHDF group and 28,6% in the IHD group (p = 0.23) (level I) 30

32 Guidelines CRRT In a single-center RCT (n = 125) patients were randomized into CVVHDF or daily IHD treatment [53]. IHD was started gently with a low blood flow and small hemofilter and removing small amounts of fluid, to avoid hemodynamic instability. The treatment doses were comparable between groups. Hospital mortality was 47% in the CVVHDF group and 51% in the IHD group (p = 0.72). Unfortunately, the study was underpowered due to the pre-terminal end and the smaller than expected number of patients included (level II). A single-center RCT (n = 80) that compared CVVHD with alternate day IHD showed no survival or renal recovery advantages between groups, despite a significant decrease in MAP for patients on IHD therapy not seen in those on CVVHD therapy (level II) [50]. However, the study was not sufficiently powered for survival as an endpoint. A single-center RCT (n = 104) showed no differences in survival or MAP between patients receiving CVVH and patients treated with daily IHD (level II) [51]. Again, this study was not adequately powered to detect small differences between modalities. Furthermore, the majority of patients (n = 33) in the CVVH group were treated with low-volume CVVH (18 ml/kg/h) and this may also have adversely affected the outcome in the CVVH group. Kellum et al. [55] performed a metaanalysis, including 13 studies (n = 1400) comparing CRRT with IHD, and did not find a statistically significant impact of dialysis modality on survival and renal recovery in haemodynamic stable patients (level I). Tonelli et al. [56] included six studies in their metaanalysis (n = 624) and concluded that in unselected critically ill patients with ARF, CRRT does not improve survival or renal recovery (level I). In a large (n = 839) prospective, multicenter cohort study mortality was comparable between the patients undergoing IHD and the patients undergoing CRRT, however patients undergoing IHD had lower Logistic Organ Dysfunction Scores (level III) [57]. Likewise, in another large (n = 587) observational prospective multicenter study RRT was not found to have any prognostic value [58]; however, patients selected for CRRT had a higher number of organ dysfunction at admission and at the time of ARF (level III). Two smaller observational studies [59,60] reported improved survival with CRRT, even though CRRT patients were sicker at baseline (level III). Two retrospective studies [61,62] in critically ill patients with ARF showed comparable mortality between the IHD group and the CRRT group, but patients with severe illness were preferentially selected for CRRT (level IV). 31

33 Chapter 2 Discussion None of the level I or level II studies showed a survival advantage for CRRT as compared with conventional IHD [50-56]. However, the largest RCT [52] suggest that CRRT is associated with increased complete renal recovery (level I). Although most of the studies did not report on the delivered treatment dose per group, none of the studies seem to have achieved the higher dose associated with a better survival in the CRRT studies [26,32]. The study of Mehta et al. [52], suggests that there is a physician s bias for CRRT being the treatment of choice for patients in shock and this was also suggested in numerous prospective observational and retrospective studies [57-62]. Indeed, beneficial effects on cardiovascular stability, cerebral edema and intestinal acidosis have been reported during CRRT therapy in comparison with conventional IHD [50;63-67] (level II). On the other hand, the study of Uehlinger et al. [53], suggests that hemodynamic instability during IHD can be avoided even in unstable patients, as long as gentle IHD is applied (daily sessions using low blood flow, small surface filter and discrete fluid removal). Final recommendations Because of the quality of the studies recommendation grades are low. Comparison among the studies is complicated by the use of various definitions of ARF. Furthermore, strategies of timing, dose and RRT mode are likely to interact. However, most of the studies only investigate one of these items and do not report on the others, making it difficult to draw firm conclusions. The below recommendations concern critically ill patients with ARF It is recommended to define ARF according to the RIFLE classification system into ARF risk, ARF injury and ARF failure. It is recommended to base the decision when to start RRT not only on the severity of ARF, but also on the severity of other organ failure (Grade E). Initiation of RRT is to be considered in oliguric patients (RIFLE risk-oliguria or RIFLE ) injury-oliguria, despite adequate fluid resuscitation, and/or a persisting steep rise in serum creatinine, in addition to persisting shock (Grade E). RRT may be postponed when the underlying disease is improving, other organ failure recovering and the slope in the serum creatinine rise declines, in order to see if renal function is also recovering (Grade E). It is recommended to continue RRT as long as the criteria defining severe oliguric ARF (RIFLE failure-oliguria ) are present (grade E). If the clinical condition improves, 32

34 Guidelines CRRT it may be considered to wait before connecting a new circuit to see whether renal function recovers. RRT should be restarted in case of clinical or metabolic deterioration. The recommended delivered (not prescribed) ultrafiltrate (dialysate) flow during CVVH(D) is 35 ml/kg/h in postdilution (Grade A). A higher dose applied for a short period may be considered in sepsis/sirs (grade E). The dose needs to be adjusted for predilution using the dilution factor, and for filter down time. In non-shock patients, continuous and intermittent treatments are equivalent regarding survival (level I). However, CRRT is recommended over IHD for patients with ARF who have, or are at risk for, cerebral edema (Grade C). CRRT is preferred in the management of patients with ARF and shock (Grade E). CRRT should be applied in the venovenous mode (Grade B). HF in patients with sepsis or SIRS without ARF is not supported by enough evidence to be recommended in daily clinical practice. Samenvatting aanbevelingen De aanbevelingen hebben betrekking op de ernstig zieke IC patiënt met acute nierinsufficiëntie. Het advies is om acute nierinsufficiëntie volgens het RIFLE classificatie system te definiëren in de categorieen ARF risk, ARF injury and ARF failure. Het advies is om de beslissing om met nierfunctie vervangende therapie (NVT) te beginnen niet alleen te laten afhangen van de ernst van het acute nierfalen maar ook van de ernst van het overig orgaanfalen (niveau E). Starten van RRT kan worden overwogen bij oligure patiënten (RIFLE risk-oliguria of RIFLE injury- oliguria ), en/of bij patiënten met een aanhoudende snelle stijging in het kreatinine gehalte in combinatie met aanhoudende shock (niveau E). Uitstel van NVT kan worden overwogen indien de onderliggende ziekte aan het verbeteren is, overig orgaan falen herstellende en het kreatinine gehalte aan het aftoppen is, om te zien of de nierfunctie ook herstellende is (niveau E). Het advies is om de NVT voort te zetten zolang er voldaan wordt aan de ernstige oligurie criteria (RIFLE failure-oliguria ) (niveau E). Men kan overwegen het aansluiten van een nieuw circuit uit te stellen indien de klinische conditie aan het verbeteren is om te beoordelen of de nierfunctie ook aan het herstellen is. Het advies is om tijdens CVVH(D) in postdilutie daadwerkelijk een ultrafiltraat (dialysaat) flow van 35 ml/kg/h te geven (niveau A). Bij sepsis/sirs kan men 33

35 Chapter 2 overwegen gedurende korte tijd te behandelen met een hogere ultrafiltraat flow (Niveau E). De dosis moet worden gecorrigeerd voor predilutie met de verdunnings factor en voor de uren dat de filtratie (dialyse) niet loopt. Bij patiënten zonder shock is geen verschil in overleving aangetoond tussen continue en intermitterende behandeling (level I). Echter, bij patiënten met hersenoedeem of een verhoogd risico hierop wordt CRRT aanbevolen (niveau C). Continue behandelingen verdienen de voorkeur bij patiënten met shock (niveau E). Voor continue behandelingen moeten venoveneuze technieken worden toegepast (niveau B). Er bestaat onvoldoende bewijs om hemofiltratie te adviseren bij patiënten met sepsis of SIRS zonder acute nierinsufficiëntie. Appendix Committee of nephrology and intensive care of the Dutch Society of Intensive Care (NVIC): Heleen M. Oudemans-van Straaten, Catherine C.S. Bouman, Anne-Cornelie J.M. de Pont, A.B. Johan Groeneveld, Miet Schetz, Arend Jan Woittiez. Committee of quality of the Dutch Federation of Nephrology (NFN): Robert Zietse, Jeroen Kooman, Coen A. Stegeman. 34

36 Guidelines CRRT References Oudemans-van Straaten H, Wester J. Resultaten van de enquete naar de praktijk van nierfunctievervangende behandeling op de intensive care in Nederland. Neth J Crit Care 2002;6: Uchino S, Kellum JA, Bellomo R,et al. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA 2005;294: Kramer P, Wigger W, Rieger J, et al. Arteriovenous haemofiltration: a new and simple method for treatment of over-hydrated patients resistant to diuretics. Klin Wochenschr 1977;55: Kellum JA, Levin N, Bouman C, et al. Developing a consensus classification system for acute renal failure. Curr Opin Crit Care 2002;8: Ronco C, Kellum JA, Mehta R. Acute dialysis quality initiative (ADQI). Nephrol Dial Transplant 2001;16: Bellomo R, Ronco C, Kellum JA, et al. Acute renal failure - definition, outcome measures, animal models, fluid therapy and information technology needs: the Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 2004;8:R204-R212. Damen J, van Diejen D, Bakker J, et al. NVIC-standpunten, NVIC-richtlijnen en de juridische implicaties. Neth J Crit Care 2002;6: Ricci Z, Ronco C, D Amico G, et al. Practice patterns in the management of acute renal failure in the critically ill patient: an international survey. Nephrol Dial Transplant 2006;21: Brivet FG, Kleinknecht DJ, Loirat P, et al. Acute renal failure in intensive care units-causes, outcome, and prognostic factors of hospital mortality; a prospective, multicenter study. French Study Group on Acute Renal Failure. Crit Care Med 1996;24: Cole L, Bellomo R, Silvester W, et al. A prospective, multicenter study of the epidemiology, management, and outcome of severe acute renal failure in a closed ICU system. Am J Respir Crit Care Med 2000;162: Himmelfarb J, Tolkoff RN, Chandran P, et al. A multicenter comparison of dialysis membranes in the treatment of acute renal failure requiring dialysis. J Am Soc Nephrol 1998;9: Bouman CS, Oudemans-van Straaten HM, Tijssen JG, et al. Effects of early high-volume continuous venovenous hemofiltration on survival and recovery of renal function in intensive care patients with acute renal failure: a prospective, randomized trial. Crit Care Med 2002;30:

37 Chapter Jiang HL, Xue WJ, Li DQ, et al. Influence of continuous veno-venous hemofiltration on the course of acute pancreatitis. World J Gastroenterol 2005;11: Demirkilic U, Kuralay E, Yenicesu M, et al. Timing of replacement therapy for acute renal failure after cardiac surgery. J Card Surg 2004;19: Elahi MM, Lim MY, Joseph RN, et al. Early hemofiltration improves survival in postcardiotomy patients with acute renal failure. Eur J Cardiothorac Surg 2004;26: Gettings LG, Reynolds HN, Scalea T. Outcome in post-traumatic acute renal failure when continuous renal replacement therapy is applied early vs. late. Intensive Care Med 1999;25: Piccinni P, Dan M, Barbacini S, et al. Early isovolaemic haemofiltration in oliguric patients with septic shock. Intensive Care Med 2006;32: Honore PM, Jamez J, Wauthier M, et al. Prospective evaluation of short-term, highvolume isovolemic hemofiltration on the hemodynamic course and outcome in patients with intractable circulatory failure resulting from septic shock. Crit Care Med 2000;28: NKF-DOQI clinical practice guidelines for hemodialysis adequacy. National Kidney Foundation. Am J Kidney Dis 1997;30:S15-S66. Gotch FA, Sargent JA, Keen ML. Whither goest Kt/V? Kidney Int Suppl 2000;76:S3-18. Marshall MR. Current status of dosing and quantification of acute renal replacement therapy. Part 2: dosing paradigms and clinical implementation. Nephrology 2006;11: Marshall MR. Current status of dosing and quantification of acute renal replacement therapy. Part 1: mechanisms and consequences of therapy under-delivery. Nephrology 2006;11: Ricci Z, Salvatori G, Bonello M, et al. In vivo validation of the adequacy calculator for continuous renal replacement therapies. Crit Care 2005;9:R266-R273. Cole L, Bellomo R, Journois D, et al. High-volume haemofiltration in human septic shock. Intensive Care Med 2001;27: Phu NH, Hien TT, Mai NT, et al. Hemofiltration and peritoneal dialysis in infectionassociated acute renal failure in Vietnam. N Engl J Med 2002;347: Ronco C, Bellomo R, Homel P, et al. Effects of different doses in continuous venovenous haemofiltration on outcomes of acute renal failure: a prospective randomised trial. Lancet 2000;356: Schiffl H, Lang SM, Fischer R. Daily hemodialysis and the outcome of acute renal failure. N Engl J Med 2002;346:

38 Guidelines CRRT Brause M, Neumann A, Schumacher T, et al. Effect of filtration volume of continuous venovenous hemofiltration in the treatment of patients with acute renal failure in intensive care units. Crit Care Med 2003;31: Gotloib L, Barzilay E, Shustak A, et al. Hemofiltration in septic ARDS. The artificial kidney as an artificial endocrine lung. Resuscitation 1986;13: Cole L, Bellomo R, Hart G, et al. A phase II randomized, controlled trial of continuous hemofiltration in sepsis. Crit Care Med 2002;30: Laurent I, Adrie C, Vinsonneau C, et al. High-volume hemofiltration after out-ofhospital cardiac arrest: a randomized study. J Am Coll Cardiol 2005;46: Saudan P, Niederberger M, De Seigneux S, et al. Martin PY: Adding a dialysis dose to continuous hemofiltration increases survival in patients with acute renal failure. Kidney Int 2006;70: Joannes-Boyau O, Rapaport S, Bazin R, et al. Impact of high volume hemofiltration on hemodynamic disturbance and outcome during septic shock. ASAIO J 2004;50: Oudemans-van Straaten HM, Bosman RJ, van der Spoel JI, et al. Outcome of critically ill patients treated with intermittent high-volume haemofiltration: a prospective cohort analysis. Intensive Care Med 1999;25: Ratanarat R, Brendolan A, Piccinni P, et al. Pulse high-volume haemofiltration for treatment of severe sepsis: effects on hemodynamics and survival. Crit Care 2005;9: R294-R302. Cornejo R, Downey P, Castro R, et al. High-volume hemofiltration as salvage therapy in severe hyperdynamic septic shock. Intensive Care Med 2006;32: Oudemans-vanStraaten HM, Bouman CS, Zandstra DF. Survival in acute renal failure. Intensive Care Med 2005;31: Palevsky P. the Renal Failure Trial Network (ATN) study. ATNStudy org. Accessed on August Bellomo R. Augmented versus normal renal replacement therapy in severe acute renal failure. gov/ct/show/nct ?order=3. Accessed on August Joannes-Boyau O: the IVOIRE study. gov/ct/show/nct ?order=1. Accessed on August Marshall MR, Golper TA, Shaver MJ, et al. Hybrid renal replacement modalities for the critically ill. Contrib Nephrol 2001; Kumar VA, Yeun JY, Depner TA, et al. Extended daily dialysis vs. continuous hemodialysis for ICU patients with acute renal failure: a two-year single center report. Int J Artif Organs 2004;27:

39 Chapter Marshall MR, Golper TA, Shaver MJ, et al. Sustained low-efficiency dialysis for critically ill patients requiring renal replacement therapy. Kidney Int 2001;60: Ronco C, Bellomo R. Basic mechanisms and definitions for continuous renal replacement therapies. Int J Artif Organs 1996;19: Ricci Z, Ronco C, Bachetoni A, et al. Solute removal during continuous renal replacement therapy in critically ill patients: convection versus diffusion. Crit Care 2006;10:R67. Canaud B, Garred LJ, Christol JP, et al. Pump assisted continuous venovenous hemofiltration for treating acute uremia. Kidney Int Suppl 1988;24:S154-S156. Tam PY, Huraib S, Mahan B, et al. Slow continuous hemodialysis for the management of complicated acute renal failure in an intensive care unit. Clin Nephrol 1988;30: Macias WL, Mueller BA, Scarim SK, et al. Continuous venovenous hemofiltration: an alternative to continuous arteriovenous hemofiltration and hemodiafiltration in acute renal failure. Am J Kidney Dis 1991;18: Storck M, Hartl WH, Zimmerer E, et al. Comparison of pump-driven and spontaneous continuous haemofiltration in postoperative acute renal failure. Lancet 1991;337: Augustine JJ, Sandy D, Seifert TH, et al. A randomized controlled trial comparing intermittent with continuous dialysis in patients with ARF. Am J Kidney Dis 2004;44: Gasparovic V, Filipovic-Grcic I, Merkler M, et al. Continuous renal replacement therapy (CRRT) or intermittent hemodialysis (IHD) - what is the procedure of choice in critically ill patients? Ren Fail 2003;25: Mehta RL, McDonald B, Gabbai FB, et al. A randomized clinical trial of continuous versus intermittent dialysis for acute renal failure. Kidney Int 2001;60: Uehlinger DE, Jakob SM, Ferrari P, et al. Comparison of continuous and intermittent renal replacement therapy for acute renal failure. Nephrol Dial Transplant 2005;20: Vinsonneau C, Camus C, Combes A, et al. Continuous venovenous haemodiafiltration versus intermittent haemodialysis for acute renal failure in patients with multipleorgan dysfunction syndrome: a multicentre randomised trial. Lancet 2006;368: Kellum JA, Angus DC, Johnson JP, et al. Continuous versus intermittent renal replacement therapy: a meta-analysis. Intensive Care Med 2002;28: Tonelli M, Manns B, Feller-Kopman D. Acute renal failure in the intensive care unit: a systematic review of the impact of dialytic modality on mortality and renal recovery. Am J Kidney Dis 2002;40:

40 Guidelines CRRT Metnitz PG, Krenn CG, Steltzer H, et al. Effect of acute renal failure requiring renal replacement therapy on outcome in critically ill patients. Crit Care Med 2002;30: Guerin C, Girard R, Selli JM, et al. Intermittent versus continuous renal replacement therapy for acute renal failure in intensive care units: results from a multicenter prospective epidemiological survey. Intensive Care Med 2002;28: Bellomo R, Farmer M, Parkin G, et al. Severe acute renal failure: a comparison of acute continuous hemodiafiltration and conventional dialytic therapy. Nephron 1995;71: Kruczynski K, Irvine-Bird K, Toffelmire EB, et al. A comparison of continuous arteriovenous hemofiltration and intermittent hemodialysis in acute renal failure patients in the intensive care unit. ASAIO J 1993;39:M778-M781. Gangji AS, Rabbat CG, Margetts PJ. Benefit of continuous renal replacement therapy in subgroups of acutely ill patients: a retrospective analysis. Clin Nephrol 2005;63: van Bommel E, Bouvy ND, So KL, et al. Acute dialytic support for the critically ill: intermittent hemodialysis versus continuous arteriovenous hemodiafiltration. Am J Nephrol 1995;15: Davenport A, Will EJ, Davison AM, et al. Changes in intracranial pressure during machine and continuous haemofiltration. Int J Artif Organs 1989;12: Davenport A, Will EJ, Davison AM, et al. Changes in intracranial pressure during haemofiltration in oliguric patients with grade IV hepatic encephalopathy. Nephron 1989;53: Ronco C, Bellomo R, Brendolan A, et al. Brain density changes during renal replacement in critically ill patients with acute renal failure. Continuous hemofiltration versus intermittent hemodialysis. J Nephrol 1999;12: Van der SG, Diltoer M, Laureys M, et al. Intermittent hemodialysis in critically ill patients with multiple organ dysfunction syndrome is associated with intestinal intramucosal acidosis. Intensive Care Med 1996;22: John S, Griesbach D, Baumgartel M, et al. Effects of continuous haemofiltration vs intermittent haemodialysis on systemic haemodynamics and splanchnic regional perfusion in septic shock patients: a prospective, randomized clinical trial. Nephrol Dial Transplant 2001;16:

41

42 Chapter 3 Effects of early high-volume continuous venovenous hemofiltration on survival and recovery of renal function in intensive care patients with acute renal failure a prospective randomized trial Catherine S.C. Bouman, Heleen M Oudemans-van Straaten, Jan G.P. Tijssen, Durk F Zandstra, Jozef Kesecioglu Crit Care Med 2002;30:

43 Chapter 3 Abstract Objective To study the effects of initiation time of CVVH and of ultrafiltrate rate, in patients with circulatory and respiratory insufficiency developing early oliguric ARF. The primary end points were mortality at 28 days and recovery of renal function. Design A randomized controlled two-center study. Patients and interventions A total of 106 ventilated severely ill patients who were oliguric despite massive fluid resuscitation, inotropic support and high dose intravenous diuretics, were randomized into three groups. Thirty five patients were treated with early high-volume hemofiltration (EHV) (72 96 L/24h), 35 patients with early low-volume hemofiltration (ELV) (24 36 L/24h) and 36 patients with late low-volume hemofiltration (LLV) (24 36 L/24h). Results Median ultrafiltrate rate was 48.2 ( ) ml/kg/h in EHV, 20.1 ( ) ml/kg/h in ELV and 19.0 ( ) ml/kg/h in LLV. Survival at day 28 was 74.3% in EHV, 68.8% in ELV and 75.0% in LLV (p = 0.80). On average, hemofiltration started 7 h after inclusion in the early groups and 42 h after inclusion in the late group. All hospital survivors had recovery of renal function at hospital discharge, except for one patient in ELV. Median duration of renal failure in hospital survivors was 4.3 ( ) days in EHV, 3.2 ( ) days in ELV and 5.6 ( ) days in LLV (p = 0.25). Conclusions In the present study of critically ill patients with oliguric ARF, survival at 28 days and recovery of renal function were not improved using high ultrafiltrate volumes, or early initiation of hemofiltration. 42

44 Early high-volume CVVH Introduction Oliguric ARF is a frequent complication in patients with septic or cardiogenic shock. Pending recovery of renal function, temporary RRT is required in most cases. In daily practice there is substantial variation in the policies regarding initiation time of RRT and in the way it is performed. Apart from IHD, other techniques including peritoneal dialysis, CAVH(D) and CVVH(D) are being used. It is generally accepted that in ICU patients with ARF the continuous techniques are superior to IHD, in particular with respect to hemodynamic stability [1,2]. Despite the implementation of continuous techniques, patient outcome is still very poor. Most studies of ARF in ICU patients reported mortality between 60 and 80% [3-6]. Low clearance techniques were used in these studies and renal replacement was started late in the course of renal failure. Nonrandomized studies suggest that both earlier initiation of RRT, and the use of higher ultrafiltrate rates, might improve survival and recovery of renal function [7,8]. In a recent prospective randomized study, improvement of survival was reported by increasing the ultrafiltrate rate [9]. The aim of the present study was to evaluate the effects of initiation time of hemofiltration, and of ultrafiltrate rate, in patients with circulatory and respiratory insufficiency and early ARF. The primary end points were 28-day mortality and recovery of renal function. Methods Study design The study was designed as a RCT comparing three treatment strategies: early highvolume hemofiltration (EHV), early low-volume hemofiltration (ELV), and late low-volume hemofiltration (LLV). The Academic Medical Center, a university hospital with a 28-bed closed format multidisciplinary ICU, and the Onze Lieve Vrouwe Gasthuis, a teaching hospital with an 18-bed closed format multidisciplinary ICU participated in the study. Both centers practiced cardiosurgery. The institutional Review Boards of both hospitals approved of the protocol. Patients were eligible if they fulfilled the following criteria: a) urine output of < 30 ml/h for > 6 hours despite aggressive fluid resuscitation (PAOP or CVP of > 2 mm Hg), hemodynamic optimization with dopamine and/or dobutamine (> 5 μg/kg/min), phosphodiesterase inhibitors or norepinephrine in any dose, and the administration of high dose diuretics (> 500 mg of frusemide infusion in 6 h); b) creatinine clearance of < 20 ml/min, calculated from a 3-h urine portion; c) mechanical ventilation; d) age between 18 and 90 yrs; e) intention to provide full intensive treatment for at least three days. Patients were excluded if they fulfilled one of the following criteria: 43

45 Chapter 3 a) preexisting renal disease with creatinine clearance of < 30 ml/min according to the formula of Cockcroft and Gauld [10]; b) ARF caused by permanent occlusion and/ or surgical lesion of renal artery; c) ARF caused by (glomerulo)nephritis, interstitial nephritis, or vasculitis; d) ARF caused by postrenal obstruction; e) CHILD class C liver cirrhosis [11]; f) AIDS with a CD4 count of < 0.05 x 10 9 /L; g) non-witnessed arrest with Glasgow Coma Score < 5; h) hematological malignancy with neutrophiles < 0.05 x 10 9 /L; i) no hemofiltration machine free for use at the moment of inclusion. Very early hemofiltration (before the onset of oliguria) in case of therapy resistant septic or cardiogenic shock was not performed during the study period. Time of inclusion (T incl. ) was defined as the time at which all inclusion criteria were fulfilled. Patients were randomized into one of the three groups: Early high-volume hemofiltration (EHV) Treatment was started within 12 h after T incl.. Blood flow rate was maintained at 200 ml/min and minimal ultrafiltrate production was 72 L/day. Hemofilter and tubing set were changed routinely every 24 h, to prevent decay in membrane permeability and loss of ultrafiltration capacity. Treatment was allowed to be interrupted for a maximum of 12 h between two runs. Early low-volume hemofiltration (ELV) Treatment was started within 12 h after T incl. Blood flow was maintained at 100 to 150 ml/min, and ultrafiltrate production was at least 24 L/day and 36 L/day at the maximum. The hemofilter and tubing set were changed when signs of clotting of the extracorporeal system occurred. Late low-volume hemofiltration (LLV) Treatment was started when the patient fulfilled the conventional reasons for RRT: plasma urea level of > 40 mmol/l, potassium of > 6.5 mmol/l or severe pulmonary edema, defined as CVP or PAOP of > 16 mm Hg and lung edema on radiograph in all quadrants with positive end expiratory pressure of 10 cm H 2 O and PO 2 /FiO 2 ratio of < 150 mm Hg. Blood flow was maintained at 150 ml/min, and ultrafiltrate production was at least 24 L/day but 36 L/day at the maximum. The hemofilter and tubing set were changed when signs of clotting of the extracorporeal system occurred. In all three treatment groups, hemofiltration was allowed to be discontinued when urine output recovered ( 60 ml/h). Treatment was restarted if renal clearance remained insufficient (blood urea level of > 50 mmol/l). When a second period of oliguria occurred the patient remained in the same group. The definite time of recovery was taken as the time of final recovery. 44

46 Early high-volume CVVH Continuous venovenous hemofiltration Hemofiltration was performed with computer controlled, fully automated hemofiltration machines (Diapact, B-Braun, Melsungen, Germany or Haemoprocessor, Sartorius, Boebingen, Germany). Vascular access was obtained by cannulation of the femoral, jugular or subclavian vein using the Seldinger technique and a double-lumen catheter (GamCath, Gambro, Hechingen, Germany). A 1.9 m 2 cellulose triacetate hollow fiber membrane was used (CT-190G, Baxter Healthcare Corporation, IL, USA; UF 205, Nissho Nipro Europe nv, Zaventem, Belgium).The SC from this hemofilter is 0.81 for 2 microglobulin. The extracorporeal circuit was anticoagulated with heparin (Leo Pharma, Ballerup, Denmark) or nadroparin (Fraxiparine, Sanofi Synthelabo, Paris, France). In case of severe contraindications for anticoagulation, hemofiltration was performed without anticoagulation. In case of heparin-induced thrombocytopenia, danaparoid (Orgaran, Organon, Oss, The Netherlands) was used. Warmed bicarbonated substitution fluids were administered in postdilution (SH-bic 35/Combibic 10, B-Braun, Melsungen, Germany), containing 140 mmol/l of Na ++, 2.0 mmol/l of K +, 1.25 mmol/l of Ca ++, 0.75 mmol/l of Mg ++, 111 mmol/l of Cl, 35 mmol/l of HCO 3 and 1.0 g/l of glucose. Measurements The following data were obtained: at ICU admission, the MPM 0, APACHE II score, APACHE III score and the SAPS II score [12-15]; MPM h after ICU admission [12]; time between ICU admission and T incl. ; at T incl. the hemodynamic data, the plasma lactate level, the Goris score, the SOFA and the LODS [16-18]; the duration between T incl. and first hemofiltration treatment; plasma urea level at the beginning of the first hemofiltration treatment; the number of hours receiving hemofiltration; temperature before and after each hemofiltration session; the development of HIT; bleeding complications with transfusion of at least 2 units of packed red cells; survival 28 days after T incl. ; ICU survival; hospital survival; duration of mechanical ventilation from T incl. until extubation; length of ICU stay, defined as the number of days from T incl. until ICU discharge; length of hospitalization, defined as the number of days from T incl. until hospital discharge; duration of renal failure, defined as the number of days from T incl. until the end of the last hemofiltration session, or in case of spontaneous recovery, the number of days between T incl. until recovery of diuresis of 60 ml/h. In case of death before recovery of renal function, duration of renal failure was defined as the number of days from T incl. until death. Study end points The primary end points of the study were survival at day 28 after inclusion and recovery of renal function. Secondary end points consisted of ICU survival, hospital survival, duration of mechanical ventilation, length of ICU stay, and length of hospitalization. 45

47 Chapter 3 Statistical analysis On account of historical controls treated with low- and high-volume hemofiltration [6-8] it was calculated that a sample of 105 patients would give the study 90% power with an overall type I error rate of 0.05 to detect a 40% absolute difference between EHV and LLV [19]. Analysis was done by intention to treat. Data are presented as mean ± SD or median and quartiles. Normally distributed variables were compared using one way analysis of variance Bonferroni s correction for multiple comparisons. For significant findings, post hoc t-tests were applied. Kruskal-Wallis one-way analysis of variance was used to compare non-normally distributed variables. Chi-square testing was used to test frequencies between groups. All testing was two-tailed and p <.05 was considered statistically significant. Statistical analysis was performed with the aid of SPSS for Windows, release 7.5 (SPSS, Chicago, IL, USA). Results During the study period (from May 1998 until March 2000), a total of 372 patients (6.3% of all ICU admissions) received CVVH; 248 patients had oliguric ARF. Of these, 142 did not fulfill the entry criteria for the following reasons: preexisting renal disease with documented clearance of < 30 ml/min (n = 45); referral from another hospital with oliguria existing longer than 18 hours (n = 31), Child C liver cirrhosis (n = 3), AIDS with a low CD4 count (n = 4), nonwitnessed arrest (n = 2), hematological malignancy with bone marrow aplasia (n = 1), no CVVH machine available at T incl. (n = 4), no intention to provide full intensive treatment for > 3 days (n = 25), no informed consent (n = 29). A total of 106 patients were randomized in the study: 35 in the EHV group, 35 in the ELV group and 36 in the LLV group (Figure 1). Patient characteristics Patient characteristics at ICU admission and at T incl. were comparable among the three groups (Table 1). There were no essential differences among patients from all three groups in terms of age, sex distribution, creatinine clearance before ICU admission, severity of illness and duration between ICU admission and inclusion. The ELV group had more cardiosurgical patients (75%) compared to EHV (51.4%) and LLV (50%) groups, but the difference was not statistically significant. Predicted mortality rates are not presented since the percentage of cardiosurgical cases is high. 46

48 Early high-volume CVVH 5919 ICU admissions Oliguric ARF n = 248 Non-oliguric ARF n = 130 Not randomize in study n = 142 Randomize in study n = 106 EHV n = 35 ELV n = 35 LLV n = 36 Hemofiltration n = 372 No hemofiltration n = 6 Figure 1. Trial profile. ICU, intentive care unit; ARF, acute renal failure; CVVH, continuous venovenous hemofiltration; EHV, Early high-volume hemofiltration; ELV, early low-volume hemofiltration; LLV, late low-volume hemofiltration. Hemofiltration treatment All patients in the EHV and ELV groups were hemofiltered. In the LLV group 30 patients were hemofiltered but six patients were not: two patients died before they fulfilled the criteria for starting hemofiltration, and in four patients renal function recovered spontaneously, meaning that diuresis recovered and urea levels never reached 40 mmol/l. The characteristics of the hemofiltration treatments are shown in Table 2. On average, hemofiltration was started within 7 h after T incl. in the early groups and 42 h after T incl. in the late group (p < 0.001). In the LLV group, 15 patients started hemofiltration because of blood urea level of 40 mmol/l and 15 patients started because of severe pulmonary edema, with urea levels of < 40 mmol/l. Median ultrafiltrate clearance, expressed as volume of filtrate per body weight, was approximately 19 ml/kg/h in the low-volume groups and 48 ml/kg/h in the high-volume group (p< 0.001). Median number of hours on hemofiltration was not significantly different among the three groups (p = 0.20). The median filter life span was significantly higher in LLV compared to EHV and ELV 47

49 Chapter 3 (p < 0.01). The percentage of hemofiltration sessions without anticoagulation was 16.5% in EHV, 19.6% in ELV and 16.5% in LLV (p = 0.61). Table 1. Patient characteristics. EHV (n = 35) ELV (n = 35) LLV (n = 36) Age, yrs 68 ± ± ± Male sex % Previous creatinine clearance, ml/min a 67 ± ± ± Type of admission, % Cardiosurgical Postoperative surgical / medical Severity of illness at ICU admission MPM ± ± ± MPM ± ± ± SAPS II 47.7 ± ± ± APACHE II 23.5 ± ± ± APACHE III 85.0 ± ± ± Days between ICU admission and T incl. 1.2 ( ) 1.6 ( ) 1.2 ( ).50 Severity of illness at study inclusion Goris score 6.3 ± ± ± SOFA score 10.3 ± ± ± LODS score 8.0 ± ± ± Dopamine, μg/kg/min 11.9 ± ± ± Norepinephrine, μg/kg/min 0.14 ± ± ± Enoximone, μg/kg/min 1.76 ± ± ± Dobutamine, μg/kg/min 2.21 ± ± ± IACD, % MAP, mm Hg 70 ± 9 70 ± ± Lactate, meq/l 4.0 ± ± ± Creatinine clearance, ml/min b 7 ± 6 5 ± 4 6 ± 5.46 Values are mean ± SD, median and interquartile range or number of patients in %. EHV, early highvolume hemofiltration; ELV, early low-volume hemofiltration; LLV, late low-volume hemofiltration; Tincl., the time at which all inclusion criteria were fulfilled; IACD, intra-aortic counterpulsation device; MAP, mean arterial pressure. a Estimated using the Cockcroft-Gault formula [12]. b Calculated from a three hours urine portion. p 48

50 Early high-volume CVVH Table 2. Hemofiltration treatment characteristics EHV (n = 35) ELV (n = 35) LLV (n = 36) Hours between T incl. and first CVVH session 6.0 ( ) 7.0 ( ) 41.8 ( ) a Urea before first CVVH, mmol/l 16.3 ( ) 17.1 ( ) 37.4 ( ) a Filtration rate, ml/kg/h 48.2 ( ) b 20.1 ( ) 19.0 ( ) Hours on hemofiltration 68.5 ( ) 94.0 ( ) 69.5 ( ) Filter life span, h 13.6 ( ) 16.1 ( ) 24.3 ( ) a EHV, early high-volume hemofiltration; ELV, early low-volume hemofiltration; LLV, late low-volume hemofiltration; T incl., the time at which all inclusion criteria were fulfilled. a p < difference between LLV and EHV or between LLV and ELV. b p < difference between EHV and ELV or between EHV and LLV. Survival Figure 2 shows the Kaplan Meier estimation of survival rates in the three groups. The 28-day survival was 74.3% in EHV, 68.8% in ELV and 75.0% in LLV (p = 0.80). ICU survival was 71.4% in EHV, 62.9% in ELV and 69.4% in LLV (p = 0.73). Hospital survival was 62.9% in EHV, 48.6% in ELV, and 61.1% in LLV (p = 0.42). ICU survival of the overall population of hemofiltered patients during the study period was 57%. Main causes of death were progressive cardiac failure in 25 patients (55.6%), bowel ischemia in seven patients (15.6%), severe encephalopathy in five patients (11.1%), septic shock and multiple organ failure in three patients (6.7%), pulmonary failure in two patients (4.4%), liver failure in one patient (2.2%) and two patients deteriorated on the surgical ward and did not want further ICU treatment (4.4%). Recovery of renal function Renal function had recovered in all hospital survivors at hospital discharge, except for one patient in ELV. In this patient, dialysis could be discontinued 3 weeks after discharge (71 days after inclusion). At hospital discharge mean creatinine and urea level were respectively 88.1 ± 41.7 μmol/l and 8.5 ± 3.9 mmol/l and comparable among the three groups. Overall median durations of renal failure were not significantly different among the three groups (Table 3.). In hospital survivors, median duration of renal failure was 4.3 ( ) days in EHV, 3.2 ( ) days in ELV and 5.6 ( ) days in LLV (p = 0.25). 49

51 Chapter Survival (%) EHV ELV LLV Days after inclusion Figure 2. Kaplan Meier estimation of survival rates in the three groups. The 28-day survival was 74.3% in early high-volume hemofiltration (EHV), 68.8% in early low-volume hemofiltration (ELV), and 75.0% in late low-volume hemofiltration (LLV) (p = 0.80). 50

52 Early high-volume CVVH Duration of treatment Including survivors and nonsurvivors, median durations of mechanical ventilation, lengths of ICU stay and lengths of hospitalization were not significantly different among the three groups (Table 3). Median duration of mechanical ventilation among ICU survivors was 8.0 ( ) days in EHV, 10.0 ( ) days in ELV and 14.0 ( ) days in LLV (p=0.85). Median length of ICU stay among ICU survivors was 10.0 ( ) days in EHV, 12.0 ( ) days in ELV and 15.0 ( ) days in LLV (p=0.89). Median length of hospitalization in ICU survivors was 31.5 ( ) days in EHV, 34.0 ( ) days in ELV and 42.0 ( ) days in LLV (p=0.42). Table 3. Survival, length of intensive care nit (ICU) stay, length of hospitalization, duration of renal failure, and duration of mechanical ventilation. EHV (n = 35) ELV (n = 35) LLV (n = 36) p Survival, % Day ICU Hospital Duration, days Renal failure 5.5 ( ) 5.7 ( ) 6.6 ( ).55 Mechanical ventilation 8.0 ( ) 11.0 ( ) 12.0 ( ).94 ICU stay 10.0 ( ) 13.0 ( ) 13.5 ( ).96 Hospitalization 27.0 ( ) 27.0 ( ) 35.5 ( ).72 Values are percentage of patients or median and interquartile range. Adverse events During hemofiltration, bleeding complications occurred in 13 of the 106 (12.3%) patients: Three patients in EHV, seven patients in ELV and three patients in LLV (p = 0.23). Mean packed red cell transfusion per bleeding patient was 3.3 ± 1.2 units in EHV, 6.4 ± 2.8 units in ELV, and 5.0 ± 1.7 units in LLV (p = 0.10). All patients with bleeding complications were anticoagulated: four patients were anticoagulated because of hemofiltration, nine patients were anticoagulated for clinical reasons other than hemofiltration. The bleeding location was gastrointestinal in six patients, surgical wounds in five patients, thoracic in one patient and retroperitoneal in one patient. Five of the 106 (4.7%) patients developed HIT: one patient in EHV, three patients in ELV and two patients in LLV (p = 0.59). Hypothermia (< 35 o C) occurred during six of 622 hemofiltration sessions without differences among groups (p = 0.54). None of the patients had bleeding complications, pneumothorax or infections related to the double-lumen catheter. 51

53 Chapter 3 Discussion This RCT in critically ill patients with circulatory and respiratory failure, developing early oliguric ARF, evaluated the effects of three different modes of hemofiltration (EHV, ELV and LLV) on survival, recovery of renal function and duration of treatment. No significant differences were observed among the three groups. However, some findings might be important. First, survival was favorable in all groups compared to literature [3-7,9]. Several factors might have contributed to this survival rate. It might be argued that, of 372 patients hemofiltered during the study period, only 106 patients were enrolled into the study and that this might have caused our higher survival rate. However, ICU survival of the total hemofiltered population was 57%, and these patients were treated for ARF, since both units refrained from the use of very early hemofiltration for severe shock or pulmonary edema. Our study was powered on the basis of the historical mortality rate of 78%, in one of the participating centers (AMC), using low clearance techniques [6]. The survival rate of our LLV group was much better. Although this might suggest that the present study is underpowered, the similarity between the outcomes in the early high volume and late low volume groups is such that even a significant larger study would not be expected to alter the survival findings. The higher survival rate might further be influenced by case mix. Over half our patients were cardiac surgery patients. Although in general cardiac surgery patients tend to do better than septic patients, this is not the case once ARF has developed. If ARF develops following cardiac surgery, it usually occurs along with multiple organ failure and carries a high mortality due to cardiac failure [20]. Our finding that mortality was highest (although not significantly) in the group with the highest percentage of cardiosurgical cases, also contradict that cardiosurgical patients with ARF do better than septic patients. We think that the low mortality observed in our centers might partially be attributed to the closed format of the units. Several authors have reported a reduction in mortality when the organization of an ICU was changed from open to closed [21,22]. Moreover, our ICU nurses are now responsible for all hemofiltration activities, leading to efficient solving of technical problems, rapid, around the clock replacement of hemofiltration systems in case of clotting, and an improved continuity of the treatment. Furthermore, lower mortality rate in the present study might also be attributed to the use of a biocompatible highly permeable membrane, the use of bicarbonate-buffered substitution, and the fact that the initiation of treatment was not as late and the volume not as low as in earlier CAVH studies [23-26]. In this study, no difference in survival was observed among EHV, ELV and LLV. This finding seems to contradict other studies, which have suggested that mortality reduces 52

54 Early high-volume CVVH if the rate of ultrafiltration is increased and RRT is started early [7-9]. In all groups of the present study, mortality was comparable to mortality in the previous high volume or early groups. In the prospective, RCT by Ronco et al. [9], the higher rate of ultrafiltration improved median survival time, a post hoc end point, significantly. In our patients, median survival time was longer than in the patients of Ronco et al. [9], 91 days in the ELV group and not determinable in the other two groups, since > 50% of the patients were discharged from the hospital alive. Case mix in our study and that by Ronco et al. [9] were different. In the study by Ronco et al. [9], all patients with acute oliguric renal failure were enrolled, most patients were postsurgical and the number of failing organs is not mentioned. In contrast, all our patients had at least three failing organ systems (we only enrolled ventilated patients with inotropic support, in whom oliguria existed no longer than 18 h), and more than half of our patients were postcardiosurgical. Ultrafiltrate rate in the study by Ronco et al. [9] study was differently prescribed than in our study. Ronco et al. [9] prescribed ultrafiltrate rate per body weight, respectively 20 ml/kg/h and 45 ml/kg/h in the low- and high-volume groups, but we are not informed about the exact delivered ultrafiltrate rates [9]. In our low-volume groups, treatment dose was based on the average dose delivered in previous reports [3-6]. The treatment dose in the high-volume group was based on the study of Oudemans-van Straaten et al. [8]. Mean actual delivered ultrafiltrate rate was 52 ± 16 ml/kg/h in EHV, 20 ± 4 ml/kg/h in ELV and 19 ± 4 ml/kg/h in LLV. Therefore, treatment doses are comparable between the two randomized studies and do not explain the different results. The better survival in our low-volume groups might also be attributed to the closed format logistics in our units, or to the use of bicarbonate-buffered substitution fluid in our study compared with lactate-buffered solutions in the Ronco et al. study [24]. Recovery of renal function was 100% in all survivors of all groups. This might indicate that all three methods are tolerated well in patients with circulatory failure. Our results are in accordance with the findings of Ronco et al. [9], who found no difference in recovery of renal function among low-volume, medium-volume and high-volume treatment groups, and with the findings of Gettings et al. [7], who found no difference in recovery of renal function between early starters and late starters. Although the definition of recovery of renal function was based in our study on independency of RRT, renal function mostly recovered to pre-morbid level. Recovery of renal function among the nonsurvivors in our study was low and this is in accordance with other authors [7,9]. Most patients die with renal failure. Duration of renal failure was comparable among EHV, ELV and LLV. In contrast to this, a correlation between recovery of diuresis and ultrafiltrate volume, as was reported in a small retrospective study including patients with isolated toxic ARF, could not be confirmed [27]. 53

55 Chapter 3 Four LLV patients had spontaneous recovery of renal function without hemofiltration, and all survived. On the other hand, two LLV patients died before the initiation of hemofiltration. Although this situation might be interesting to analyze in further studies, to determine whether early hemofiltration is unnecessary or essential, the number of included patients in this study prevent us from further speculations on this matter. In this report, median values for duration of mechanical ventilation, length of ICU stay and length of hospitalization were lower in EHV than in the other two groups, but none of these findings were statistically significant. Short-term favorable effects of hemofiltration on gas exchange have been reported after 24 hours of hemofiltration, even with a net positive fluid balance [8,28-30]. These studies evaluated the short term effects of hemofiltration, without the use of a control group. In an experimental model of endotoxic shock in pigs, it was shown that oxygenation improved significantly during high-volume zero-balanced hemofiltration, as compared with a control group [31]. In this study, the pigs were treated very early in the course of septic shock and ultrafiltration rate (predilution) was as high as 240 ml/kg/h, a situation which is not comparable to our patient population. It should be noted that the present study evaluated the effect of EHV in patients with early oliguric ARF. As a result of this criterion, delay time between ICU admission and start of hemofiltration was > 1.5 days. The highest benefit from a high-volume exchange might be earlier in the acute phase of shock. Furthermore, it is our impression that the presently used high-volume dose, developed for very early hemofiltration in severe shock, might be higher than necessary for renal replacement purposes. We would advice either to increase the interval between sessions or to decrease dose when hemodynamics and metabolism are under control. The most important complication in our study was bleeding. In total 13 patients (12.3%) had bleeding complications. Differences among the three groups were not statistically significant. The majority of patients had to be anticoagulated for cardiovascular indications as well. The rate of bleeding in our study is higher than in the study of Ronco et al. [9], who reported only 5% bleeding complications. However, in an earlier report, the same author reported 18 bleeding complications and 8 hematomas in a total of 212 patients (12.3%) [32]. Heparin-induced thrombocytopenia occurred in 3.8% of the patients, and this is comparable with previous reports [33,34]. By using warmed substitution fluid, the incidence of hypothermia during hemofiltration in our study was low, and it was comparable among groups. Median filter life span was significantly longer in the LLV group compared to the EHV group and ELV group. The number of hemofiltration sessions without anticoagulation was comparable between groups, and cannot explain the differences in filter life span. In EHV, filters were routinely changed 54

56 Early high-volume CVVH after 24 h, which was not the case in ELV or LLV, and this has certainly contributed to the shorter filter life span in EHV compared to LLV. Conclusions In the present, prospective RCT, investigating patients with circulatory and respiratory insufficiency developing oliguric ARF, survival and duration of renal failure were not different among EHV, ELV and LLV groups. For all groups, survival was high compared to previous reports and recovery of renal function was 100% in all hospital survivors. Several factors might account for the high survival rate, including ICU organization, continuity of hemofiltration treatment, substitution fluid and the type of hemofilter. In the present population of critically ill patients with early oliguric ARF, we could not show a significant survival benefit for early hemofiltration or high filtration rates; however, no deleterious effects were seen either. High-volume hemofiltration might have logistic advantages. Due to the high clearance, high-volume hemofiltration can be interrupted. In the meantime, the hemofiltration machine is available for other patients and, in between sessions, the patients can be mobilized. Acknowledgements This study is dedicated to the memory of Christiaan P. Stoutenbeek MD, PhD, former director of both our Intensive Care Units. We thank the medical and nursing staff of the Intensive Care Units of Academic Medical Center and Onze Lieve Vrouwe Gasthuis for their cooperation and support. 55

57 Chapter 3 References Davenport A, Will EJ, Davidson AM. Improved cardiovascular stability during continuous modes of renal replacement therapy in critically ill patients with acute renal and hepatic failure. Crit Care Med 1993;21: Bellomo R, Boyce N et al. Continuous venovenous hemodiafiltration compared with conventional dialysis in critically ill patients with acute renal failure. ASAIO J 1993;39 M794-M797 Schwilk B, Wiedeck H, Stein B, et al. Epidemiology of acute renal failure and outcome of haemodiafiltration in intensive care. Intensive Care Med 1997;23: Jones CH, Richardson D, Goutcher E, et al. Continuous venovenous high flux dialysis in multiorgan failure: a 5-year single-center experience. Am J Kidney Dis 1998; 31: Baudouin SV, Wiggins J, Keogh BF, et al. Continuous veno-venous haemodiafiltration following cardio-pulmonary bypass. Intensive Care Med 1993; 19: Douma CE, Redekop WK, van der Meulen JHP, et al. Predicting mortality in intensive care patients with acute renal failure treated with dialysis. J Am Soc Nephrol 1997;8: Gettings LG, Reynolds HN, Scalea T. Outcome in post-traumatic acute renal failure when continuous renal replacement therapy is applied early vs. late. Intensive Care Med 1999;25: Oudemans-van Straaten HM, Bosman RJ, van der Spoel JI, et al. Outcome in critically ill patients treated with intermittent high-volume haemofiltration: a prospective cohort analysis. Intensive Care Med 1999;25: Ronco C, Bellomo R, Homel P, et al. Effects of different doses in continuous venovenous haemofiltration on outcomes of acute renal failure: a prospective randomized trial. Lancet 2000;356:26-30 Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976;16:13-16 Pugh RNH, Murray-Lyon IM, Dawson JL, et al. Transsection of the oesophagus for bleeding oesophageal varices. Br J Surg 1973;60: Lemeshow S, Teres D, Pastides H, et al. A method for predicting survival and mortality of ICU patients using objectively derived weights. Crit Care Med 1985;3: Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of disease classification system. Crit Care Med 1985;13: Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system: Risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991;100:

58 Early high-volume CVVH Le Gall JR, Lemeshow S, Saulnier F. A new simplified acute physiology score (SAPS II) based on a European/North American multicentre study. JAMA 1993;270: Goris RJA, te Boekhorst TPA, Nuytinck JKS, et al. Multiple-Organ Failure. Arch Surg 1985;120: Vincent J-L, Moreno R, Takala J, et al. The SOFA (Sepsis-related organ failure assessment) score to describe organ dysfunction/failure. Intensive Care Med 1996;22: Le Gall J, Klar J, Lemeshow S, et al. The Logistic Organ Dysfunction system. A new way to assess organ dysfunction in the intensive care unit. ICU Scoring Group. JAMA 1996;276: Cohen J. Statistical power analysis for the behavioral sciences. New York: Academic Press 1977; Chertow GM, Levy EM, Hammermeister KE, et al. Independent association between acute renal failure and mortality following cardiac surgery. Am J Med 1998;104: Carson SS, Stocking C, Podsadecki T, et al. Effects of organizational change in the medical intensive care unit of a teaching hospital. A comparison between open and closed formats. JAMA 1996;276: Costas RC, Costa JL, Assad JA, et al. Impact of a coordinated team of intensivists on mortality in critical care: A prospective study over twelve months. Crit Care Med (suppl) 1996;24:A56 Schiffl H, Lang SM, König A, et al. Biocompatible membranes in acute renal failure: prospective case controlled study. Lancet 1994;334: Heering P, K Ivens, Thümer O, et al. The use of different buffers during continuous hemofiltration in critically ill patients with acute renal failure. Intensive Care Med 1999;25: Davenport A, Will EJ, Davison AM. Hyperlactataemia and metabolic acidosis during hemofiltration using lactate-buffered fluids. Nephron 1991;59: Himmelfarb J, Tolkoff RN, Chadran P, et al. A multicenter comparison of dialysis membranes in the treatment of acute renal failure requiring dialysis. J Am Soc Nephrol 1998;9: Journois D, Chanu D, Safran D. Pump-driven haemofiltration: Lancet 1991;337: 985 Bellomo R, Farmer M, Boyce N. Combined acute respiratory and renal failure: management by continuous hemodiafiltration. Resuscitation 1994;28: Gotloib L, Barzilay E, Shustak A, et al. Hemofiltration in septic ARDS. The artificial kidney as an artificial endocrine lung. Resuscitation 1986;13: Coraim FJ, Coraim HP, Ebermann R, et al. Acute respiratory failure after cardiac surgery: Clinical experience with the application of continuous arteriovenous hemofiltration. Crit Care Med 1986;14:

59 Chapter Grootendorst AF, van Bommel EFH, van der Hoven B, et al. High volume hemofiltration improves right ventricular function of endotoxin-induced shock in the pig. Intensive Care Med 1992;18: Ronco C. Continuous renal replacement therapies for the treatment of acute renal failure in intensive care patients. Clin Nephrol 1993;40(4): Ananthasubramaniam K, Shurafa M, Prasad A. Heparin induced thrombocytopenia and thrombosis. Prog Cardiovasc Dis 2000;42(4): Mammen EF. Low molecular weight heparins and heparin-induced thrombocypenia. Clin Appl Thromb Hemost 1999;5:S72-S75 58

60 Chapter 4 The effects of continuous venovenous hemofiltration on coagulation activation Catherine S.C. Bouman, Anne-Cornélie J.M. de Pont, Joost C.M. Meijers, Kamran Bakhtiari, Dorina Roem, Sacha Zeerleder, Gertjan Wolbink, Johanna C. Korevaar, Marcel M. Levi, Evert de Jonge Crit Care 2006;10:R150

61 Chapter 4 Abstract Introduction The mechanism of coagulation activation during CVVH has not been elucidated yet. Insight into the mechanism(s) of hemostatic activation within the extracorporeal circuit could result in a more rational approach to anticoagulation. Aim of the present study was to investigate whether CVVH using cellulose triacetate filters causes activation of the contact factor pathway or the tissue factor pathway of coagulation. In contrast to previous studies CVVH was performed without anticoagulation. Methods Ten critically ill patients were studied prior to the start of CVVH and at 5, 15, 30 min and 1, 2, 3 and 6 h thereafter, for measurement of F1+2, stf, fviia, TFPI, kallikrein-c1- inhibitor and factor XIIa-C1-inhibitor complexes, t-pa, PAI-1, PAP complexes, protein C and antithrombin. Results During the study period, F1+2 levels increased significantly in four patients (defined as group A) and did not change in six patients (defined as group B). Group A also showed a rapid increase in transmembrane pressure, indicating clotting within the filter. At baseline aptt, PT kallikrein-c1-inhibitor complex and factor XIIa-C1-inhibitor complex levels were significantly higher in group B, whereas platelet count was significantly lower. For the other studied markers the differences between group A and group B at baseline were not statistically significant. During CVVH, the difference in time course between group A and B was not statistically significant for the markers of the TF system (stf, fviia and TFPI), the contact system (kallikrein- and factor XIIa-C1-inhibitor complexes) and the fibrinolytic system (PAP complexes, t-pa and PAI-1). Conclusions Early thrombin generation was detected in a minority of intensive care patients receiving CVVH without anticoagulation. Systemic concentrations of markers of the TF system and contact system did not change during CVVH. To elucidate the mechanism of clot formation during CVVH we suggest that future studies are needed that investigate the activation of coagulation directly at the site of the filter. Early coagulation during CVVH may be related to lower baseline levels of markers of contact activation. 60

62 Effect of CVVH on coagulation Introduction Acute renal failure requiring RRT occurs in approximately 4% of patients admitted to the ICU and often these patients are treated with some form of CRRT [1]. Continuous techniques require anticoagulation to allow the passage of blood through the extracorporeal circuit over a prolonged period. Maintenance of CRRT circuits for sufficient duration is important for efficacy, cost effectiveness and minimization of blood component loss. On the other hand, the systemic anticoagulation techniques used to prevent clotting of the circuit are important causes of morbidity in CRRT. Understanding the mechanisms involved in premature clotting of the filtration circuit is mandatory to optimize anticoagulation and maintain filter patency. Several studies have addressed the pathophysiology of circuit thrombogenesis, but the exact mechanism by which it occurs, has not yet been elucidated. Multiple factors may play a role: the extracorporeal circuit itself, treatment modalities, platelet factors, coagulation factors, natural anticoagulants and fibrinolysis [2,3]. Clotting of CRRT circuits could be caused by increased activation of coagulation, initiated by either the (intrinsic) contact activation pathway or the (extrinsic) tissue factor/factor VIIa pathway, or by low activity of the endogenous anticoagulant pathways, such as the antithrombin system, the protein C/protein S system and the tissue factor pathway inhibitor system. In addition, decreased fibrinolysis could also contribute to clotting of extracorporeal circuits. Although much is known about the effect of a single hemodialysis treatment on the coagulation system, there are very few prospective studies that have monitored the effects of repeated passage of blood through a CRRT circuit and these studies were always performed with concurrent administration of anticoagulants, usually unfractionated heparin or low molecular weight heparin [4-7]. As heparin influences both tissue-factormediated coagulation, contact-activated coagulation [7] and fibrinolysis [8], however, studies on the activation of coagulation during CRRT should ideally be performed without anticoagulation. In the present study in critically ill patients with ARF we studied the effects of CVVH without the use of anticoagulation on the activation of coagulation and fibrinolysis. 61

63 Chapter 4 Material and methods Patients The study was approved by the institutional review board and written informed consent was obtained from all participants or their authorized representatives. A cohort of ten critically ill patients with ARF requiring CVVH was studied. Patients were excluded if they fulfilled one of the following criteria: 1) treatment with coumarins or platelet aggregation inhibitors within one week prior to starting CVVH; 2) unfractionated heparin within 12 h prior to starting CVVH or low molecular weight heparin within 48 h prior to starting CVVH; 3) treatment with extracorporeal techniques within 48 h prior to starting CVVH; 4) discontinuation of CVVH for any reason other than clotting of the circuit, e.g. transfer for a computed tomography scan. Continuous venovenous hemofiltration Vascular access was obtained by insertion of a 14 F double lumen catheter (Duo-Flow 400 XL, Medcomp, Harleysville PA, USA) into a large vein (femoral, subclavian, or internal jugular vein). Hemofiltration was performed with computer controlled; fully automated hemofiltration machines (Diapact, Braun AG, Melsungen, Germany). A 1.9 m 2 cellulose triacetate hollow-fiber membrane with a 2-microglobulin SC of approximately 0.82 was used (CT190G, Baxter, Mc Gaw Park, IL, USA). The blood flow rate was 150 ml/min and warmed substitution fluid was added in predilution mode at a flow rate of 2 L/h. The hemofiltration run continued until the extracorporeal circuit clotted. No anticogulant was used during CVVH, and neither was the extracoorporeal aircut primed with any anticoagulant. Blood collection Blood was drawn from the venous limb of the hemofiltration catheter before starting hemofiltration and, at 5, 15 and 30 minutes and at 1, 2, 3 and 6 hours after commencement. For the determination of contact activation, 4.8 ml blood was collected in siliconized vacutainer tubes to which 0.2 ml of a mixture of EDTA (0.25 M), benzamidine (0.25 M) and soybean-trypsin inhibitor (0.25%) was added to prevent in vitro contact activation and clotting. All other blood samples were collected in citrated vacutainer tubes. Plasma was prepared by centrifugation of blood at 2500 g twice for 20 minutes at 16 C, followed by storage at 80 C until assays were performed. Assays The plasma concentrations of F1+2 were measured by ELISA (Dade Behring, Marburg, Germany). Soluble tissue factor was also determined by ELISA (American Diagnostica, 62

64 Effect of CVVH on coagulation Greenwich CT, USA). The plasma concentration of fviia was determined on a Behring Coagulation System (BCS, Dade Behring) with the StaClot VIIa-rTF method from Diagnostica Stago (Asnières-sur-Seine, France). Tissue factor pathway inhibitor activity was measured on the BCS as described by Sandset et al. [9]. Kallikrein-C1-inhibitor and factor XIIa-C1-inhibitor complexes were measured as described by Nuijens et al. [10]. Tissue-type plasminogen activator antigen and PAI-1 antigen was assayed by ELISA (Innotest PAI-1, Hyphen BioMed, Andrésy, France). Antithrombin activity was determined with Berichrom Antithrombin (Dade Behring) on a BCS. Plasmin-antiplasmin (PAP) complexes were determined with a PAP micro ELISA kit (DRG, Berlin Germany). Protein C was determined using the Coamatic protein C activity kit from Chromogenix (Mölndal, Sweden). Statistical analysis Values are given as median and ranges. We used the Mann-Whitney U test to analyze the difference between baseline variables and linear mixed models to evaluate the difference over time between groups. Data were analyzed using the Statistical Package for the Social Sciences for Windows, version 11.0 (SPSS, Chicago IL, USA). A p-value < 0.05 was considered significant. Hemofilter survival times were compared using the Kaplan-Meier method and the log-rank test (GraphPad Prism 4.0, GraphPad software inc., San Diego CA, USA). Results Baseline characteristics The baseline characteristics of the 10 enrolled patients are shown in table 1. Thrombin generation and clotting of the circuit Nine out of ten patients showed coagulation activation before the initiation of CVVH, as reflected by increased F1+2 levels. Figure 1 shows the F1+2 levels during CVVH for each patient. The concentrations of F1+2 increased in patient 1, 2, 8 and 9 (defined as group A) and did not change in the other patients (defined as group B) (p < 0.001). One hour after the onset of CVVH the relative increase in transmembrane pressure was significantly higher (p = 0.01) in group A compared with group B (respectively 57% (42% 80%) in group A and 2% (2% 7%) in group B). In group A the lifespan of the circuit was less than 4.3 hours in 3 patients, but one patient had an unexpected long circuit run of 22.5 hours and the difference in circuit life span was not significantly different between the two groups (figure 2). 63

65 Chapter 4 Table 1. Patient characteristics. Trial no. Age (year) Gender Diagnosis APACHE a) II Cause ARF Type of ARF Urea b) (mmol/l) Creatinine b) (μmol/l) Duration of the CVVH circuit studied (h) Filter lifespan (h ) Outcome Group A (patients with increased thrombin generation) 1 44 Male Subarachnoidal hemorrhage 29 Nonseptic anuric Died 2 52 Male Lung cancer and pneumonia 10 Septic oliguric Died 8 68 Female Thoracic aortic prosthesis 15 Nonseptic Nonoliguric Survived 9 64 Male Ruptured abdominal aortic aneurysm 14 Nonseptic Nonoliguric Survived Group B (Patients without increased thrombin generation) 3 65 Male Ruptured abdominal aortic aneurysm 28 Nonseptic anuric Survived 4 48 Male 23 Septic oliguric Survived Streptococcal sepsis Male Myocardial infarction 23 Nonseptic anuric Died 6 65 Male Bowel ischemia 18 Septic oliguric Died 7 67 Male Non-Hodgkin lymphoma 24 Nonseptic Nonoliguric Died Male 23 septic oliguric Died Peritonitis 23.9 a) APACHE II; the acute physiology and chronic health evaluation II score at ICU admission [26]. ARF, Acute renal failure; b) Before CVVH 64

66 Effect of CVVH on coagulation Baseline coagulation parameters Coagulation parameters before the initiation of CVVH are shown in table 2, along with their reference values. By comparison to group B, baseline levels of aptt, kallikrein-c1- inhibitor complex and factor XIIa-C1-inhibitor complex were significantly lower in group A whereas platelet count was significantly higher in group A. Table 2. Baseline levels of coagulation markers. Group A Group B p Reference values F1+2 (nmol/l) 2.5 ( ) 4.1 ( ) stf (pg/ml) 126 (73 216) 207 (30 322) fviia (mu/ml) 61 (14 141) 97 (20 267) TFPI (ng/ml) 167 ( ) 127 (56 200) Antithrombin (%) 78 (46 100) 45 (16 81) Protein C (%) 63 (29 164) 41 (16 89) PAP complexes (ng/ml) 682 ( ) 727 ( ) t-pa (ng/ml) 14.3 ( ) 11.7 ( ) PAI-1 (ng/ml) 135 (16 275) 526 ( ) Kallikrein-C1-inhibitor complex (mu/ml) Factor XIIa-C1-inhibitor complex (mu/ml) 8.2 ( ) 11.1 ( ) ( ) 2.4 ( ) 0.02 < 0.6 < 0.5 Platelet count (*10 9 /L) 136 (90 329) 63 (30 101) PT (sec) 13.5 ( ) 17.5 ( ) aptt (sec) 25 (21 29) 37 (27 57) Group A, patients with increased thrombin generation; Group B, patients without increased thrombin generation. Values are median and ranges; F1+2, prothrombin fragment F1+2; stf, soluble tissue factor; fviia, activated factor VII; TFPI, tissue factor pathway inhibitor; PAP, pasmin-antiplasmin; t-pa, tissue type plasminogen activator; PAI-1, plasminogen activator inhibitor type 1. Coagulation parameters during CVVH The time courses of the coagulation markers are shown in figures 2-4. Data points are shown as a percentage of the initial concentration for those markers that were not significantly different at baseline (figure 3 and figure 5), whereas data points are shown as absolute values for those markers that were significantly different at baseline (figure 4). Analysis of the difference in the time course between group A and group B was limited to the first three hours after the start of CVVH, because in group A, only 1 patient was still on CVVH at 6 h. 65

67 Chapter 4 F1+2, Group A F1+2, Group B % initial concentration Time (h) Time (h) Figure 1. Prothrombin fragment F1+2 during hemofiltration. Curves represent values of individual patients. Group A, patients demonstrating an increase in thrombin generation. Group B, patients with a constant level of thrombin generation. 100 Hemofilter survival (%) p = Time to hemofilter failure (h) Figure 2. Kaplan-Meier survival function indicating hemofilter survival times between patients with increased thrombin (Group A, closed circles) and patients without increased thrombin generation (Group B, open circles). 66

68 Effect of CVVH on coagulation stf factor VIIa TFPI % initial concentration p = Time (h) p = Time (h) p = Time (h) Figure 3. Soluble tissue factor (stf), factor VIIa and tissue factor pathway inhibitor (TFPI) during hemofiltration. Data points are median and interquartile ranges. The closed circles represent patients with thrombin generation (group A) and the open circles represent patients without thrombin generation (group B). The p-value represents the difference in time course between both groups by linear mixed models and during the first three hours of hemofiltration. Kallikrein-C1-inhibitor complex Factor XIIa-C1-inhibitor complex mu/ml p = p = Time (h) Time (h) Figure 4. Concentrations of kallikrein-c1 inhibitor and factor XIIa-C1-inhibitor complexes during hemofiltration. Levels are absolute values in order to display the significant (p 0.02) difference at baseline. Data points are median and interquartile ranges. The closed circles represent patients with thrombin generation (group A) and the open circles represent patients without thrombin generation (group B). The p-value represents the difference in time course between both groups by linear mixed models and during the first three hours of hemofiltration. 67

69 Chapter 4 PAP t-pa PAI % initial concentration p = Time (h) p = Time (h) p = Time (h) Figure 5. Plamin-antiplasmin (PAP) complexes, tissue plasminogen activator (t-pa), and plasminogen activator inhibitor type 1 (PAI-1) during hemofiltration. Data points are median and interquartile ranges. The closed circles represent patients with thrombin generation (group A) and the open circles represent patients without thrombin generation (group B). The p-value represents the difference in time course between both groups by linear mixed models during the first three hours of hemofiltration. The difference in time course between group A and B was not significant for the tissue factor system (figure 3) and the contact system (figure 4). Levels of t-pa and PAI-1 were also not significantly different between group A and group B during CVVH (figure 5). PAP complex levels tended to increase in group A during CVVH (p=0.07). Discussion In the present study in critically ill patients, we investigated the early effects of CVVH without anticoagulation on systemic markers of coagulation activation and fibrinolysis. During the first six hours of CVVH, increased thrombin generation was found in only four out of ten patients. An early increase in transmembrane pressure, indicating filter clotting, was exclusively seen in the four patients with thrombin generation. Premature clotting of the circuit was found in three of these four patients, necessitating replacement of the circuit. CVVH without anticoagulation did not change the systemic concentrations of markers of the intrinsic or extrinsic pathway, nor did CVVH affect the systemic concentrations of fibrinolysis markers. Thrombin generation on an artificial surface, such as the filter membrane, has traditionally been attributed to contact activation of the intrinsic pathway of coagulation 68

70 Effect of CVVH on coagulation that starts upon exposure of contact factors (factor XII, high molecular weight kallikrein and prekallikrein) to a negatively charged surface and their subsequent activation. We did not find any change in plasma levels of factor XIIa-C1-inhibitor complex and kallikrein-c1-inhibitor complex, making initiation of coagulation via this pathway less likely. This finding confirms the results of Salmon et al. who did not find an increase in contact activation during CVVH using a polyacrilonitrile membrane and systemic heparinization [6]. Interestingly, in our study baseline levels of factor XIIa-C1-inhibitor complex and kallikrein-c1-inhibitor complex were relatively lower in patients with early increased thrombin generation during CVVH. Several authors have described the role of factor XIIa and kallikrein in the activation of fibrinolysis [11,12]. Factor XII is able to activate fibrinolysis by three different pathways: 1) it activates prekallikrein, which in turn activates urokinase-type plasminogen activator, 2) following the activation of prekallikrein, the kallikrein generated can liberate t-pa and 3) factor XII activates plasminogen directly. The role of contact activation-dependent fibrinolysis in vivo is unclear, but a relationship between contact activation-dependent fibrinolysis and thromboembolic complications has been described [13,14]. Thus, low baseline activation of the contact system may be associated with lower fibrinolysis and an increased risk of filter clotting. However, in our study fibrinolysis during CVVH was not decreased in group A. On the contrary, we observed a trend towards increased PAP levels during CVVH in patients with early clotting of the filter. This PAP level increase is most probably caused by activated coagulation leading to plasmin generation from plasminogen on the formed fibrin. In this respect, therefore PAP levels may be more an indication of coagulation than of fibrinolytic activity itself. Alternatively, one could speculate that patients with higher baseline levels of factor XIIa- C1-inhibitor complex and kallikrein-c1-inhibitor complex have higher baseline thrombin generation. Baseline F1+2 levels were higher in group B, although the difference was not statistically significant, possibly due to the small number of patients. Thrombin is required for activation of the endogenous anticoagulant protein C system [15,16]. In patients with higher levels of factor XIIa-C1-inhibitor complex and kallikrein-c1-inhibitor complex, it is conceivable that coagulation activation during CVVH is decreased following increased endogenous anticoagulant activity. Indeed, an anticoagulant effect of thrombin infusion has been reported in a dog model [16]. In the present study protein C levels were not different at baseline between the two groups, however we did not measure the activated protein C levels. A contribution of the extrinsic pathway to thrombin generation on artificial surfaces is unexpected at first sight since TF is normally not found on the surface of cells in 69

71 Chapter 4 contact with blood. Monocytes, however, can express TF under certain pathophysiologic conditions, mostly associated with increased endotoxin and/or cytokine levels [17]. Based on the measurements of circulating fviia, stf and TFPI, we did not find signs of activation of coagulation via the extrinsic pathway. Our findings are in contrast with another study that concluded that activation of TF/VIIa mediated coagulation took place in critically ill patients treated with CVVH [4]. This conclusion was based on increased levels of thrombin-antithrombin complexes and fviia and on decreased levels of TFPI during CVVH. The change in circulating fviia and TFPI levels, however, was relative to values just after start of CVVH with concurrent administration of heparin. No change in fviia and TFPI levels was found when they where compaqred with pre-cvvh values. The observed changes in TFPI and fviia in this study may represent the effects of heparin, rather than activation of the TF/VIIa mediated pathway of coagulation, because the concentration of TFPI increases after administration of heparin [18], and high TFPI levels may bind fviia. In our study, CVVH was performed without administration of heparin and no changes in markers of TF/VIIa mediated coagulation were observed. What is the mechanism of increased thrombin generation in the absence of detectable activation of the extrinsic and intrinsic coagulation system? One explanation could be a lack of sensitivity of the systemic markers such as stf, fviia and TFPI. The total volume of blood in the extracorporeal circuit is only approximately 300 ml. The absolute amount of thrombin formation may therefore be too low to lead to detectable increases in plasma levels of precursor proteins, such as soluble TF or fviia. In that case, different study designs are needed to show the pathophysiologic mechanism underlying coagulation during CVVH, e.g. studies analyzing tissue factor expression on monocytes in pre- and post-filter samples or studies directly analyzing the clot formed in the hemofilter. Alternatively, an increase in systemic coagulation markers could be prevented by the removal of markers across the filter membrane into the ultrafiltrate or secondary to adsorption to the membrane. The high molecular weight ( 35 kda) and polarity of coagulation factors, however, should significantly prevent marker removal during hemofiltration [19]. In our previous in vitro hemofiltration study using the same cellulose triacetate membrane as in the present study we found only minimal filtration of IL-6 (MW kda) and the calculated sieving coefficient was approximately 0.1 in predilution mode [20]. In general the process of adsorption to the membrane is rapidly saturated, but we cannot rule out some adsorption to the membrane during the first hour of CVVH. Finally, it is also conceivable that alternative pathways of thrombin generation are responsible for filter clotting, including the direct activation of factor X, either on the surface of activated platelets or by the integrin receptor MAC-1 on leukocytes [3]. 70

72 Effect of CVVH on coagulation Low levels of natural anticoagulants have been suggested to contribute to early filter clotting. In the randomized CRRT study by Kutsogiannis et al. [21], comparing regional citrate anticoagulation with heparin anticoagulation, decreasing antithrombin levels were an independent predictor of an increased risk of filter failure. In the retrospective study by du Cheyron et al. [22], in sepsis patients requiring CRRT and with acquired antithrombin deficiency, anticoagulation with unfractionated heparin plus antithrombin supplementation prevented premature filter clotting. In our own experience, treatment with recombinant human activated protein C obviated additional anticoagulation during CVVH in patients with severe sepsis [23]. In the present study, however, baseline levels of antithrombin and protein C were not extremely low and no significant difference between patients with and without early thrombin generation was found. Another natural defense mechanism against activated coagulation is the fibrinolytic system, and a disturbance of the normal balance between fibrinolysis and anti-fibrinolysis might play a role in thrombosis of the CVVH circuit. In our study the difference in PAP complex, t-pa and PAI-1 levels before and during CVVH were not statistically significant in patients with and without thrombin generation, but our study is limited by the small number of patients. At baseline, the platelet count was significantly lower and PT and aptt were significantly longer in those patients without subsequent coagulation activation. The association of a low platelet count with decreased risk of filter clotting confirms our finding in earlier studies [24]. The association also confirms the findings of Holt et al., who showed an association between the starting aptt and time to circuit clotting [25]. Conclusion We conclude that activation of coagulation can be detected in a minority of intensive care patients treated with CVVH without anticoagulation. Systemic concentrations of markers of the TF/VIIa system and contact system did not change during CVVH. We suggest that different studies investigating the activation of coagulation directly at the site of the filter, are needed to elucidate the mechanism of clot formation during CVVH. 71

73 Chapter 4 References Uchino S, Kellum JA, Bellomo R, et al. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA 2005;294: Davenport A. The coagulation system in the critically ill patient with acute renal failure and the effect of an extracorporeal circuit. Am J Kidney Dis 1997;30:S20-S27. Schetz M. Anticoagulation in continuous renal replacement therapy. Contrib Nephrol 2001; Cardigan RA, McGloin H, Mackie IJ, et al. Activation of the tissue factor pathway occurs during continuous venovenous hemofiltration. Kidney Int 1999;55: Klingel R, Schaefer M, Schwarting A, et al. Comparative analysis of procoagulatory activity of haemodialysis, haemofiltration and haemodiafiltration with a polysulfone membrane (APS) and with different modes of enoxaparin anticoagulation. Nephrol Dial Transplant 2004;19: Salmon J, Cardigan R, Mackie I, et al. Continuous venovenous haemofiltration using polyacrylonitrile filters does not activate contact system and intrinsic coagulation pathways. Intensive Care Med 1997;23: Wendel HP, Heller W, Gallimore MJ. Influence of heparin, heparin plus aprotinin and hirudin on contact activation in a cardiopulmonary bypass model. Immunopharmacology 1996;32: Urano T, Ihara H, Suzuki Y, et al. Coagulation-associated enhancement of fibrinolytic activity via a neutralization of PAI-1 activity. Semin Thromb Hemost 2000;26: Sandset PM, Abildgaard U, Pettersen M. A sensitive assay of extrinsic coagulation pathway inhibitor (EPI) in plasma and plasma fractions. Thromb Res 1987;47: Nuijens JH, Huijbregts CC, Eerenberg-Belmer AJ, et al. Quantification of plasma factor XIIa-Cl(-)-inhibitor and kallikrein-cl(-)-inhibitor complexes in sepsis. Blood 1988; 72: Braat EA, Dooijewaard G, Rijken DC. Fibrinolytic properties of activated FXII. Eur J Biochem 1999;263: Schousboe I, Feddersen K, Rojkjaer R. Factor XIIa is a kinetically favorable plasminogen activator. Thromb Haemost 1999;82: Himmelreich G, Ullmann H, Riess H, et al. Pathophysiologic role of contact activation in bleeding followed by thromboembolic complications after implantation of a ventricular assist device. ASAIO J 1995;41:M790-M794. Jespersen J, Munkvad S, Pedersen OD, et al. Evidence for a role of factor XII-dependent fibrinolysis in cardiovascular diseases. Ann N Y Acad Sci 1992;667: Esmon CT. Protein C anticoagulant pathway and its role in controlling microvascular thrombosis and inflammation. Crit Care Med 2001;29:S48-S51. 72

74 Effect of CVVH on coagulation Taylor FB, Jr., Chang A, Hinshaw LB, et al. A model for thrombin protection against endotoxin. Thromb Res 1984;36: Osterud B. Tissue factor expression by monocytes: regulation and pathophysiological roles. Blood Coagul Fibrinolysis 1998;9 Suppl 1:S9-14. Tobu M, Ma Q, Iqbal O, et al. Comparative tissue factor pathway inhibitor release potential of heparins. Clin Appl Thromb Hemost 2005;11: Guth HJ, Klingbeil A, Wiedenhoft I, et al. Presence of factor-vii and -XIII activity in ultrafiltrate during hemofiltration. Int J Artif Organs 1999;22: Bouman CS, van Olden RW, Stoutenbeek CP. Cytokine filtration and adsorption during pre- and postdilution hemofiltration in four different membranes. Blood Purif 1998;16: Kutsogiannis DJ, Gibney RT, Stollery D, et al. Regional citrate versus systemic heparin anticoagulation for continuous renal replacement in critically ill patients. Kidney Int 2005;67: du Cheyron D, Bouchet B, Bruel C, et al. Antithrombin supplementation for anticoagulation during continuous hemofiltration in critically ill patients with septic shock: a case-control study. Crit Care 2006;10:R45. de Pont AC, Bouman CS, de Jonge E, et al. Treatment with recombinant human activated protein C obviates additional anticoagulation during continuous venovenous hemofiltration in patients with severe sepsis. Intensive Care Med 2003;29:1205. de Pont AC, Oudemans-van Straaten HM, Roozendaal KJ, et al. Nadroparin versus dalteparin anticoagulation in high-volume, continuous venovenous hemofiltration: a double-blind, randomized, crossover study. Crit Care Med 2000;28: Holt AW, Bierer P, Bersten AD, et al. Continuous renal replacement therapy in critically ill patients: monitoring circuit function. Anaesth Intensive Care 1996;24: Knaus WA, Wagner DP, Draper EA, et al. The APACHE III prognostic system. Risk prediction of hospital mortality for critically ill hospitalized adults. Chest 1991;100:

75

76 Chapter 5 Predilution versus postdilution during continuous venovenous hemofiltration: a comparison of circuit thrombogenesis Anne-Cornélie J.M. de Pont, Catherine S.C. Bouman, Kamran Bakhtiari, Marianne C.L. Schaap, Rienk Nieuwland, Augueste Sturk, Barbara A. Hutten, Evert de Jonge, Margreeth B. Vroom, Joost C.M. Meijers, Harry R. Büller ASAIO Journal 2006;52:

77 Chapter 5 Abstract Background During CVVH, predilution can prolong circuit survival time, but the underlying mechanism has not been elucidated. Aim of the present study was to compare predilution to postdilution with respect to circuit thrombogenesis. Methods Eight critically ill patients were treated with both pre- and postdilutional CVVH in a crossover fashion. A filtration flow of 60 ml/min was used in both modes. We chose blood flows of 140 and 200 ml/min during pre- and postdilution respectively, to keep the total flow through the hemofilter constant. Extracorporeal circuit pressures were measured hourly and samples of blood and ultrafiltrate were collected at five different timepoints. Thrombin-antithrombin complexes and prothrombin fragments F1+2 were measured by ELISA and platelet activation was assessed by flow cytometry. Results No signs of thrombin generation or platelet activation were found during either mode. During postdilution, baseline platelet count and maximal prefilter pressure had a linear relationship, while both parameters were inversely related with circuit survival time. Conclusion Predilution and postdilution did not differ with respect to extracorporeal circuit thrombogenesis. During postdilution, baseline platelet count and maximal prefilter pressure were inversely related with circuit survival time. 76

78 Pre- versus postdilution CVVH Introduction During the last decades CVVH has become the treatment of choice in critically ill patients needing RRT. However, thrombosis in the extracorporeal circuit has always been a limiting factor [1]. Several systemically administered anticoagulants have been used to limit the activation of coagulation in the extracorporeal circuit, such as unfractionated and low molecular weight heparins, danaparoid, hirudin and nafamostat. However, the use of these anticoagulants is limited by the risk of bleeding. Another technique to limit thrombosis in the extracorporeal circuit is regional anticoagulation of the hemofilter, using either citrate before and calcium after the filter, or heparin before and protamine after the filter. Citrate anticoagulation has recently gained popularity, although it carries the risk of metabolic disorders [2]. Regional anticoagulation using heparin and protamine carries the risk of protamine toxicity [3]. Historically, predilution has been suggested as another method to limit coagulation in the extracorporeal circuit, because it lowers hematocrit level, platelet count, and concentration of coagulation factors in the hemofilter [4]. However, trials comparing predilution with postdilution are limited [5,6]. Moreover, the mechanisms by which predilution can prolong circuit survival time, have not been determined. In the present randomized crossover study we compared the effects of predilution and postdilution CVVH on thrombin generation and platelet activation.- Material and methods Patients The study was approved by the institutional review board, and written informed consent was obtained from all participants or their authorized representatives. Critically ill adult patients with an indication for RRT were eligible for the study. Exclusion criteria were recent bleeding, treatment with aspirin within one week before enrollment, treatment with therapeutic doses of unfractionated or low-molecular-weight-heparin within 12 hours before enrollment, and results of routine coagulation tests such as PT and aptt exceeding twice the upper limit of normal. Procedure Eligible patients were randomly assigned to either predilution or postdilution for their first CVVH run. After finishing a first run in predilution mode, the second was performed after an interval of 12 hours in postdilution mode and vice versa. To obtain vascular access, a double lumen catheter (Duo-Flow 400XL, 14F x 6 (15 cm), Medcomp, 77

79 Chapter 5 Harleysville PA, USA) was inserted into a large vein (femoral, subclavian, or internal jugular vein). CVVH was performed using a Diapact hemofiltration machine (Braun AG, Melsungen, Germany) and a cellulose triacetate hemofilter (CT.190G, Baxter Healthcare Corp., Deerfield, IL, USA). Warmed bicarbonate buffered substitution fluid with a flow of 60 ml/min was used in both predilution and postdilution. The blood flow was set at 200 ml/min during postdilution and at 140 ml/min during predilution to keep the total flow through the hemofilter constant at 200 ml/min in both modes. A negative fluid balance was allowed. Extracorporeal circuit pressures were measured every hour and limits were preset as follows: arterial pressure: 200 mmhg, prefilter pressure: 400 mmhg, transmembrane pressure: 450 mmhg. Circuit survival time was defined as the time during which the preset ultrafiltration rate was achieved, i.e. when extracorporeal circuit pressures exceeded the preset pressure limits, leading to automatic reduction of the ultrafiltration rate, this was considered as an end point. Anticoagulation The circuit was primed with 2 L NaCl 0.9%, to which IU of nadroparin (Sanofi- Synthelabo, Paris France) were added. Before starting CVVH, a loading dose of 2850 IU nadroparin was administered intravenously, followed by a continuous prefilter infusion of 456 IU/h. Blood and ultrafiltrate collection Blood was collected from the hemofiltration catheter before CVVH and from the extracorporeal lines immediately before and after the hemofilter at 0.5, 6, 12 and 18 hours during CVVH. Ultrafiltrate was obtained at the same time points. Blood for the determination of hemoglobin, hematocrit, leukocyte and platelet counts was collected in K3-EDTA tubes and for the determination of urea in lithium heparin tubes. Blood for flow cytometry and coagulation assays was collected in 0.32% trisodium citrate and processed within 15 minutes. Plasma was prepared by centrifugation at 2500 g twice for 20 minutes at 16 C, followed by storage at 80 C until assays were performed. Laboratory assays PT and aptt were performed on an automated coagulation analyzer (Behring Coagulation System, Dade Behring, Marburg, Germany). Antithrombin activity was determined with Berichrom Antithrombin (Dade Behring) on a Behring Coagulation System. Thrombinantithrombin complexes and F1+2 were measured by ELISA (Dade Behring, Marburg, Germany). Factor VII antigen levels were determined with an ELISA from Diagnostica Stago (Asnières-sur-Seine, France). Platelet activation was assessed by means of flow cytometric measurements, using a method adapted from Maquelin et al. [7] Aliquots of 78

80 Pre- versus postdilution CVVH 5 l citrated blood were diluted in 35 l N-2-hydroxyethylpiperazine-N-2-ethanesulfonic acid (HEPES) buffer to which 5 l fluorescein isothiocyanate (FITC)-labeled antiglycoprotein (GP) Ib and 5 l of a second phycoerythrin (PE)-labeled monoclonal antibody were added: anti-cd62p (P-selectin) (Beckman Coulter, Miami FL, USA), anti- CD63 (GP53) (Beckman Coulter) and anti-fibrinogen (Biopool, Umeå, Sweden). After mixing and incubation for 30 minutes at room temperature in the dark, further staining was stopped by adding 2.5 ml HEPES buffer containing 0.3% paraformaldehyde. Flow cytometric measurements were performed in a FACScan flow cytometer with CellQuest software (Becton Dickinson, San Jose CA, USA). Forward (FSC) and sideward (SSC) light scatter were set at logarithmic gain. Platelets were identified on FSC, SSC and binding of FITC-labeled anti-gpib. The surface expression of activation markers was determined in a population of 5000 platelets. The threshold for platelet activation was arbitrarily set at 2% with a PE-labeled IgG 1 control-antibody. All coagulation and flow cytometric results are reported without correction for the dilution used in the extracorporeal circuit, unless stated otherwise. Correction of laboratory values for hemoconcentration across the hemofilter To assess the actual change in levels of TAT, F1+2 and platelet count across the hemofilter, values measured at the filter inlet were corrected for hemoconcentration across the hemofilter using a correction factor, based on ultrafiltration rate (Qf) and hematocrit (Ht). Taking into account that TAT levels were measured in plasma, we used the following formula to correct TAT levels for hemoconcentration across the hemofilter: TAT corr = TAT pre x 200 (1-Ht pre ) 140 (1-Ht post ) where TAT corr is the TAT value corrected for the hemoconcentration across the hemofilter, TAT pre is the TAT value at the hemofilter inlet, Ht pre is the Ht at the hemofilter inlet and Ht post is the Ht at the hemofilter outlet. The amount of fluid entering the filter inlet per minute (200 ml) is multiplied by the plasmafraction (1-Ht pre ) to obtain the amount of plasma in which TAT pre was measured, and this value is divided by the plasmafraction to which the amount of fluid was concentrated across the hemofilter [140 (1-Ht post)]. To correct levels of F1+2, a similar formula was used. As platelet counts were measured in whole blood, we used the following formula to correct platelet counts for hemoconcentration across the hemofilter: 79

81 Chapter 5 Platelets corr = platelets pre x where platelets pre is the concentration of platelets at the filter inlet and platelets corr the concentration of platelets corrected for hemoconcentration across the hemofilter. Clearance calculations The filter clearance of urea was calculated at t = 0.5, 6, 12 and 18 h using the following formula: Kf urea = urea uf x UF volume urea before filter where Kf urea is the filter clearance of urea (in ml/min), urea UF is the urea concentration in the ultrafiltrate (in mmol/l), UF volume is the volume ultrafiltrated (in ml/min) and urea before filter is the uncorrected concentration of urea in the fluid entering the hemofilter (in mmol/l). The filter extraction of urea Ef urea was calculated by dividing the filter clearance of urea Kf urea by the actual filter flow of 200 ml/min in both predilution and postdilution: Ef urea = Kf urea urea uf x UF volume = x urea before filter The extracorporeal urea clearances K expost for postdilution and K expre for predilution were calculated by multiplying the filter extraction Ef urea by the actual blood flow (200 ml/min for postdilution and 140 ml/min for predilution): Kf expost urea = urea uf x UF volume urea before filter Kf expre urea = 140 x urea uf x UF volume 200 x urea before filter 80

82 Pre- versus postdilution CVVH Statistical analysis Data were analyzed on an intention to treat basis, using the Statistical Package for the Social Sciences (SPSS) for Windows, version 11.0 (SPSS, Chicago IL, USA). Differences between pre- and postdilution were tested by analysis of repeated measures, using mixed linear models. Changes from baseline to a certain time point within the same group were analyzed by a paired Student s t-test. Regression analysis was used to determine the influence of different parameters on circuit survival time. Values are given as means ± SD or medians and range if appropriate. Significance was defined as p < Results Patient characteristics A total of eight patients were enrolled in the study. Baseline patient characteristics are shown in Table 1 and baseline coagulation parameters in Table 2. Except for baseline platelet count, all baseline coagulation parameters were similar in both modes. Table 1. Baseline characteristics of the patients Characteristic Total 8 Age 63 ± 13 Male sex (%) 3 (38) Mean body weight (kg) 80 ± 19 APACHE II score 23 ± 8 Mechanical ventilation 6 Vasopressor use 6 Number of dysfunctional organ systems 4 ± 1 Clinical settings Infectious disease 6 Cardiac surgery 1 Heart failure 1 APACHE, acute physiology and chronic health evaluation. Data represent mean ± SD or number of patients. 81

83 Chapter 5 Table 2. Baseline coagulation parameters during predilution and postdilution Characteristic Predilution Postdilution p value Baseline hematocrit (l/l) 0.30 ± ± 0.04 NS PT (sec) 14.9 ± ± 1.9 NS aptt (sec) 24 ± 5 22 ± 6 NS AT (%) 58 ± ± 19 NS TAT ( g/l) 24.7 ± ± 17.8 NS F1+2 (nmol/l) 1.86 ± ± 1.5 NS FVII Ag (%) 55 ± ± 30 NS Baseline platelet count (x 10 9 /l) 168 ± ± 67 < 0.05 P-selectin expressing platelets (%) 5.1 ± ± 2.1 NS GP-53 expressing platelets (%) 4.9 ± ± 2.8 NS Fibrinogen expressing platelets (%) 3.2 ± ± 0.6 NS PT, prothrombin time; aptt, acivated partial thromboplastin time; AT, antithrombin; TAT, thrombin-antithrombin complex; F1+2, prothrombin fragment F1+2; FVII Ag, factor VII antigen; GP-53, glycoprotein 53. Data represent mean ± SD. Circuit survival time In the 8 enrolled patients, 15 hemofiltration runs were performed, 8 in predilution and 7 in postdilution mode. In one patient, renal function recovered after the first hemofiltration run, obviating the need for a second run. Median circuit survival time during predilution was 28 hours (range hours) and during postdilution 15 hours (range 1 65 hours) (NS) (Figure 1). Four patients had their first run in predilution mode and four in postdilution mode. Median circuit survival time for the first run was 20 hours (range 1 65 hours) and for the second run 14 hours (range 1 44 hours) (NS). Although not statistically significant, a greater number of hemofilters expired within one hour during postdilution (2 out of 7) compared with predilution (0 out of 8). Hematocrit Predilution decreased the hematocrit from 0.30 ± 0.04 to 0.18 ± 0.02 (p < 0.001). Due to ultrafiltration, the hematocrit increased across the hemofilter to preprocedure levels (0.29 ± 0.04 vs 0.30 ± 0.04, NS). During postdilution, the hematocrit increased across the hemofilter to approximately 1.4 times baseline (0.38 ± 0.08 vs 0.28 ± 0.05, p < 0.01). Hematocrit levels measured before the filter during the procedure were similar to those measured before the procedure (0.30 ± 0.04 vs 0.28 ± 0.04, NS). 82

84 Pre- versus postdilution CVVH 100 predilution postdilution filter survival (%) 50 p = time(h) Figure 1. Survival curve of extracorporeal circuit during predilution and postdilution hemofiltration. Clotting times During both predilution and postdilution, PT values increased significantly after administration of the nadroparin bolus, remaining significantly higher before the filter than after the filter (Table 3). Similarly, aptt values increased significantly after administration of the nadroparin bolus in both predilution and postdilution. During predilution, the aptt value remained significantly higher before the filter than after the filter throughout the study period (p < 0.01), whereas during postdilution, the APTT values both before and after the filter gradually returned to normal in the course of the study period. Thrombin generation In both modes, there was no significant change in levels of TAT and F1+2, neither across the hemofilter, nor over time, indicating that no activation of coagulation was observed (Table 3). When levels of TAT and F1+2 were corrected for hemoconcentration across the hemofilter, there was no significant difference between the values measured at the filter outlet and the values corrected for hemoconcentration across the hemofilter, indicating the absence of thrombin generation across the hemofilter. 83

85 Chapter 5 Table 3. Maximal and minimal values of coagulation parameters during pre- and postdilution. Characteristic Predilution Postdilution before filter after filter before filter after filter PT min (s) 14.9 ± ± ± ± 0.7 PT max (s) 20.4 ± 3.6 a,c 15.7 ± 1.9 d 15.9 ± 2.0 a,d 13.9 ± 1.4 aptt min (s) 24 ± 5 32 ± 4 22 ± 6 25 ± 4 aptt max (s) 58 ± 13 a,c 37 ± 8 c 41 ± 10 b,c 31 ± 8 c TAT min ( g/l) 10 ± ± ± ± 10.2 TAT max ( g/l) 24.7 ± ± ± ± 20.4 F1+2 min (nmol/l) 0.7 ± ± ± ± 1.7 F1+2 max (nmol/l) 1.9 ± ± ± ± 4.7 Platelet count min (% baseline) 39 ± 10 a,c 59 ± 15 c 100 ± ± 0 Platelet count max (% baseline) 100 ± 0 91 ± ± 43 d 182 ± 83 GP-53 expressing platelets min (%) 4.9 ± ± ± ± 3.5 GP-53 expressing platelets max (%) 7.4 ± ± ± ± 4.8 a before vs after filter p < 0.01, b before vs after filter p < 0.05, c vs baseline p < 0.01, d vs baseline p < 0.05 PT, prothrombin time; aptt, acivated partial thromboplastin time; TAT, thrombin-antithrombin complex; F1+2, prothrombin fragment F1+2; GP-53 glycoprotein 53. Data represent mean ± SD. Platelet counts Because baseline values were higher in predilution mode, platelet counts were expressed as a percentage from baseline. During predilution, the platelet count measured before the filter decreased to 39 ± 10% baseline over 18 hours (p < 0.01), whereas measured after the filter, it decreased to 59 ± 15% baseline over 18 hours (p < 0.01). At all timepoints, platelet counts were significantly lower before the filter than after the filter (p < 0.01). When platelet counts were corrected for hemoconcentration across the hemofilter, actually measured platelet counts at the hemofilter outlet during predilution were significantly lower than predicted based on hemoconcentration occurring across the hemofilter. During postdilution, the platelet count measured before the filter did not change significantly over 18 hours, whereas measured after the filter, the platelet count increased to 143 ± 42% baseline at t = 0.5 h (p < 0.05), without a significant change thereafter. When platelet counts were corrected for hemoconcentration across the hemofilter, actually measured platelet counts at the hemofilter outlet during postdilution were not significantly different from the platelet counts predicted, based on hemoconcentration across the hemofilter. 84

86 Pre- versus postdilution CVVH Platelet activation In both modes, there was no change in the percentage of platelets expressing GP53, neither across the hemofilter, nor over time (Table 3). There was also no change in percentage platelets expressing either P-selectin or fibrinogen (data not shown). Thus, no signs of platelet activation were observed. Extracorporeal circuit pressures In Table 4, minimal and maximal pressures measured in the extracorporeal circuit are summarized for both modes. During postdilution, minimal levels of arterial, prefilter and transmembrane pressures were significantly higher than during predilution. Moreover, both minimal and maximal pressure drop over the hemofilter were significantly higher during postdilution. Interestingly, maximal transmembrane pressures were higher during predilution. All predilution runs were stopped because of ultrafiltrate reduction due to high transmembrane pressures. During postdilution, 3 out of 7 runs were stopped because of high arterial or prefilter pressures and only 4 out of 7 runs because of ultrafiltrate reduction due to high transmembrane pressures. Table 4. Extracorporeal circuit pressures during pre- and postdilution Predilution Postdilution P value PA min (mmhg) 27 ± ± 30 < 0.01 PA max (mmhg) 85 ± ± 102 < 0.05 PBE min (mmhg) 134 ± ± 87 < 0.05 PBE max (mmhg) 257 ± ± 89 NS PV min (mmhg) 73 ± ± 82 NS PV max (mmhg) 168 ± ± 74 NS P min (mmhg) 59 ± ± 1 < 0.01 P max (mmhg) 57 ± ± 14 < 0.05 TMP min (mmhg) 35 ± ± TMP max (mmhg) 360 ± ± 105 < PA, arterial pressure; min, minimal; mmhg, millimeter of mercury; max, maximal; PBE, prefilter pressure; PV, venous pressure; P, pressure drop over the hemofilter; TMP, transmembrane pressure. Data represent mean ± SD. Factors influencing circuit survival time Baseline platelet count was inversely related with circuit survival time. This effect was significant for postdilution (p < 0.05), but not for predilution (Figure 2). Other baseline coagulation parameters did not correlate with circuit survival time. During postdilution, maximal prefilter pressure was also inversely related with circuit survival 85

87 Chapter 5 time (r 2 = 0.73, p = 0.01). The relationship between baseline platelet count and maximal prefilter pressure during postdilution was linear (r 2 = 0.93, p = 0.001) (Figure 3). 80 predilution circuit survival time (h) r 2 = 0.68 p = 0.02 postdilution r 2 = 0.01 p = baseline platelet count (x10 9 /l) Figure 2. Correlation between baseline platelet count and circuit survival time during hemofiltration in predilution ( ) and postdilution ( ) mode. Data represent mean ± SD. 450 PBE max (mmhg) r 2 = 0.93 p = baseline platelet count (x10 9 /l) Figure 3. Correlation between baseline platelet count and maximal prefilter pressure (PBE) during postdilution hemofiltration. 86

88 Pre- versus postdilution CVVH Urea clearance Mean urea filter clearance during both predilution and postdilution was constant over time, ranging from 62.5 ± 13.8 ml/min at t = 0.5 h to 67.1 ± 3.5 ml/min at t = 18 h during predilution and from 59.8 ± 1.9 at t = 0.5 h to 59.9 ± 2.6 ml/min at t = 18 h during postdilution. There was no statistically significant difference in mean filter clearance between the two modes. Mean filter extraction ratios were 32 ± 0.02 and 30 ± 0.01 in predilution and postdilution respectively. Extracorporeal clearances were 44.6 ± 2.9 and 60.4 ± 1.7 ml/min for predilution and postdilution respectively (p < 0.01). Discussion In the present crossover study, predilution and postdilution did not differ with respect to extracorporeal circuit thrombogenesis. During postdilution however, minimal extracorporeal pressure levels were higher, and circuit survival time was inversely related with both baseline platelet count and maximal prefilter pressure. Interestingly, a linear relationship between the latter two parameters was found. In two previous studies, predilution was associated with an increased circuit life. Uchino et al. [5] found a median circuit life of 18 hours during predilution, as compared to 13 hours during postdilution. In this study, predilution was a significant independent predictor of increased filter life, together with platelet count and heparin dose. Van der Voort et al. [6] found a median circuit survival time of 45.7 hours during predilution, compared with 16.1 hours during postdilution, using a method very similar to ours. In the present crossover study, the superiority of predilution with respect to circuit survival time could not be confirmed. However, the sample size of our study was small, carrying the risk of a type II error. Indeed, during postdilution, a relative high number of hemofilters (2 out of 7) expired within one hour, compared with none during predilution. Moreover, baseline platelet count was significantly lower in the postdilution group, which might have attenuated the difference in circuit survival time between predilution and postdilution. During postdilution, the mean hematocrit increased across the hemofilter from 0.28 ± 0.05 to 0.38 ± 0.08 (p < 0.01), whereas during predilution, the diluted mean hematocrit of 0.18 ± 0.02 at the hemofilter inlet increased to 0.29 ± 0.04 at the hemofilter outlet (p < 0.01). However, this hemoconcentration across the hemofilter did not result in thrombogenesis, as no significant increase in levels of TAT and F1+2 was detected during either mode, neither across the hemofilter, nor over time. These results confirm the 87

89 Chapter 5 results of Stefanidis et al. [8], who did not find a significant difference in fibrinopeptide A and TAT complexes between patients with a hematocrit > 30% and patients with a hematocrit < 30% during postdilution hemofiltration. However, the present findings are in contrast with the results of Cardigan et al. [9], who found an increase in TAT levels over time in 8 out of 12 patients during hemofiltration, with an inverse relationship between TAT levels and circuit survival time. During postdilution, platelet counts increased significantly at t = 0.5 h, remaining constant thereafter. There was no significant difference in actually measured and predicted platelet counts at the hemofilter outlet, based on hemoconcentration across the hemofilter. During predilution however, platelet counts measured both before and after the filter decreased significantly over time. Moreover, actually measured platelet counts at the hemofilter outlet were significantly lower than predicted, based on hemoconcentration across the hemofilter. The reason for this disproportionate decrease in platelet count during predilution is unclear. No change in indicators of platelet activation was found, neither across the hemofilter, nor throughout the procedure. These findings confirm those of Kozek-Langenecker et al. [10], who demonstrated that the extent of platelet activation as measured by the monoclonal antibodies PAC-1 and anti-cd62, remained constant during 24 hours of hemofiltration. The hypothesis that platelet activation plays a role in the initiation of thrombosis during hemofiltration, is supported by the fact that circuit survival time can be prolonged by adding prostaglandin I2 or E1 to the anticoagulation with unfractionated heparin [11]. In the present study however, no evidence for an influence of platelet activation on circuit survival time was found. In the design of the present study, total flow across the hemofilter was kept constant at 200 ml/min in both modes, because we anticipated thrombogenesis to be proportionate to prefilter pressure and thus to flow rate across the hemofilter, based on the results of earlier studies. However, despite the equal flow across the hemofilter in both predilution and postdilution, minimal extracorporeal circuit pressures were higher during postdilution. During hemofiltration, the blood flow through the hollow fibers of the hemofilter is governed by the Hagen-Poiseuille equation: Qb = P/(8 L/ r 4 ), where Qb is blood flow, P is pressure drop over the hemofilter, is blood viscosity, L is hollow fiber length and r is hollow fiber radius [12]. As blood flow resistance (R) equals P/Qb, it is proportionate to blood viscosity, which is increased during postdilution, due to the increase in hematocrit [13]. To overcome the increased blood flow resistance in the hemofilter while maintaining a constant blood flow, the prefilter pressure during 88

90 Pre- versus postdilution CVVH postdilution increases. The pressure limit is more easily reached, causing the bloodpump to stop. Baldwin et al. [14] demonstrated that interruptions of the blood flow are likely to promote clotting of the CVVH circuit. They nicely showed that the frequency of medium intensity (34 66%) flow reductions per hour was inversely related with filter life. Moreover, this correlation was much stronger than that seen with the anticoagulation variables normally monitored during CVVH. High pressure circumstances entailing blood flow interruptions might have been the reason for shorter circuit survival times during postdilution in previous studies. Interestingly, maximal transmembrane pressures were higher during predilution. Indeed, all predilution runs were terminated because of ultrafiltrate reduction due to high transmembrane pressures, whereas during postdilution, only half of the runs were terminated for this reason. During predilution, the lower pressure profile might allow the system to run long enough to be limited by protein accumulation in the filter micropores, leading to elevation of the transmembrane pressure. During postdilution, we found an inverse relationship between circuit survival time and both baseline platelet count and maximal prefilter pressure. Moreover, a linear relationship between the latter two parameters was demonstrated. The observed inverse correlation between circuit survival time and baseline platelet count during postdilution, confirms our finding in a previous study [15]. To our knowledge however, this is the first time a linear relationship is demonstrated between baseline platelet count and maximal prefilter pressure during postdilution. There are a few possible explanations for this phenomenon. First, platelet count has been shown to be an independent denominator of blood viscosity [16]. Second, platelet cohesion is promoted by increased shear stress, which equals shear rate multiplied by viscosity [17-19]. Shear rate is a measure of how rapid fluid layers are flowing past each other and equals dv/dr, where V is the fluid velocity and r is the hollow fiber radius. The shear rate is zero at the vessel center and maximal at the vessel wall. When blood flows through the hollow fibers of the hemofilter, platelets are pushed towards the wall, which increases both the local platelet concentration and the shear stress exerted on these platelets [17,19]. It is conceivable that a higher baseline platelet count facilitates the platelet cohesion induced by increased shear stress. Moreover, the increase in platelet count during postdilution might further potentiate platelet cohesion, leading to increased blood flow resistance and hence increased prefilter pressure. In this study, mean extracorporeal urea clearance was 60.4 ± 1.7 ml/min during postdilution and 44.6 ± 2.9 ml/min during predilution. This difference was expected, as 89

91 Chapter 5 an ultrafiltration rate of 60 ml/min was used during both predilution and postdilution, whereas a higher blood flow was used during postdilution (200 vs 140 ml/min). Thus, according to the clearance calculation by David et al. [20], the anticipated urea clearance was 60 ml/min during postdilution and 42 ml/min during predilution (60 x 140/200). We found a higher urea clearance than expected during predilution, which may be explained by the fact that in the predicted clearance, we did not take into account hematocrit, plasma protein content and equilibrium distribution coefficient between red cells and plasma. Indeed, lower values of hematocrit and total plasma protein content and a higher urea gradient between red cells and plasma yield a higher urea clearance. David et al. [20] report a 6% higher clearance when these factors are taken into account. In the present study, the measured extracorporeal urea clearance was indeed 6% higher than calculated (44.6 ± 2.9 vs 42 ml/min). Determination of the optimal blood flow and substitution rate to obtain the highest clearance at the lowest rate of filter obstruction during predilution, remains a challenge for future studies. The impact of enhanced urea clearance on platelet function during postdilution is unclear, as the influence of uremia on platelet function is complex. Decreased thromboxane A2 production, abnormal intracellular calcium mobilization, increased intracellular cyclic AMP and cyclic GMP and abnormal aggregability have all been described in uremic platelets and may all contribute to defective platelet function in uremic patients [21]. In the present study, we studied only three markers of platelet actvation: P selectin, GP53 and fibrinogen. We did not find a difference in the three activation markers studied during either predilution or postdilution. Apparently, the enhanced urea clearance during postdilution did not influence the studied markers of platelet activation. However, the influence of enhanced urea clearance on other determinants of platelet function was not studied. Conclusion Predilution and postdilution did not differ with respect to extracorporeal circuit thrombogenesis. During postdilution, extracorporeal circuit pressures were significantly higher and circuit survival time was inversely related with both baseline platelet count and maximal prefilter pressure. Moreover, a linear relationship between the latter two parameters was demonstrated. This suggests that baseline platelet count has an important impact on maximal prefilter pressure and thus on circuit survival time during postdilution. For predilution, additional studies are needed to determine the optimal blood flow and substitution rate to obtain the highest clearance at the lowest rate of filter obstruction. 90

92 Pre- versus postdilution CVVH References Abramson S, Niles JL. Anticoagulation in continuous renal replacement therapy. Curr Opin Nephrol Hypertens 1999;8: Gabutti L, Marone C, Colucci G, et al. Citrate anticoagulation in continuous venovenous hemodiafiltration: a metabolic challenge. Intensive Care Med 2002;28: Horrow JC: Protamine. a review of its toxicity. Anesth Analg 1985;64: Kaplan AA. Predilution versus postdilution for continuous arteriovenous hemofiltration. Trans Am Soc Artif Intern Organs 1985;31:28-32 Uchino S, Fealy N, Baldwin I et al. Pre-dilution vs post-dilution during continuous venovenou hemofiltration: impact on filter life and azotemic control. Nephron Clin Pract 2003;94:c94-98 van der Voort PH, Gerritsen RT, Kuiper MA, et al. Filter run time in CVVH: pre-versus postdilution and nadroparin versus regional heparin-protamine anticoagulation. Blood Purification 2005;23: Maquelin KN, Nieuwland R, Lentjes EG, et al. Aprotinin administration in the pericardial cavity does not prevent platelet activation. J Thorac Cardiovasc Surg 2000;120: Stefanidis I, Heintz B, Frank D, et al. Influence of hematocrit on hemostasis in continuous venovenous hemofiltration during acute renal failure. Kidney International 1999;56:S51-S55 Cardigan RA, McGloin H, Mackie IJ, et al. Activation of the tissue factor pathway occurs during continuous venovenous hemofiltration. Kidney International 1999;55: Kozek-Langenecker SA, Spiss CK, Michalek-Sauberer A, et al. Effect of prostacyclin on platelets, polymorphonuclear cells, and heterotypic cell aggregation during hemofiltration. Crit Care Med 2003;31: Kozek-Langenecker SA, Spiss CK, Gamsjager T, et al. Anticoagulation with prostaglandins and unfractionated heparin during continuous venovenous hemofiltration: a randomized controlled trial. Wien Klin Wochenschr. 2002;114: Clark WR, Gao D. Properties of membranes used for hemodialysis therapy. Semin Dial 2002;15: Lee AJ, Mowbray PI, Lowe GD, et al. Blood viscosity and elevated intima-media thickness in men and women. The Edinburgh artery study. Circulation 1998;97: Baldwin I, Bellomo R, Koch B. Blood flow reductions during continuous renal replacement therapy and circuit life. Intensive Care Med 2004;30:

93 Chapter de Pont AC, Oudemans-van Straaten HM, Roozendaal KJ, et al. Nadroparin versus dalteparin anticoagulation in high-volume continous venovenous hemofiltration: A double-blind, randomized, crossover study. Crit Care Med 2000;28: Crowley JP, Metzger J, Assaf A, et al. Low density lipoprotein cholesterol and whole blood viscosity. Ann Clin Lab Sci 1994;24: Kroll MH, Hellums JD, McIntire LV, et al. Platelets and shear stress. Blood 1996;88: Goto S, Ikeda Y, Saldivar E, et al. Distinct mechanisms of platelet aggregation as a consequence of different shearing flow conditions. J Clin Invest 1998;101: Sakariassen KS, Hanson SR, Cadroy Y. Methods and models to evaluate shear-dependent and surface-reactivity dependent antithrombotic activity. Thromb Res 2001;104: David S, Boström M, Cambi V. Predilution hemofiltration.clinical experience and removal of small molecular weight solutes. Int J Artif Organs 1995;18: , 1995 Weigert AL, Schafer AI. Uremic bleeding: Pathogenesis and therapy. Am J Med Sci 1998;316:

94 Chapter 6 Discrepancies between observed and predicted CVVH removal of antimicrobial agents in critically ill patients and the effects on dosing Catherine C.S. Bouman, Hendrikus J.M. van Kan, Richard P. Koopmans, Johanna C. Korevaar,Marcus J. Schultz, Margreeth B. Vroom Intensive Care Med 2006;Oct 17 (Epub ahead of print)

95 Chapter 6 Abstract Objectives Drug dosing during continuous venovenous hemofiltration (CVVH) is based partly upon the CVVH clearance (Cl CVVH ) of the drug. Cl CVVH is the product of the sieving coefficient (SC) and ultrafiltration rate (Q uf ). Although it has been suggested that the SC can be replaced by the fraction of a drug not bound to protein (F up ), the F up values as reported in the literature may not reflect the protein binding in critically ill patients with renal failure. We compared the observed Cl CVVH (SC x Q uf ) with the estimated Cl CVVH (estimated F UP x Q uf ) determined the effect on the maintenance dose multiplication factor (MDMF). Design and setting Clinical study in a mixed ICU in a university hospital. Patients 45 oligoanuric patients on CVVH (2 L/h). Interventions Timed blood and ultrafiltrate samples. Measurements and main results Amoxicillin, ceftazidime, ciprofloxacin, fluconazole, metronidazole, and vancomycin were easily filtered (mean SC > 0.7), but not flucloxacillin (mean SC 0.3). Predicted and observed Cl CVVH corresponded only for fluconazole and metronidazole. The difference (mean and 95% CI) between observed and predicted MDMF was small for all drugs, with the exception of ceftazidime (0.25 ( 0.96 to 1.48)) and vancomycin (0.05 ( 1.34 to 1.45)). However, this difference was clinically relevant only for vancomycin, because of its narrow therapeutic index. Conclusions Dosing based on predicted CVVH removal provides an as reliable estimate than that based on observed CVVH removal, except for those antibiotics that have both a narrow therapeutic index and a predominantly renal clearance (e.g. vancomycin). 94

96 Antibiotics & CVVH Introduction The majority of critically ill patients receive some form of antimicrobial treatment during their stay in the intensive care unit (ICU). The importance of antimicrobial drug dose optimization in these patients has recently been highlighted [1]. While underdosing may lead to treatment failures and the selection of resistant microorganisms, overdosing may cause drug toxicities and higher costs. Correct dosing of antimicrobial drugs requires not only knowledge of several general pharmacokinetic parameters (such as absorption rate, bio-availability, distribution including protein and tissue binding, and renal and non-renal clearance) but also of extracorporeal clearance by renal replacement therapy if applied [2]. Continuous venovenous hemofiltration (CVVH) is widely used to control complications and manifestations of renal failure in the ICU. To calculate the drug maintenance dose for patients on CVVH the dose recommended for anuric patients not receiving renal support therapy has to be multiplied by a maintenance dose multiplication factor (MDMF) which is based on the total body clearance and the extracorporeal clearance [3]. The CVVH clearance (Cl CVVH ) of a drug is the product of the ultrafiltration rate (Q uf ) and the sieving coefficient (SC) for that drug. The SC is the ratio between the drug concentration in the ultrafiltrate and that in blood and is a measure for the ability of a drug to pass through a membrane and varies between 0 (for drugs that do not pass the membrane) and 1 (for drugs that pass the membrane freely). The SC is influenced by several factors, such as drug-membrane interaction, size and charge of molecules and the characteristics of the hemofilter, but the main determinant limiting drug sieving is protein binding [4]. It has been suggested that for predicting Cl CVVH, the SC can be replaced by the unbound protein fraction of a drug (F up ) [5,6]. However, the F up values reported in the literature are obtained in healthy volunteers and may not reflect the protein binding in critically ill patients [7]. Clinical data on the removal of antimicrobial drugs by CVVH are scarce. This is even more strongly the case when higher ultrafiltrate doses are applied, doses that are now recommended for critically ill patients with acute renal failure [8]. In critically ill patients with oliguric renal failure we compared the observed Cl CVVH (calculated from measured data) and the predicted Cl CVVH (calculated from the F up ) for seven frequently used antimicrobial drugs. In addition, we determined whether dose adjustment according to the predicted CVVH removal provides an as reliable estimate than that according to the observed CVVH removal. Parts of our results were previously published as an abstract [9]. 95

97 Chapter 6 Methods and materials Patients The study was conducted in a 28-bed multidisciplinary closed format ICU in a university hospital. We examined 45 oligoanuric critically ill patients on CVVH and receiving intravenous amoxycillin, ceftazidime, ciprofloxacillin, flucloxacillin, fluconazole, metronidazole, or vancomycin to treat a known or suspected infection (anuric n = 31, oliguric n = 13, i.e., diuresis < 400 ml in 12 h despite hemodynamic optization and highdose frusemide). During the sampling period we administered a single antimicrobial drug to 31 patients, two to 9 patients, and a combination of three to 5 patients. The antimicrobial agents were given in a variety of doses (Table 1). Of the 64 collection periods samples were collected at three time points in 55 (86%), at two time points in 8 (two each for ceftazidime, ciprofloxacin, and metronidazole, one each for flucloxacillin and fluconazole), and at a single time point in one (flucloxacillin). The observations in this study were made in the context of standardized protocol for routine patient care. The Medical Ethics Committee of our institution waived a formal approval procedure for the study. No drug-related adverse effects were reported during the study. Continuous venovenous hemofiltration Hemofiltration was performed with computer controlled fully automated hemofiltration machines (Diapact, B.Braun, Melsungen, Germany). Vascular access was obtained by cannulation of the femoral, jugular or subclavian vein using the Seldinger technique and a double lumen catheter (GamCath, Gambro, Hechingen, Germany). A 1.9 m 2 high-flux cellulose triacetate hollow-fiber membrane (in vivo cutoff 40 kda, ultrafiltration coefficient 37 ml/h/mmhg, SC for 2 microglobulin 0.81) was used (CT-190G, Baxter Healthcare Corporation, IL, USA). Hemofilter replacement was not standardized. Nevertheless, the hemofilter was not replaced within two hours prior to drug administration and the hemofilters were not replaced during sampling. The extracorporeal circuit was anticoagulated with heparin (Heparin Leo, Leo Pharma, Ballerup, Denmark). In case of severe contraindications for anticoagulation, hemofiltration was performed without anticoagulation. Blood flow rate was 150 ml/min and warmed substitution fluids (SH 19, B-Braun, Melsungen, Germany; SH 53-HEP, B-Braun, Melsungen, Germany) were administered in predilution at a flow rate of 2000 ml/h. In case a negative fluid balance was required the ultrafiltration flow was increased and the substitution flow was constant. 96

98 Antibiotics & CVVH Table 1. Sieving coefficient, protein unbound fraction and characteristics of patients by antimicrobial agent. Amoxycillin (n = 12) Ceftazidime (n = 7) Ciprofloxacin (n = 16) Flucloxacillin (n = 5) Fluconazole (n = 9) Metronidazole (n = 9) Vancomycin (n = 6) Dose (mg) Interval (h) Age (years) 61 ± ± ± ± 4 61 ± ± 8 66 ± 17 APACHE II a 24.5 ± ± ± ± ± ± ± 7.6 Albumin (g/l) a 20.9 ± ± ± ± ± ± ± 4.4 Filter life before sampling (h) a 22 (3 35) 17 (9 32) 11 (6 21) 16 (12 53) 15 (12 25) 22 (14 36) 24 (4 28 ) UF flow (ml/kg/h) a 29 ± 7 25 ± 7 27 ± 5 23 ± 7 27 ± 7 26 ± 3 32 ± 9 SC 0.71 ± ± ± ± ± ± ± 0.19 F up Values are mean ± SD or median and quartiles. a Differences among agents statistically nonsignificant. APACHE, acute physiology and chronic health evaluation at admission [27]; UF, ultrafiltrate; SC, sieving coefficient; Fup the protein unbound fraction (derived from the literature) [3,11]. 97

99 Chapter 6 Sample collection Blood samples were collected in glass tubes (Vacutainer Systems, Becton Dickinson, Plymouth, UK) from the afferent (prehemofilter), and efferent (posthemofilter) line of the extracorporeal circuit and from the ultrafiltrate line. Samples were collected at three different time points between two successive intravenous doses of antibiotics (at 2, 4 and 6 h for agents given every 6 h; at 2, 4 and 8 h for agents given every 8 h; and at 2, 6 and 10 h for agents given every 12 or 24 h). Blood and ultrafiltrate samples were stored on ice until centrifugation. Serum was prepared by centrifugation (3600 x g for 15 minutes). Serum and ultrafiltrate were stored at 80ºC, until analysis. Drug assays Amoxicillin, ciprofloxacin, ceftazidime, flucloxacillin, fluconazole, and metronidazole serum and ultrafiltrate concentrations were measured with validated methods using high-performance liquid chromatography with ultraviolet detection. These methods have been validated according to good laboratory practice guidelines providing an intermediate imprecision and inaccuracy below 10% in the measured range for all methods. Vancomycin concentrations were measured using a validated fluorescence polarization immunoassay (Axsym System, Abbott Laboratories, Abbott Park, Illinois, USA) with a maximum intermediate imprecision of 7% and a maximum inaccuracy of 6%. Calculations The prefilter serum concentration was multiplied by the dilution factor to correct for the dilution effects of predilution hemofiltration [10]. Dilution factor = Q b / (Q b + Q inf ) Where Q b is the blood flow rate and Q inf is the infusion rate of the substitution fluid. The following equations were used to calculate the predicted and observed Cl CVVH and the effect on drug dosing [3]. SC = 2 x C uf / (C pre + C post ) Observed Cl CVVH = SC x Q uf Predicted Cl CVVH = F up x Q uf Fr CVVH = Cl CVVH / (Cl CVVH + Cl anuric ) MDMF = 1 / (1-Fr CVVH ) 98

100 Antibiotics & CVVH where F up is the unbound fraction in healthy volunteers as reported in the literature, C uf the drug concentration in ultrafiltrate, C pre the drug concentration in prefilter serum corrected for predilution, C post the drug concentration in postfilter serum, Q uf the ultrafiltrate flow rate, Fr CVVH, the fractional CVVH clearance, Cl anuric the estimated total body clearance in anuric patients not receiving renal replacement therapy as reported in the literature [3,11] and MDMF the maintenance dose multiplication factor [3]. Statistical Analysis Data were analyzed using the Statistical Package for the Social Sciences for windows, version 11.0 (SPSS, Chicago IL, USA). Results are presented as mean ± SD or median and quartiles. Patient characteristics for each antibiotic were compared using one way analysis of variance for normally distributed data and Kruskal-Wallis one-way analysis of variance to compare nonnormally distributed data. Observed and predicted data were compared using Bland-Altmann analysis. The correlations between observed and predicted data were determined by Pearson s correlation technique. Results Predicted and observed CVVH clearance The SC did not change significantly over time for any of the antimicrobial drugs. Therefore the mean individual data were used for further analysis and are summarized in table 1. All agents were easily filtered (mean SC > 0.70) except for flucloxacillin. The SC of flucloxacillin was approximately 0.2 in 4 patients, but one patient had an unexpectedly high SC of 0.94). Figure 1a shows the correlation between observed and predicted Cl CVVH. The correlation between observed and predicted clearance was significant (p = 0.003) only when all drugs were combined, not for the individual antimicrobial drugs. Despite the nonsignificant correlation the difference between predicted and observed clearance for all drugs was small as depicted in the Bland-Altmann plots (Figure 1b), with the exception of ceftazidime, ciprofloxacin and vancomycin, because of the high interpatient variability in observed values. Predicted and observed maintenance dose multiplication factor The Cl anuric value as reported in the literature was used to calculate the observed and predicted Fr CVVH and MDMF (Table 2). The observed MDMF was significantly correlated with the predicted MDMF when all drugs were combined (p < 0.05) but not for the individual drugs. The differences between observed and predicted data are shown in figure 2. For amoxicillin, ciprofloxacin, flucloxacillin, and metronidazole the differences 99

101 Chapter 6 between observed and predicted values were small and a little more pronounced for fluconazole (Figure 2). In contrast, for vancomycin and ceftazidime the differences between observed and predicted MDMF showed wide interpatient variability (Figure 3). Discussion The present study in critically ill oligoanuric patients compared the observed Cl CVVH (i.e., clearance calculated from the observed SC) and the predicted Cl CVVH (i.e., clearance calculated from the F up derived from the published literature) for seven antimicrobial drugs in order to determine the clinical implications on drug dosing adjustment. All the studied agents were easily filtered (SC > 0.7) with the exception of flucloxacillin. The most important finding of our study is that for clinical practice dose adjustment according to the predicted CVVH removal provides an as reliable estimate than that according to the observed CVVH removal. However, because of interpatient variability observed in the clearance of many antibiotics, plasma levels should be monitored when CVVH is performed for those antibiotics that are eliminated predominantly by the kidney, and that have a low therapeutic threshold. Although there is interpatient variability between the observed and predicted Cl CVVH for some antibiotics, its effect on dosing strategies is not necessarily clinically relevant. Two factors are responsible for this apparent contradiction: the non-renal elimination route and the therapeutic range of antimicrobial agents. The elimination of a drug through CVVH is thought to be clinically relevant if more than 25% of the overall elimination of the drug is by this route [3]. Ciprofloxacin and flucloxacillin have an important nonrenal elimination route and therefore the Fr CVVH for both agents were extremely low and not affected by the large interindividual variations in observed Cl CVVH. This remains true as long as the nonrenal elimination routes are intact. The therapeutic range of a drug also plays an important role in the decision whether the difference between observed and predicted Cl CVVH is clinically important. The therapeutic range of ceftazidime is wide, and therefore we believe it is safe to use the predicted Cl CVVH for dose adjustment, despite the interpatient variability in observed Cl CVVH. In contrast, vancomycin dosing during CVVH should be based on frequent serum monitoring rather than on predicted clearances, because of its narrow therapeutic range and the interpatient variability in observed Cl CVVH, confirming the findings of previous studies [12,13]. 100

102 Antibiotics & CVVH 50 All drugs 50 Ceftazidime Amoxycillin Ciprofloxacin r = 0.37 p = r = 0.30 p = r = 0.32 p = r = 0.10 p = Flucloxacillin Fluconazole Metronidazole 50 Vancomycin r = 0.23 p = r = 0.03 p = r = 0.28 p = r = 0.06 p = Figure 1a. Correlation between observed (X-axis) and predicted (y-axis) hemofiltration clearance (ml/min) of seven antibiotics during CVVH. 101

103 Chapter All drugs Amoxycillin 30 Ceftazidime Ciprofloxacin Flucloxacillin Fluconazole Metronidazole Vancomycin Figure 1b. Bland-Altmann analysis of observed and predicted hemofiltration clearance (ml/min); X-axis, observed clearace (ml/min), Y-axis, difference between observed and predicted clearance (ml/min). 102

104 Antibiotics & CVVH Table 2. Predicted and observed hemofiltration clearance fraction and maintenance dose multiplication factor. Amoxycillin (n = 12) Ceftazidime (n = 7) Ciprofloxacin (n = 16) Flucloxacillin (n = 5) Fluconazole (n = 9) Metronidazole (n = 9) Vancomycin (n = 6) Cl nonrenal (ml/min) Fr CVVH Predicted 0.41 ± ± ± ± ± ± ± 0.00 Observed 0.37 ± ± ± ± ± ± ± 0.04 MDMF Predicted 1.70 ± ± ± ± ± ± ± 0.09 Observed 1.61 ± ± ± ± ± ± ± 1.32 Cl nonrenal, total body clearance in anuric patients not receiving dialysis treatment (as reported in the literature [3;11]); Fr CVVH, hemofiltration clearance; MDMF, maintenance multiplication factor (factor by which the usual maintenance dose for anuric patients must be multiplied [3]). 103

105 Chapter 6 3 All drugs 3 Ceftazidime 3 Vancomycin 3 Fluconazole Amoxicillin Ciprofloxacin Flucloxacillin Metronidazole Figure 2. Bland-Altmann analysis of observed and predicted maintenance dose multiplication factor (MDMF). X-axis, observed MDMF; Y-axis, difference between observed and predicted MDMF. 104

106 Antibiotics & CVVH Another notably finding in our study was the effective extracorporeal removal of fluconazole, exceeding the normal renal clearance and requiring a high daily maintenance dose. This finding was reported earlier in critically ill patients receiving hemodiafiltration [14] and a possible explanation for this phenomenon is that in normal kidneys fluconazole is not only excreted but also partly reabsorbed. Our study is the first to provide data on the removal of seven antimicrobial agents using a modern cellulose triacetate membrane and an ultrafiltration rate of approx. 2 L/h. Two other studies compared the observed and predicted drug removal through CVVH, however using different membranes and lower (< 13 ml/min) ultrafiltration rates [6,15]. In contrast to our findings, these studies report a good correlation between observed and predicted data for all the studied agents. Several factors may account for the discrepancies between our study and those of the two other studies, including differences in CVVH mode, drugs under study and case mix of the studied population. In addition, for some drugs the number of patients studied was smaller than in our study and in the study by Joos et al. the interpatient variability per drug is not reported [6,15]. We observed a high interindividual variability in the SC values of the studied drugs, in particular for ceftazidime, ciprofloxacin and vancomycin and to a lesser degree for amoxycillin and flucloxacillin. The SC characterizes the permeability of the hemofilter for a substance and should be more or less constant. The hemofilter used in this study should allow the removal of molecules up to a molecular weight of 40 kda. Although the studied agents have molecular weights far below the cutoff point, they are not freely filtered due, for example, to protein binding and the gradual formation of a protein layer on the membrane decreasing effective pore size. One possible reason for the high interindividual variability is that the hemofilter has important changes in sieving characteristics over time. In our patients filter life before sampling varied enormously but no significant correlation was found between filter life and SC (p = 0.20). Moreover, samples were collected at three different time points with two successive doses of antibiotics, but SC did not change significantly over time. Another possible explanation is the differences in the severity of SIRS/sepsis which may affect distribution and protein binding. All our patients had low albumin levels, but the difference among patients was not statistically significant. It is possible that the limited number of observations does not allow us to evaluate which factors, if any, influenced the wide interpatient variability. The small number of patients is an important limitation of our study, but the number of patients studied was higher than most other studies evaluating the CVVH removal 105

107 Chapter 6 of amoxicillin [15], ceftazidime [15-17], ciprofloxacin [15,18,19], fluconazole [20,21], vancomycin [12,13,15,22,23] and flucloxacillin [15,24]. Another limitation of our study is that we only studied one CVVH mode and our results cannot be extrapolated to other designs of renal replacement therapy. Factors such as dilution mode, amount of substitution fluids or ultrafiltration flow, combination with dialysis (hemodiafiltration) and filter material are likely to influence the results. In addition, we did not study the potential for adsorption of drugs to the membrane. In general, dosing guidelines do not account for adsorption effects. Adsorption of ceftriaxone and aminoglycosides was reported with the AN69 membrane [25,26], and adsorption of flucloxacillin was reported with the polyamide membrane [24]. To our knowledge, adsorption of antimicrobial agents to cellulose triacetate membranes has not been reported. It is also important to note that our findings on dose adjustment are based on anuric clearances as reported in the literature, but the assumption that anuric clearance of healthy volunteers is a reliable tool in critically ill remains unproved. However, data on nonrenal clearance in critically ill patients with renal failure are scarce and even the results from pharmacokinetic studies performed in critically ill patients cannot be generalized. In conclusion, our study in critically ill patients with renal failure and treated with CVVH and a cellulose triacetate membrane showed no significant correlation between the predicted and observed CVVH drug removal. However, for clinical practice dose adjustment according to the predicted CVVH removal provided an as reliable estimate than that according to the observed CVVH removal for amoxicillin, ceftazidime, ciprofloxacin, flucloxacillin, fluconazole, and metronidazole. In contrast, vancomycin dosing during CVVH should be based on frequent serum monitoring because of its narrow therapeutic range and the interpatient variability of the extracorporeal clearance. 106

108 Antibiotics & CVVH References Mehrotra R, De Gaudio R, Palazzo M. Antibiotic pharmacokinetic and pharmacodynamic considerations in critical illness. Intensive Care Med 2004;30: Subach RA, Marx MA. Drug dosing in acute renal failure: the role of renal replacement therapy in altering drug pharmacokinetics. Adv Ren Replace Ther 1998;5: Reetze-Bonorden P, Bohler J, Keller E. Drug dosage in patients during continuous renal replacement therapy. Pharmacokinetic and therapeutic considerations. Clin Pharmacokinet 1993;24: Lau AH, Kronfol NO. Determinants of drug removal by continuous hemofiltration. Int J Artif Organs 1994;17: Bohler J, Donauer J, Keller F. Pharmacokinetic principles during continuous renal replacement therapy: drugs and dosage. Kidney Int Suppl 1999;72:S24-S28. Golper TA, Wedel SK, Kaplan AA, et al. Drug removal during continuous arteriovenous hemofiltration: theory and clinical observations. Int J Artif Organs 1985;8: Bugge JF. Pharmacokinetics and drug dosing adjustments during continuous venovenous hemofiltration or hemodiafiltration in critically ill patients. Acta Anaesthesiol Scand 2001;45: Ronco C, Bellomo R, Homel P, et al. Effects of different doses in continuous venovenous haemofiltration on outcomes of acute renal failure: a prospective randomised trial. Lancet 2000;356: Bouman, C, Kan, H, Schultz, M, et al. The removal of antibiotics during continuous venovenous hemofiltration (CVVH) in critically ill patients with acute renal failure: measured versus estimated CVVH clearance (abstract). Critical Care 2005; 9(supplement):14 Clark WR, Ronco C. CRRT efficiency and efficacy in relation to solute size. Kidney Int Suppl 1999;S3-S7. Schetz M, Ferdinande P, van den Berg G, et al. Pharmacokinetics of continuous renal replacement therapy. Intensive Care Med 1995;21: Boereboom FT, Ververs FF, Blankestijn PJ, et al. Vancomycin clearance during continuous venovenous haemofiltration in critically ill patients. Intensive Care Med 1999;25: Matzke GR, O Connell MB, Collins AJ, et al. Disposition of vancomycin during hemofiltration. Clin Pharmacol Ther 1986;40: Yagasaki K, Gando S, Matsuda N, et al. Pharmacokinetics and the most suitable dosing regimen of fluconazole in critically ill patients receiving continuous hemodiafiltration. Intensive Care Med 2003;29:

109 Chapter Joos B, Schmidli M, Keusch G. Pharmacokinetics of antimicrobial agents in anuric patients during continuous venovenous haemofiltration. Nephrol Dial Transplant 1996;11: Matzke GR, Frye RF, Joy MS, et al. Determinants of ceftazidime clearance by continuous venovenous hemofiltration and continuous venovenous hemodialysis. Antimicrob Agents Chemother 2000;44: Traunmuller F, Schenk P, Mittermeyer C, et al. Clearance of ceftazidime during continuous venovenous haemofiltration in critically ill patients. J Antimicrob Chemother 2002;49: Bellmann R, Egger P, Gritsch W, et al. Pharmacokinetics of ciprofloxacin in patients with acute renal failure undergoing continuous venovenous haemofiltration: influence of concomitant liver cirrhosis. Acta Med Austriaca 2002;29: Malone RS, Fish DN, Abraham E, et al. Pharmacokinetics of levofloxacin and ciprofloxacin during continuous renal replacement therapy in critically ill patients. Antimicrob Agents Chemother 2001;45: Muhl E, Martens T, Iven H, et al. Influence of continuous veno-venous haemodiafiltration and continuous veno-venous haemofiltration on the pharmacokinetics of fluconazole. Eur J Clin Pharmacol 2000;56: Valtonen M, Tiula E, Neuvonen PJ. Effect of continuous venovenous haemofiltration and haemodiafiltration on the elimination of fluconazole in patients with acute renal failure. J Antimicrob Chemother 1997;40: Shah M, Quigley R. Rapid removal of vancomycin by continuous veno-venous hemofiltration. Pediatr Nephrol 2000;14: Uchino S, Cole L, Morimatsu H, et al. Clearance of vancomycin during high-volume haemofiltration: impact of pre-dilution. Intensive Care Med 2002;28: Meyer B, Ahmed eg, Delle KG, et al. How to calculate clearance of highly protein-bound drugs during continuous venovenous hemofiltration demonstrated with flucloxacillin. Kidney Blood Press Res 2003;26: Cigarran-Guldris S, Brier ME, Golper TA. Tobramycin clearance during simulated continuous arteriovenous hemodialysis. Contrib Nephrol 1991;93: Matzke GR, Frye RF, Joy MS, et al. Determinants of ceftriaxone clearance by continuous venovenous hemofiltration and hemodialysis. Pharmacotherapy 2000;20: Knaus WA, Draper EA, Wagner DP, et al. APACHE II: a severity of disease classification system. Crit Care Med 1985;13:

110 Chapter 7 Cystatin C in critically ill patients treated with continuous venovenous hemofiltration Marije C. Baas, Catherine S.C. Bouman, Frans J. Hoek, Raymond T. Krediet, Marcus J. Schultz Hemodialysis International 2006;10:S11-S15

111 Chapter 7 Abstract Background Assessment of residual renal function in critically ill patients with ARF treated with CVVH is difficult. Cystatin C is a low-molecular weight protein (13.3 kda) removed from the body by glomerular filtration. Its serum concentration has been advocated for assessment of renal function in patients with kidney disease. Objective To investigate whether the removal of CysC by CVVH is likely to influence its serum concentration. Patients and methods Concentrations of CysC were measured in three consecutive samples in 18 patients with oliguric ARF treated with predilution CVVH (2 L/h). Samples were taken from the afferent and efferent blood lines and from the ultrafiltrate line. Results Concentrations of CysC did not change during the time interval studied. The mean serum concentrations of CysC were 2.25 ± 0.45 mg/l in the afferent and 2.19 ± 0.56 mg/l in the efferent samples (NS); ultrafiltrate concentrations of CysC were 1.01 ± 0.45 mg/l. The sieving coefficient of CysC was 0.52 ± 0.20; clearance of CysC was 17.3 ± 6.6 ml/min; removed quantity of CysC averaged 2.13 mg/h. Conclusion During predilution CVVH (2 L/h) the removed quantity of CysC is less than 30% of its production and no rapid changes in its serum concentration are observed. Therefore CVVH (2 L/h) is unlikely to influence serum concentrations of CysC significantly, which suggests that it can be used to monitor residual renal function during CVVH. 110

112 Cystatin C & CVVH Introduction Critically ill patients are at risk of acquiring ARF. Its incidence is 15% to 20% of all intensive care admissions, of which 4% to 6% require some form of RRT [1]. During RRT, adequate monitoring of renal function is severely hampered because plasma creatinine and urea are removed by the hemofilter. This problem could be overcome by the use of CysC. Cystatin C has been advocated as a marker of renal function. It is a low-molecular-weight protein (13.3 kda), produced at a constant rate by nucleated cells [2] and removed from the body by glomerular filtration [3]. Cystatin C production is not influenced by gender, age, bodyweight or muscle mass [4,5] and several studies have shown that CysC is a more sensitive indicator of mild reductions of renal function than creatinine [5-11]. Moreover, CysC was not removed during IHD [12] or CVVHDF [13]. However, critically ill patients with ARF will often receive CVVH. In contrast to hemodialysis, removing substances by diffusion, clearance of solutes with hemofiltration is achieved by convection (ultrafiltration) and adsorption [14]. With the use of conventional membranes convection is associated with higher removal of middle- and high-molecular-weight-molecules. The aim of this study was to investigate whether the removal of CysC by CVVH is likely to influence its serum concentration. If not, serum concentrations of CysC can be used as a marker for GFR in patients treated with CVVH. Methods Patients The present study was conducted as part of a wider study on the effect of CVVH on antimicrobial dosing [15]. The study was conducted in a 28-bed multidisciplinary closedformat intensive care in a university hospital and was performed in accordance with the guidelines of the local ethics committee. Consecutively admitted critically ill patients who required CVVH for oliguric ARF of any cause, in whom antimicrobial therapy was started to treat a known or suspected infection, were included. Continuous venovenous hemofiltration Hemofiltration was performed with computer-controlled fully automated hemofiltration machines (Diapact, Braun, Melsungen, Germany). Vascular access was obtained by cannulation of the femoral, jugular or subclavian vein using the Seldinger technique and a double lumen catheter (GamCath, Gambro, Hechingen, Germany). A 1.9 m 2 cellulose 111

113 Chapter 7 triacetate hollow fiber membrane (in vitro cutoff 60 kda, ultrafiltration coefficient 37 ml/h/mmhg, SC for 2-microglobulin 0.81) was used (CT-190G, Baxter Healthcare Corporation, IL, USA). The extracorporeal circuit was anti-coagulated with heparin (Heparin Leo, Leo Pharma, Ballerup, Denmark). In case of severe contraindications for anticoagulation, hemofiltration was performed without anticoagulation. Blood flow rate was 150 ml/min and warmed substitution fluids (SH 19, B-Braun; SB 53-HEP, B-Braun) were administered in predilution mode at a flow rate of 2 L/h. Samples Samples were obtained from the afferent (prehemofilter) and efferent (posthemofilter) line of the extracorporeal circuit and from the ultrafiltrate line. They were collected at three different time points, 1.5 to 4 h apart, between two successive intravenous doses of antibiotics. Serum and ultrafiltrate were stored at 80 C. Assay Cystatin C was measured with the N Latex Cystatin C test kit, a particle-enhanced immunonephelometric method, on a BN ProSpec analyser (Dade Behring, Leusden, the Netherlands). Normal serum CysC values range from 0.50 to 0.96 mg/l. Urea and creatinine concentrations were measured by standard clinical chemical methods. Calculations The prefilter serum concentration was multiplied by the dilution factor to correct for the dilution effects of predilution HF. Dilution factor = Q b / (Q b + Q inf ) Where Q b is the blood flow rate and Q inf is the infusion rate of the substitution fluid. The following equations were used to calculate the SC, CVVH clearance (Cl CVVH ) and total mass removed in ultrafiltrate (M uf ): SC = 2 x C uf / (C aff + C eff ) Cl CVVH (ml/min) = SC x Q uf M uf (mg/h) = C uf x Q uf Where C uf is the concentration in ultrafiltrate, C aff and C eff are the concentrations in the afferent and efferent blood line respectively, and Q uf de ultrafiltrate flow rate. 112

114 Cystatin C & CVVH Statistics Data were analyzed using the Statistical Package for the Social Sciences for Windows version 11 (SPSS, Chicago IL, USA). Results are presented as mean ± SD. To examine changes over time, we used linear mixed models. This analysis studies average changes in subjects, taking into account the association between variables for individual patients at separate time points. To compare afferent and efferent values, we used the paired t- test. Differences at the level of p < 0.05 were considered to be statistically significant. Results Eighteen patients were included (Table 1). For each patient, 3 afferent, 3 efferent and 3 ultrafiltrate samples were available. In 1 patient, only 2 afferent samples were available and in 2 patients only 2 efferent samples were present. Figure 1 shows the afferent CysC concentrations in the 3 consecutive samples. Changes over time were not statistically significant for CysC, creatinine and urea. Therefore, the mean individual data were used for further analysis and are summarized in Table 2. Figure 2 shows the individual data for CysC. The SC for creatinine averaged 0.9 and that of urea 1.0. The mean SC for CysC was When corrected for possible incomplete mixing by using the SC of urea, which should be 1.0 by definition, a value of 0.52 ± 0.20 was found. Likewise, the corrected CysC hemofilter clearance was 17.3 ± 6.6 ml/min (Figure 3). Serum CysC concentration was 2.21 ± 0.48 mg/l in the 12 patients receiving corticosteroids and 2.34 ± 0.41 mg/l in the others (NS). Table 1. Patient characteristics (n = 18). Male / Female 9 / 9 Age (years) 62.0 ± 14.0 Weight (kg) 79.2 ± 18.2 Height (cm) ± 8.4 Urine production (ml/24 h) 84.7 ± 98.8 APACHE-II score 21.7 ± 6.4 Admission type Medical 10 Surgical 8 Corticoid therapy 12 Values are mean ± SD or number. 113

115 Chapter Cystatin C (mg/l) I II III mean Figure 1. CysC concentrations in three consecutive afferent samples (horizontal lines represent mean ± SD). No statistically significant differences were observed over the three collection periods (P = 0.334) Cystatin C (mg/l) afferent line efferent line Figure 2. Mean individual CysC concentrations in the afferent and efferent samples 1n 18 patients. 114

116 Cystatin C & CVVH Table 2. Concentrations, sieving coefficient, hemofilter clearances and removed quantity of CysC, creatinine and urea. (Values are mean ± SD). CysC Creatinine Urea Afferent concentration 2.25 ± 0.45 mg/l 166 ± 99 mol/l 13.0 ± 5.6 mmol/l Efferent concentration 2.19 ± 0.56 mg/l 154 ± 90 mol/l 12.4 ± 5.3 mmol/l Ultrafiltrate concentration 1.01 ± 0.45 mg/l 160 ± 124 mol/l 13.5 ± 9.0 mmol/l Sieving coefficient 0.47 ± ± ± 0.45 Clearance 15.8 ± 6.5 ml/min 30.1 ± 14.9 ml/min 33.3 ± 15.0 ml/min Removed quantity 2.13 ± 0.95 mg/h 340 ± 263 mol/h 28.5 ± 19.0 mmol/h 25 Clearance (ml/min) Figure 3. Mean corrected clearance of CysC by CVVH. Clearance was calculated from the ureacorrected sieving coefficient (horizontal lines represent mean ± SD). Discussion In the present study, the corrected SC for CysC is 0.52 and the CVVH clearance is 17 ml/min. Adsorptive removal is unlikely because there is no difference in CysC level between the afferent and efferent concentrations. The removed quantity averages 2.13 mg/h. Data from the literature show that the generation rate of CysC is 7.44 ± 1.44 mg/h [16]. Consequently, CVVH CysC removal is less than 30% of its generation and unlikely to influence serum CysC levels in a clinically significant way. This is confirmed by the absence of rapid changes in the CysC concentrations in individual patients. Our study is the first study to evaluate the removal of CysC during CVVH in critically ill patients with ARF. Balik et al. [13] studied the effects of CVVHDF with polysulfone membranes in critically ill patients and concluded that CysC is not removed during 115

117 Chapter 7 CVVHDF to a significant extent. However, solute removal during hemodialysis is based on diffusion and not convection as in CVVH. Our study has several limitations. The conclusion is based on the generation rate of CysC as reported in the literature in non critically ill patients [16]. Several conditions may have an effect on CysC levels, in particular thyroid disease [17,18] and the use of corticosteroids [19-21]. It is possible that the generation rate of CysC is influenced by critical illness; however, this would affect our conclusion only in case of a reduced generation rate. In our study the difference in CysC levels between the patients with and without corticosteroids is not statistically significant; however, the number of patients per group is small. We studied one filter and one ultrafiltration rate in the predilution mode. Membrane material and pore size might affect the SC for CysC. Convective removal of CysC was reported earlier during in vitro hemofiltration [22]. In that study, the SC for CysC was somewhat higher than in our study, most likely because high-cutoff membranes were used (in vitro cutoff 100 kda). Moreover, adsorptive removal might be more prominent with other filters, in particular the polyacrilonitrile filter [23]. In our study, applying an ultrafiltration rate of 2 L/h, the removal of CysC is not likely to affect serum levels. However, during the application of higher ultrafiltration rates, larger quantities of CysC will be removed and this may affect its concentration. Conclusion During predilution CVVH (2 L/h), CysC is removed from the circulation; however, the removed quantity is less than 30% of its production. Therefore, CVVH (2 L/h) is unlikely to influence serum concentrations of CysC, which suggests that it can be used to monitor residual renal function during CVVH. 116

118 Cystatin C & CVVH References Block CA, Manning HL. Prevention of acute renal failure in the critically ill. Am J Respir Crit Care Med 2002;165: Filler G, Bokenkamp A, Hofmann W, et al. Cystatin C as a marker of GFR--history, indications, and future research. Clin Biochem 2005;38:1-8. Tenstad O, Roald AB, Grubb A, et al. Renal handling of radiolabelled human cystatin C in the rat. Scand J Clin Lab Invest 1996;56: Donadio C, Lucchesi A, Ardini M, et al. Cystatin C, beta 2-microglobulin, and retinolbinding protein as indicators of glomerular filtration rate: comparison with plasma creatinine. J Pharm Biomed Anal 2001;24: Vinge E, Lindergard B, Nilsson-Ehle P, et al. Relationships among serum cystatin C, serum creatinine, lean tissue mass and glomerular filtration rate in healthy adults. Scand J Clin Lab Invest 1999;59: Coll E, Botey A, Alvarez L, et al. Serum cystatin C as a new marker for noninvasive estimation of glomerular filtration rate and as a marker for early renal impairment. Am J Kidney Dis 2000;36: Dharnidharka VR, Kwon C, Stevens G. Serum cystatin C is superior to serum creatinine as a marker of kidney function: a meta-analysis. Am J Kidney Dis 2002;40: Harmoinen AP, Kouri TT, Wirta OR, et al. Evaluation of plasma cystatin C as a marker for glomerular filtration rate in patients with type 2 diabetes. Clin Nephrol 1999;52: Hoek FJ, Kemperman FA, Krediet RT. A comparison between cystatin C, plasma creatinine and the Cockcroft and Gault formula for the estimation of glomerular filtration rate. Nephrol Dial Transplant 2003;18: Price CP, Finney H: Developments in the assessment of glomerular filtration rate. Clin Chim Acta 2000;297: Randers E, Erlandsen EJ. Serum cystatin C as an endogenous marker of the renal function - a review. Clin Chem Lab Med 1999;37: Campo A, Lanfranco G, Gramaglia L, et al. Could plasma cystatin C be useful as a marker of hemodialysis low molecular weight proteins removal? Nephron Clin Pract 2004;98:c79-c82. Balik M, Jabor A, Waldauf P, et al. Cystatin C as a Marker of Residual Renal Function during Continuous Hemodiafiltration. Kidney Blood Press Res 2005;28: Bellomo R, Ronco C. Continuous renal replacement therapy: continuous blood purification in the intensive care unit. Ann Acad Med Singapore 1998;27:

119 Chapter Bouman CSC, Kan HJM, Schultz MJ, et al. The removal of antibiotics during continuous venovenous hemofiltration (CVVH) in critically ill patients with acute renal failure: measered versus estimated CVVH clearance. Crit Care (Abstract) 2005; 9,S14. Sjostrom P, Tidman M, Jones I. Determination of the production rate and non-renal clearance of cystatin C and estimation of the glomerular filtration rate from the serum concentration of cystatin C in humans. Scand J Clin Lab Invest 2005;65: den Hollander JG, Wulkan RW, Mantel MJ, et al. Is cystatin C a marker of glomerular filtration rate in thyroid dysfunction? Clin Chem 2003;49: Fricker M, Wiesli P, Brandle M, et al. Impact of thyroid dysfunction on serum cystatin C. Kidney Int 2003;63: Bjarnadottir M, Grubb A, Olafsson I. Promoter-mediated, dexamethasone-induced increase in cystatin C production by HeLa cells. Scand J Clin Lab Invest 1995;55: Cimerman N, Brguljan PM, Krasovec M, et al. Serum cystatin C, a potent inhibitor of cysteine proteinases, is elevated in asthmatic patients. Clin Chim Acta 2000;300: Poge U, Gerhardt T, Bokenkamp A, et al. Time course of low molecular weight proteins in the early kidney transplantation period--influence of corticosteroids. Nephrol Dial Transplant 2004;19: Mariano F, Fonsato V, Lanfranco G, et al. Tailoring high-cut-off membranes and feasible application in sepsis-associated acute renal failure: in vitro studies. Nephrol Dial Transplant 2005;20: Bouman CS, van Olden RW, Stoutenbeek CP. Cytokine filtration and adsorption during pre- and postdilution hemofiltration in four different membranes. Blood Purif 1998; 16:

120 Chapter 8 Cytokine filtration and adsorption during predilution and postdilution hemofiltration in four different membranes Catherine S.C. Bouman, Rudolf W. van Olden, Christiaan P. Stoutenbeek Blood Purif 1998;16:

121 Chapter 8 Abstract Introduction In the present in vitro study we investigated filtration and adsorption of TNF, IL-6 and IL-8 during predilution and postdilution hemofiltration with polysulfone, polyacrylonitrile, polyamide and cellulose triacetate membranes. Results Median SC for all membranes were 0.0 for TNF, below 0.15 for Il-6 and below 0,15 for IL-8 during postdilution hemofiltration. Differences in SC between filtration modes were less than Maximal differences in SC between membranes were 0.11 for IL-6, 0.0 for TNF, and 0.11 for IL-8. Progressive decrease of cytokine concentrations, was identical between the two filtration modes and most pronounced with the polyacrylonitrile membrane (reduction 77% for IL-6, 39% for TNF and 95% for IL-8 after 4 hours of hemofiltration. The relative contribution of adsorption to reduction of cytokines was 100% for TNF for all membranes, between 53% (cellulose triacetate) and 83% (polyacrylonitrile) for IL-6, and for IL-8 between 0% (polysulfone) and 100% (polyacrylonitrile). Conclusion Reduction of TNF, IL-6 and IL-8 was most impressive with the polyacrylonitrile membrane after 4 hours of hemofiltration and was largely due to adsorption. Adsorption of TNF, IL-6 and IL-8 was also seen with the other membranes. None of the membranes filtered TNF. Sieving of IL-6 and IL-8 was low with all membranes with only marginal differences between membranes or between filtration modes. 120

122 Cytokine removal & CVVH Introduction Continuous hemofiltration permits efficient control of fluid balance and azotemia in patients with renal failure as part of septic multiple organ dysfunction syndrome [1]. Additional beneficial effects of continuous hemofiltration have been ascribed to clearance of toxic mediators [2-4]. TNF and IL-6 were found in the ultrafiltrate of septic shock patients in several studies, although reported cytokine clearances varied enormously in these studies [5-9]. Differences in cytokine clearances can partly be explained by different hemofiltration techniques, ultrafiltration rates, hemofilters and cytokine assays. Convection (hemofiltration) is preferable to diffusion (dialysis) to eliminate cytokines, as the relatively high molecular weights of cytokines limit the rate of diffusion across the membrane. Predilution hemofiltration might be superior to postdilution hemofiltration for middle-molecular-weight clearance, due to a reduced hyperviscosity within the filter, improving the SC [10]. Furthermore, cytokine filtration and adsorption probably also depend on intrinsic membrane properties (e.g. diffusivity, porosity and sorbent properties) [11]. In vitro studies may be more appropriate than in vivo studies to study cytokine clearance and membrane adsorption, because septic patients do not have a steady-state production of cytokines. In vitro studies using saline test solutions and recombinant cytokines are probably not representative for the clinical situation. Fouling of the membrane and binding of cytokines to proteins are important processes limiting the filtration of cytokines. Both processes will not occur when only saline is used. Furthermore, recombinant cytokines may behave differently than natural cytokines, due to modifications in glycosylation and phosphorylation. The use of whole blood as test solution has the disadvantage of possible membrane-induced cytokine production by leukocytes. In the present in vitro study, using erythrocytes and plasma as test solution, we investigated the filtration and adsorption of natural human TNF, IL-6 and IL-8 in four different membranes. In addition we compared the effect of predilution and postdilution on hemofiltration and adsorption. Methods Hemofiltration was conducted in vitro at room temperature. The experimental set-up consisted of a blood reservoir, an ultrafiltrate reservoir, a hemofilter and a computercontrolled fully automated hemofiltration machine (BM 25, Baxter, Unterschei heim, Germany). The venous outflow was returned to the blood reservoir. During the experiment zero-balanced hemofiltration was performed with identical rates of ultrafiltration and substitution (SH 19, Schiwa, Glandorf, Germany). The pumps were calibrated to deliver 121

123 Chapter 8 a blood flow of 150 ml/min and an ultrafiltrate flow of 30 ml/min. Hemofiltration of the test solution was started after priming the entire circuit with a heparin-containing saline solution and continued for 240 min. Filters were studied during predilution and postdilution hemofiltration, using a new filter and tubing set for each experiment. Test solution Two units of outdated packed red blood cells were added to two units of fresh frozen plasma. To this solution 20 ml of the cytokine preparation and 50,000 IU of heparin (Leo pharmaceutical products BV, Weesp, Netherlands) were added. Test solution volumes varied between 900 and 1200 ml. The solution was continuously stirred using a magnet iron. Preparation of cytokines Venous blood of a healthy volunteer was collected in Sodium heparin (Leo pharmaceutical products BV, Weesp, Netherlands) and diluted 10 times in endotoxin-free ISCoves modified Dulbeccos medium (IMDM, Biowhittaker). After addition of Staphylococcus aureus Cowan I strain (pansorbin, Calbiochem, La Jolla, Ca, USA)) at a final dilution of 1/1000, the blood was incubated for 24 h at 37 C in a CO 2 incubator. The supernatant was harvested by centrifugation. Concentrations of TNF, IL-6 and IL-8 were 7, 20 and 150 ng/ml, respectively. Membranes Four hollow fiber biocompatible membranes with a cutoff point of approximately 30 kda were studied: a polyamide membrane (Polyflux 11, Gambro, 1,1 m 2 ), a polyacrylonitrile membrane (AN 69, Hospal, 1,3 m 2 ), a cellulose triacetate membrane (CT90, Baxter, 0,9 m 2 ) and a polysulfone membrane (F60, Fresenius, 1,2m 2 ). The ultrafiltration coefficients of the four membranes were: Polyflux 11, 42 ml/h/mmhg (bovine blood Ht 0.32, TMP 200 mmhg); F60, 40 ml/h/mmhg (human blood Ht 0.28, TMP 200 mmhg); AN69, 40 ml/h/mmhg (bovine blood Ht 0.32, TMP 200 mmhg); CT 90, 19 ml/h/mmhg (Bovine blood Ht 0.25, TMP 100 mmhg). Sampling After a 30-min stabilisation period a sample was taken from the blood reservoir. Thereafter hemofiltration was started and samples (2 ml) were taken from the arterial, venous and ultrafiltrate line at 5, 10, 15, 60, 120 and 180 min. At 240 min samples were taken from the blood and ultrafiltrate reservoir. After centrifugation serum was frozen at 70 C until assayed. 122

124 Cytokine removal & CVVH Cytokine-assays TNF, IL-6 and IL-8 were determined by ELISA (Pelikine Compact kit for TNF, IL-6 and IL-8, CLB, Amsterdam). Inter- and intra-assay variations are <5%. Detection limits are 0.5 pg/ ml for IL-6, 2 pg/ml for TNF and 2 pg/ml for IL-8. According to the manufacturers these assays do not express any cross-reactivity to each other. Both monomeric and trimeric forms of TNF are detected in this assay. Calculations The following equations were used to calculate the SC, ultrafiltrate clearance, total mass loss and adsorption: SC = Cuf C p Cl (ml/min) = SC x Q uf M P (pg) = (V P x C P ) t0 (V P x C P ) t240 M uf (pg) = (V uf x C uf ) Ad (pg) = M p M uf where C uf is the solute concentration in ultrafiltrate, C P the solute concentration in plama, Cl the ultrafiltrate clearance, M P the total mass loss of solute after 4 h of hemofiltration treatment, V p the total volume in the blood reservoir, M uf the total solute mass in the ultrafiltrate after 4 h of hemofiltration treatment, V f the total filtrate volume after 4 h of hemofiltration, and Ad is adsorption of solute in the membrane. The plasma concentration during predilution hemofiltration was corrected using the dilution factor: Dilution factor = Q b Q b +Q inf where Q b is the blood flow rate and Q inf is the infusion flow rate of the substitution fluid. 123

125 Chapter 8 Predilution HF Postdilution HF TNF TNF % initial concentration % initial concentration time (min) time (min) IL-6 IL % initial concentration % initial concentration time (min) time (min) 124

126 Cytokine removal & CVVH IL-8 IL % initial concentration % initial concentration time (min) time (min) Figure 1. Individual reduction of TNF, IL-6 and IL-8 concentrations during hemofiltration comparing 4 different membranes both in the prdilution and postdilution mode. TNF, tumor necrosis factor; IL-6, interleukin-6; IL-8, interleukin-8. -x- Polyamide; - - cellulose triacetate; - - polyacrylonitrile; - - polysulfone. Statistical analysis Values of SC at different time points during treatment are given as median and ranges. The Wilcoxon rank sum test for unpaired samples was used to compare SC among the four membranes and among the different cytokines. The decrease of cytokine concentration was expressed as area under the curve. Predilution hemofiltration was compared with postdilution hemofiltration using the Wilcoxon rank sum test for unpaired samples. A p-value of < 0.05 was considered statistically significant. Results Median concentrations in the test solution were: TNF 76 (range 32 97) pg/ml, IL (range ) pg/ml, IL-8, 241 (range ) pg/ml. The decline in concentrations of TNF, IL-6 and IL-8 during 4 h of HF with four different membranes is shown in figure 1, both for the predilution and postdilution mode. The elimination rates were different for the three cytokines studied. The four membranes behaved differently in their ability to remove cytokines from the reservoir. Removal of TNF, IL-6 and IL-8 was highest with the polyacrylonitrile membrane (46, 74 and 93% reduction during 4 h predilution hemofiltration and 31, 79 and 96% reduction during 125

127 Chapter 8 postdilution hemofiltration, respectively). The smallest reduction of TNF was seen with the polysulfone membrane (7% in predilution, 17% in postdilution). The smallest IL-6 and IL-8 reduction was seen with the polyamide membrane (17 and 50% in predilution and 24 and 46% in postdilution, respectively). Predilution % reduction Polyamide Polyacrilonitrile Cellulose-triacetate Polysulfone TNF IL-6 IL-8 TNF IL-6 Il-8 TNF IL-6 IL-8 TNF IL-6 IL-8 Postdilution % reduction Polyamide Polyacrilonitrile Cellulose-triacetate Polysulfone TNF IL-6 IL-8 TNF IL-6 IL-8 TNF IL-6 IL-8 TNF IL-6 IL-8 Figure 2. Reduction (filtration and adsorption) of TNF, IL-6 and IL-8 after 4 h of hemofiltration. Black bars represent cytokine reduction due to adsorption, grey bars represent cytokine reduction due to filtration and light grey bars represent cytokine production. 126

128 Cytokine removal & CVVH Figure 2 shows the relative contribution of adsorption and filtration to the total reduction of cytokines after 4 h of hemofiltration for each membrane. The reduction is presented as the percentage of the initial cytokine concentration. For all membranes the reduction in TNF after the 4-hour study period was completely explained by adsorption. The relative contribution of adsorption to IL-6 reduction was 75% for the polyamide membrane, 83% for the polyacrylonitrile membrane, 53% for the CT90 membrane and 58% for the polysulfone membrane. The relative contribution of adsorption to IL-8 reduction was 40% for the polyamide membrane, 100% for the polyacrilonitrile membrane, 24% for the CT90 membrane and 0% for the polysulfone membrane, Production of IL-8 was seen in the polysulfone membrane. Table 1 shows the SC of each cytokine for the four membranes, both during predilution and postdilution hemofiltration. The SC of TNF for all membranes was 0 at all time points and both in the predilution and postdilution HF mode. During the 4-hour study period the polysulfone and cellulose triacetate membranes had a significantly higher SC for IL-6 than the polyacrylonitrile and polyamide membranes. The SC for IL-8 was comparable among the polysulfone, cellulose triacetate and polyamide membranes. The polyacrylonitrile membrane had a significantly lower SC for IL-8 than the other three membranes. Table 1. Sieving coefficients of TNF, IL-6 and IL-8 during 4 h of predilution (pre) and postdilution (post) hemofiltration with four different membranes. TNF 0 (0 0) IL ( ) Polyamide Cellulose triacetate Polyacrilonitrile Polysulfone Pre Post Pre Post Pre Post Pre Post 0 (0 0) 0.02 ( ) 0 (0 0) 0 (0 0) ( ) ( ) 0 (0 0) 0.06 ( ) 0 (0 0) 0.04 ( ) 0 (0 0) 0 (0 0) ( ) ( ) IL ( ) 0.09 ( ) 0.11 ( ) 0.10 ( ) 0.0 ( ) 0.0 ( ) 0.12 ( ) 0.13 ( ) Values are presented as median and ranges. Discussion In this study a reduction of TNF, IL-6 and IL-8 was seen with all membranes after 4 h of hemofiltration. The highest reduction in cytokines was observed with the AN69 membrane which was largely due to adsorption. Adsorption of TNF, IL-6 and IL-8 was also seen with the other membranes. Adsorptive removal by the membranes could have been overestimated in this study. The mass balance technique used to estimate 127

129 Chapter 8 adsorptive cytokine removal assumes that the difference between the mass appearing in the ultrafiltrate and the mass lost from the reservoir is solely due to adsorption to the membrane itself. Erythrocytes moderately bind IL-8 and proteins in the fresh frozen plasma could also have bound cytokines [12,13]. This process of binding to erythrocytes and proteins, however, will mainly be present during the stabilization period before the start of the hemofiltration procedure. None of the membranes filtered TNF within the study period. Filtration of IL-6 and IL-8 was low with all membranes with only marginal differences among the membranes and between the two filtration modes. The cellulose triacetate and polysulfone membranes filtered more IL-6 and IL-8 than the AN69 and polyamide membranes. Still, even with the cellulose triacetate and polysulfone membranes ultrafiltrate clearance was extremely low (approximately 2 to 3 ml/min). In this study relatively large membranes were used on a small test volume. This may have resulted in underestimation of the transmembrane removal of cytokines by sampling at a time when secondary membrane formation was not yet complete. Despite low clearances in the present study an appreciable reduction of IL-6 and IL-8 was achieved (>30% reduction) due to the relatively high filtration rate of 1800 ml/h on the small testsolution volume of approximately 1 liter. Our findings are in accordance with previous in vitro studies that also found remarkable adsorption of TNF, most pronounced with the AN69 membrane [5,14]. However, in these studies, sieving of TNF appeared in less than 30 min after the membrane was saturated and this was not the case in our study. These differences are probably the result of the use of monomeric recombinant TNF (17kDa) in the study of v.bommel et al. [5] and Goldfarb and Golper [14], and the use of human (trimeric) TNF in our study. The same was true for IL-6. The higher adsorptive capacity of the AN69 membrane was also demonstrated for 2 -microglobulin (11.8 kda) [15] and may be the result of the intrinsic AN69 membrane properties [16]. Our results are in contrast with results of hemofiltration studies in septic patients using the AN69 membrane. In these studies adsorption of TNF, IL-6 and IL-8 was absent and TNF clearance was either extremely low [5], or as high as 30 L/day [4]. Several factors can explain these conflicting results between in vitro and in vivo experiments. The most important difference is probably the relatively high membrane surface and ultrafiltrate flow compared to the cytokine distribution volume. Time of sampling is another important factor. Membrane adsorption of cytokines will not be detected if ultrafiltrate and blood samples are taken when the membrane is already saturated. Finally, membrane-induced cytokine production, endogenous cytokine production and endogenous cytokine clearance are confounding factors in septic patients. 128

130 Cytokine removal & CVVH In the present study local IL-8 production was suggested with the polysulfone membrane. This is a remarkable finding, as the test-solution only contained erythrocytes and freshfrozen plasma. However, as prestorage leucocyte filters were not used, we cannot exclude the possibility that the erythrocyte/plasma mixture was contaminated with a few cytokine-producing cells. Another possibilty is the release of erythrocyte bound IL- 8. This release of IL-8 from erythrocytes could be based on the erythrocyte membrane interaction which resulted in an improved convection of free IL-8. Conclusion In the present study, the highest reduction of TNF, IL-6 and IL-8 after 4 h of hemofiltration was found with the polyacrylonitrile membrane and was largely due to adsorption. Adsorption of TNF, IL-6 and IL-8 was also seen, but to a lesser degree in the polysulfone, cellulose triacetate and polyamide membranes. None of the membranes filtered TNF. Filtration of ll-6 and IL-8 was low (SC < 0.15) with all membranes with only marginal differences among the membranes or between the filtration mode. Care has to be taken to extrapolate these results to clinical practice as relative large membranes and high volume filtration volumes were used on only a small volume of cytokine solution. However, given the small amount of convective removal of all cytokines with all membranes, and the potential for adsorption to be saturated, our results suggest that there is little role for continuous hemofiltration in cytokine removal. 129

131 Chapter 8 References Ronco C. Continuous renal replacement therapies for the treatment of acute renal failure in intensive care patients. Clin Nephrol 1993;40: Grootendorst AF, van Bommel EFH, van Leengoed LAMG, et al. Infusion of ultrafiltrate from endotoxemic pigs depresses myocardial performance in normal pigs. J Crit Care 1993;8: Hoffmann JN, Faist E, Deppisch R, et al. Hemofiltration in human sepsis: Evidence for elimination of immunomodulatory substances. In Sieberth HG, Strumvoll HK, Kierdorf H eds.: Continuous extracorporeal treatment in multiple organ dysfuction syndrome. Contrib Nephrol Basel: Karger, 1995;116: Bellomo R, Tipping P, Boyce N. Tumor necrosis factor clearances during veno-venous hemofiltration with dialysis removes cytokines from the circulation of septic patients. Crit Care Med. 1993;21: Van Bommel EFH. Hesse CJ, Jutted NHPM, et al. Cytokine Kinetics (TNF-, IL-1ß, IL-6) during continuous hemofiltration: a laboratory and clinical study. In Sieberth HG, Strumvoll HK, Kierdorf H, eds.: Continuous extracorporeal treatment in multiple organ dysfunction syndrome. Contrib Nephrol Basel: Karger, 1995;116: Kierdorf H, Melzer H, Weiben D, et al. Elimination of tumor necrosis factor (TNF) by continuous venovenous hemofiltration (CVVH). Ren Fail 1992;14:98 (Abstr). Tonnesen E, Hansen MB, Hohndorf K, et al. Cytokines in plasma and ultrafiltrate during continuous arteriovenous haemofiltration. Anaesthesia and Intensive Care 1993;21: Sfyras D, Douzinas E, Perreas K et al. Effect of continuous veno-venous hemofiltration on sepsis. Blood Purif 1995;13: Sander A, Armbruster W, Sander B, et al. Hemofiltration increases Il-6 clearance in early systemic inflammatory response syndrome but does not alter Il-6 and TNF plasma concentrations. Intensive Care Med 1997;23: David S, Cambi V. Hemofiltration: Predilution versus postdilu tion. In: Shaldon S, Koch KM, eds.: Polyamide - The evolution of a synthetic membrane for renal therapy. Contrib Nephrol Basel: Karger, 1992;96 : Colton CK, Henderson LW, Ford CA, et al. Kinetics of hemodiafiltration. In vitro transport characteristics of a hollow-fiber blood ultrafilter. J Lab Clin.Med. 1975;85: Neote K, Darbonne W, Ogez J, et al. Identification of a promiscuous inflammatory peptide receptor on the receptor of red blood cells. J Biol Chem. 1993;268: Nagaki M, Hughes RD, Lau JYN, et al. Removal of endotoxin and cytokines by adsorbents and the effect of plasma protein binding. Int J Artif. Organs 1991;14:

132 Cytokine removal & CVVH Goldfarb S, Golper TA. Proinflammatory cytokines and hemofiltration membranes. J Am Soc Nephrol 1994;5: Jorstad S, Smeby LC, Balstad T, et al. Generation and removal of anaphylatoxinsduring hemofiltration with five different membranes. Blood Purif 1988;6: Pacual M, Schifferli A. Adsorption of complement factor D by polyacrylonitrile dialysis membranes. Kidney International 1993;43:

133

134 Chapter 9 Hemofiltration in sepsis and SIRS: The role of dosing and timing Catherine S.C. Bouman, Heleen M. Oudemans-van Straaten, Marcus J. Schultz, Margreeth B. Vroom Journal Crit Care (in press)

135 Chapter 9 Abstract Introduction The benefit of hemofiltration as an adjunctive treatment for sepsis or SIRS in critically ill patients is subject of severe debate. Firm conclusions on this subject are hampered by the heterogeneity in study populations and hemofiltration treatments and the lack of adequately sized RCTs. Aim of this review was to determine the importance of ultrafiltration dose and timing on the physiological and clinical effects of hemofiltration in sepsis and SIRS. In addition we discuss the issue of filter pore size. Methods Literature search was done in Embase and PubMed database for animal and human studies. Results Animal studies suggest beneficial effects of hemofiltration on hemodynamics; gas exchange; sepsis-induced immunoparalysis; histology of gut, lung and kidney; and (short term) survival. These effects were more prominent with very high ultrafiltrate rates ( 100 ml/kg/h) and early initiation of hemofiltration (ie, before or very early after the septic challenge). Three small randomized studies and 3 observational studies in patients with sepsis or SIRS show beneficial effects of short-term or pulse hemofiltration using very high ultrafiltrate rates and/or early initiation of hemofiltration on physiological endpoints and survival. However, the studies were underpowered for survival. The first observations of high permeability hemofiltration (pore size about 10 nm; in vitro cutoff, 100 kda) are promising, but so far, it has not been sufficiently examined to allow strong conclusions. Conclusion Human and animal studies suggest that early initiation and high ultrafiltrate volumes are determinants of the beneficial physiological and clinical effect of hemofiltration in sepsis and SIRS. As yet, the evidence in humans is to scarce to recommend hemofiltration as adjunctive therapy for critically ill patients with sepsis or SIRS. Regarding the many uncertainties about optimal volume (high or very high) and type of membrane, clinical studies should first focus on endpoints such as recovery from organ failure and length of treatment, before survival studies are started. 134

136 Hemofiltration & sepsis Introduction Continuous hemofiltration permits efficient control of fluid balance and azotemia in septic patients with acute renal failure (ARF) [1]. In 1984 Gotloib et al. suggested additional beneficial effects of hemofiltration, resulting from the removal of inflammatory mediators, such as proinflammatory cytokines [2]. Indeed, hemofiltration has been shown to remove several proinflammatory mediators, although generally not resulting in decreased plasma levels [3]. Nowadays, the septic syndrome is no longer seen as a disorder of uncontrolled inflammation, but more as a syndrome characterized by loss of the balance between pro- and antiinflammatory forces [4,5]. Ronco et al. proposed the peak concentration hypothesis as the mechanism by which hemofiltration in sepsis could be beneficial [6]. This concept refers to the ability of hemofiltration to lower peaks of both the pro- and antiinflammatory mediators, reducing their toxic effects such as capillary leak, vasoplegia and immunodepression. By this nonspecific removal of an overshoot of mediators, a new equilibrium might be shaped [7]. Evaluation of the efficacy of hemofiltration in sepsis is complicated by the heterogeneity of both study populations and hemofiltration strategies. In addition to the role of hemofiltration, the outcome of sepsis is determined by many factors, including the virulence of the microorganism, polymorphisms in cytokine genes of the patient, co-morbidity and concomitant treatment [4,5]. Experimental septic shock models overcome the problem of heterogeneity, although one should be careful to extrapolate the results of experimental shock to the complex clinical situation. The heterogeneity in hemofiltration operational characteristics further complicates the evaluation of the role of hemofiltration in sepsis [8]. The aim of the present review was to evaluate the role of ultrafiltrate dose, timing of hemofiltration, and membrane characteristics in sepsis or SIRS. We focused on physiological and clinical endpoints. For the effect of hemofiltration on the removal of inflammatory mediators, the reader is referred the excellent review of de Vriese et al. [9]. We decided to review both animal and human studies, because the concept has developed by an interaction of knowledge derived from both types of studies. Methods Studies were identified using the MeSH terms hemofiltration OR haemofiltration combined with the words sepsis OR septic shock OR SIRS in Embase and PubMed from 1984 until 135

137 Chapter 9 July 2005, restricting the language to English, French or German and by scanning the reference lists of publications found by database searches. The identified studies were eligible if they fulfilled all of the three following criteria: (a) Prospective randomized trials or observational studies including at least ten animals or critically ill patients with sepsis or SIRS secondary to pancreatitis, cardiac arrest, bowel ischemia or endotoxin infusion; (b) hemofiltration treatment with specified treatment characteristics, that included at least ultrafiltration rate and membrane type; (c) study endpoints included physiological or clinical effects as proposed by the ADQI consensus statement [8]. Physiologic effects include improved hemodynamic profile or decreased need for vasopressor drugs, improved vital signs, improved cardiopulmonary function, or improved markers of renal function. Clinical endpoints include organ dysfunctionfree days, ventilator-free days, intensive care unit-free days, duration of intensive care unit/hospital stay, dialysis-independent survival, cost-effectiveness analysis, survival, survival to hospital discharge, relation between improved hemodynamic responses, survival and immunological changes. We excluded ex vivo studies and in vitro studies, studies without a physiological or clinical endpoint and studies evaluating the effect of hemodiafiltration, hemodialysis or hemoadsorption and plasmapheresis. We also excluded studies applying hemofiltration during cardiac surgery and in patients with cardiac failure because these studies specifically focus on the beneficial effects of fluid removal. Ultrafiltrate rate was expressed in milliliters per kilogram per hour and was estimated if not stated to facilitate comparison of studies. Bodyweight for humans was estimated to be 75 kg in non-asian countries per 55 kg in Asian countries; in animal studies average bodyweights were used. The studies were classified into dosing studies, timing studies or permeability studies. Dosing studies were classified into low-volume hemofiltration (LV-HF, < 30 ml/kg/h) based on the conventional renal dose of 2 L/h [10], high-volume hemofiltration (HV-HF, ml/kg/h) and very high-volume hemofiltration (VHV- HF, > 50 ml/kg/h). Permeability studies were classified into studies using conventional membranes (filter pore size of about 5nm or in vitro permeability of kda) and high cutoff hemofiltration (filter pore size about 10 nm or in vitro permeability of kda). After assessment of the eligible studies we concluded that the studies were too heterogeneous to be sensibly combined in a metaanalysis. Therefore, we performed a descriptive analysis. The characteristics of the animal and human studies are summarized in the tables. All studies used (low-molecular-weight) heparin for anticoagulation of 136

138 Hemofiltration & sepsis the extracorporeal circuit; most used conventional membranes allowing the removal of molecules up to 30 kda, with the exception of the high cutoff hemofiltration studies allowing the removal of molecules up to 50 kda. The animal studies applied zerobalanced hemofiltration. Factors involved in the effects of hemofiltration in sepsis It has been suggested that the beneficial effects of hemofiltration in sepsis are related to high ultrafiltration rates, early start of hemofiltration, use of highly permeable filters, and adsorption [11-14]. In contrast to hemodialysis, removing substances by diffusion, clearance of solutes with hemofiltration is achieved by convection (ultrafiltration) and adsorption [1]. With the use of conventional membranes convection is associated with higher removal of middle- and high-molecular weight substances, substances that play important roles in the pathogenesis of sepsis [15,16]. Both membrane and solute characteristics (geometry, charge, molecular weight, and protein binding) determine the degree of removal by ultrafiltration and adsorption [17,18]. Important membrane-related determinants are pore characteristics (size, distribution and density), ph, charge and surface [18]. Convective removal is proportional to ultrafiltration rate and SC. For solutes freely crossing the membrane, SC values are equal or close to one. Conventional membranes usually have a pore size of about 5 nm (in vitro cutoff kda). This allows the removal of molecules up to a molecular weight of about 30 kda. Although many mediators of sepsis have a molecular weight between 15 and 50 kda, they are not freely filtered because of, for example, protein binding or trimer formation and the gradual formation of a protein layer on the membrane decreasing effective in vivo pore size [19]. High cutoff membranes have pore sizes of about 10 nm (in vivo cutoff kda), allowing the elimination of molecules with molecular weights up to 50 kda [20]. Filter permeability has dramatic influence on the removal of mediators [12,21-25]. For some membranes, in particular, the negatively charged polyacrilonitrile (AN69) membrane, adsorption is the main mechanism of mediator removal [3,26]. However, adsorption to the membrane is subject to saturation of binding sites over time and secondary release from the membrane [17,27]. The blood-membrane interaction during hemofiltration can also induce the release of inflammatory mediators although this process is much less with the use of biocompatible membranes nowadays used [28]. Therefore, the net effect of hemofiltration on the circulating concentration of a mediator will depend on its volume of distribution, its SC, the ultrafiltration rate, its adsorption to the membrane and its 137

139 Chapter 9 generation rate at tissue and membrane level. In the following, studies are discussed according to hemofiltration dose and timing of initiation of hemofiltration. In addition we evaluated the effects of membrane characteristics (adsorption and pore size). Studies using a highly adsorptive membrane are discussed in the section on hemofiltration dose. Studies evaluating the effect of high cutoff membranes are discussed separately. Hemofiltration dose Low-volume hemofiltration (LV-HF) studies (< 30 ml/kg/h) Animal studies Animal studies in dogs [29,30] and pigs [31-33] with endotoxin-induced or abdominal sepsis using LV-HF show marginal beneficial physiological effects on myocardial contractility [29,30], lung compliance [33] and phagocytosis [31]. Gomez et al. ascribed the improved myocardial contractility to removal of a cardiodepressant factor, because in vitro exposure of the trabeculae of healthy dogs to plasma obtained from septic dogs reduced isometric tension while post-hemofiltration, the depressant effect of plasma, was less [29]. Removal of toxic mediators was also suggested by Stein et al., because lung compliance improved without any significant effects on extravascular lung water [33]. In the study by Stein et al., the improved lung compliance did not result in improved oxygenation, nor did hemodynamics improve significantly [32,33]. The LV-HF studies in animals only evaluated physiological endpoints and not clinical endpoints (Table 1). Human studies The number of controlled studies is limited and their size is small [34-37]. Two studies included patients with severe sepsis and ARF [35,36]. Low-volume hemofiltration significantly improved survival compared to peritoneal dialysis [36]. Although the authors did not report on the renal replacement dose in the peritoneal dialysis group, we assumed that this dose was lower than the LV-HF group as metabolic control was also less. In another study, LV-HF slightly improved hemodynamics; however, the parameters of splanchnic regional perfusion did not improve [35]. Two other small randomized trials compared the effects of LV-HF to no hemofiltration in patients with septic shock without ARF [34,37]. In these 2 studies LV-HF did not improve physiological [34,37] or clinical endpoints [34], and plasma levels of TNF, interleukin IL-6 and anaphylatoxins did not decrease. Unfortunately, the study by Cole et al. was not powered to detect a tangible effect on organ dysfunction (Table 2 and 3). 138

140 Hemofiltration & sepsis Numerous uncontrolled clinical studies reported beneficial effect of LV-HF on hemodynamics [38-44] or gas exchange [45-48], but as no control groups were included, it is not clear if these effects were hemofiltration related. High-volume (HV-HF) hemofiltration (30-50 ml/kg/h) Animal studies In rats with endotoxin-induced shock, short term (4 h) survival improved (5/6 vs 0/6, p < 0.01) following HV-HF, and this was associated with the removal of thromboxane A2 [49]. However, the study did not include a nonfiltering control group, but compared hemofiltration to sham hemofiltration (hemofiltration with clamped ultrafiltrate line), and thus, it cannot be excluded whether sham hemofiltration caused negative effects because clamping of the ultrafiltrate allows the release of mediators due to the blood membrane interaction while the convective removal of mediators is inhibited. In the porcine endotoxin shock model of Murphey et al. a nonfiltering control group was included and compared to HV-HF and sham hemofiltration but the differences in hemodynamic parameters among the three groups were not statistically significant [50] (Table 1). Human studies The largest RCT (n = 425) in critically ill patients on the effects of ultrafiltrate dose on outcome was performed in patients with oliguric ARF with or without sepsis [51]. Patients were randomly assigned to groups 1 (20 ml/kg/h), 2 (35 ml/kg/h) and group 3 (45 ml/kg/h). The survival 15 days after discontinuation of hemofiltration was significantly lower in group 1 (41%) compared to group 2 (57%) and 3 (58%), indicating a minimal renal dose of 35 ml/kg/h in patients with ARF to improve survival. In the subgroup of patients with sepsis (n = 52), survival was 47% in group 3 compared to 18% in group 2 and 25% in group 1 (p = 0.23) and the authors postulated that sepsis patients might benefit from a higher ultrafiltrate dose (Table 2). In a small observational study increasing the ultrafiltrate volume resulted in increased removal of several cytokines, but was not associated with hemodynamic effects [3]. Filter replacement also resulted in increased cytokine removal but this effect was only temporary [3] (Table 3). 139

141 Chapter 9 Table 1. Characteristics animal studies. Study [ref] Animal model Randomization (number of animals) Sepsis HF delay [min] Fluid resuscitation Filter [m 2 ] UF rate [ml/kg/h] Duration HF [min] Physiological effect Clinical effect Gomez [29] Dog EColi sepsis HF (9); S+HF (8); Hydralazine+HF (5) 240 target pressure colloid PS nr Mink [30] Dog EColi sepsis S (10); S+HF (10); shamhf (7) S+Phenylefrine (10); S+HF+Phenylefrine (10); 240 target pressure Colloid PA 0, nr Stein [32;33] Pig LPS 18,5 ug/kg S (10); S+HF (10) 90 7 ml/kg/h crystalloid PS ± nr Discipio [31] Pig Peritonitis S (8); S+HF (8) 0 12 ml/kg/h colloid PS nr Heidemann [49] Rat LPS 10mg/kg S+HF (6); S+shamHF (6) 15 not reported PS Murphey [50] Pig LPS 15 ug/kg S (5); S+HF (7); S+shamHF (4) 30 target Ht crystalloid PS nr Grootendorst [11] Pig LPS 0,5 mg/kg S (6); S+HF (6); S+shamHF (6) ml/h crystalloid PS nr Grootendorst [57] Pig SMA clamping S (6); S+shamHF (6) Before clamping 600 ml/h crystalloid CP Ullrich [53] Pig LPS 0,5 mg/kg S (6); S+HF (6); S+shamHF (4) 60 target pressure crystalloid PS nr Rogiers [55] Pig LPS 2 mg/kg S (5); S+HF 3L/h (5); S+HF 6L/h (5); HF (5) 30 target pressure crystalloid PS 0,7 100 and nr Wang [58] Pig Pancreatitis S (8); S+LVHF (8); S+VHVHF(8) 0 5 ml/kg/h crystalloid PAN 20 and nr Yekebas [59] Pig Pancreatitis S(12); S+LVHF (12); S+LVHF+FFR (12); S+VHVHF+FFR (12); 850 nr PAN 20 and Rogiers [13] Dog LPS 2mg/kg S (5); S+PS (5); S+PAN (5) 30 target pressure crystalloid PS 0,71 PAN nr 140

142 Hemofiltration & sepsis Bellomo [54] LPS 0,5 mg/kg S+HF (8); S (8) Before sepsis target pressure crystalloid PAN 0, nr Lee [52] Freeman [60] Lee [12] Veenman [73] Pig SAur sepsis Dog Peritonitis Pig SAur sepsis Pony LPS 2 ug/kg S+HF (20); HF (4); 60 no PS S+shamHF (20) S(7); S+HF (7); S+shamHF 60 nr PS (7) S+HF (7); S+HPHF (7) 60 no PS S (5); S+HF (5) 30 5 ml/kg/h crystalloid PMMA ,6 Mink [71] Dog Paer sepsis S+earlyHF (12); S (13); shams (5); shams+earlyhf (4); S+lateHF (13); S (13); shams (4); shams+latehf (4) 120 and ml/kg/h colloid PA nr 0,66 Yekebas [72] Yekebas [14] Pig Pancreatitis Pig Pancreatitis S (11); S+earlyHF (12); S+lateHF (12) S+lateLVHF (12); S+early- LVHF (12); S+lateLVHF+FFR (12); S (12); S+earlyVHVHF (12); S+lateVHVHF+FFR (12); S+earlyVHVHF+FFR (12) 0 and and ml/kg/h PS crystalloid nr PAN 20 and UF, ultrafiltration; Ecoli, Escherichia coli; LPS, Escherichia coli endotoxin; SMA superior mesenteric artery; Saur, Staphylococcus Aureus; Paer, Pseudomonas Aeruginosa; HF, hemofiltration; S, sepsis ; shamhf, HF with clamped ultrafiltrate line; LVHF, low-volume HF; VHVHF, very high-volume HF; FFR, frequent (every 12h) filter replacement; min, minutes; PS, polysulfone; PA, polyamide; CP, cuprophane; PAN, polyacrilonitrile; PMMA, polymethylmethacrylate; nr, not reported; see text for explanation early and late hemofiltration 141

143 Chapter 9 Table 2. Characteristics randomized human studies. Study [ref] Randomization (number of patients) Duration HF (hours) ARF UF rate (ml/kg/h) Filter (m 2 ) In vitro cutoff Blood flow (ml/min) Substitution fluid Physiol effect Clinical effect Sander [37] HF 1L/h (13); no HF (13) 48 no 13 PAN ringer s solution nr Cole [34] HF 2L/h (12); no HF (12) 48 no 27 PAN (1,2) lactate Phu [36] HF 25 L/day (34); PD (36) Duration ARF or death yes 14 PA (0,66) lactate or bicarbonate nr + John [35] HF 2 L/h (20); IHD (10) 96 yes 26 PS (1,35) lactate + nr Cole [61] HF 1L/h (11); HF 6L/h (11) 8 yes 13 and 80 PAN (1,2); PAN (1,6) ; 300 lactate + nr Ronco [51] HF 20 ml/kg/h (20); HF 35 ml/kg/h (17); HF 45 ml/kg/h (15) Duration ARF or death yes 20, 35 and 45 PS (1,3 or 1,7) lactate nr + Jiang [62] Early HF 1L/h (9); Late HF 1L/h (10); Early HF 4L/h (9); Late HF 4L/h (9) 72 h no 18 and 72 PAN (1,2) nr + + Laurent [63] Control (19); HF 200 ml/kg/h (22); HF+ HT (20); 8 no 200 PA (2,1) bicarbonate + + N, number of patients; HF, hemofiltration; PD, peritoneal dialysis; IHD, intermittent hemodialysis; HT, hypothermia; ARF, acute renal failure; UF, ultrafiltrate; PAN, polyacrilonitrile; PA, polyamide ; PS, polysulfone; early HF, within 48 h of the onset of abdominal pain; late HF, after 96 h of the onset of abdominal pain; nr, not reported. 142

144 Hemofiltration & sepsis Table 3. Characteristics observational human studies. Study [ref] Observation (number) Duration HF [hours] ARF UF rate (ml/kg/h) Filter (m 2 ) In vitro cutoff Blood flow [ml/min] Substitution fluid Physiol effect Clinical effect Gotloib [47] CAVH (24) nr + (6) nr CP (1) ringer s + nr Ossenkoppel [44] CAVH (10) PS (0,2) 30 nr no + nr Matamis [48] CAVH (20) PS (0,7) lactate + nr Meloni [43] CAVH (18) PAN (0,4) 40 nr bicarbonate + nr Heering [38] HF 1 L/h (18) PS (1,35) lactate nr Blinzler [45] HF 8 42 L/day (11) PA (0,66) lactate + nr Klouche [40] HF ml/h (11) PAN (0,8) bicarbonate + nr Kruger [41] HF L/day (19) 154 ± PA (0,88) nr + nr Level [42] HF 2500 ml/h (13) PAN (0,9) PA (1,4) bicarbonate + nr Hoffmann [39] HF 2 L/h (16) 12 + (3) 27 PA ((0,66) isotonic fluid + nr De Vriese [3] HF+Q b 100 and 200 ml/min (16) and 33 PAN (0,9) 20/ Isotonic fluid nr Joannes [66] HF ml/kg/h (24) < 96 h [6] > 96 h [18] PS (2) bicarbonate + + Oudemans [67] HF 100 L/day (91) CTA (1,9) lactate or bicarbonate + + Wang [69] HF 4 L/h (28) (7) 72 PAN (1,2) nr + + Ratanarat [68] HF 85 ml/kg/h (15) PS 35 1,8 2,0 bicarbonate + + Honore [65] HF 35 L/4h (20) 4 nr 116 PS (1,6) bicarbonate + + Morgera [74] HP HF 1 L/h (16) PA (0,6) 30/ nr + CAVH, continuous arteriovenous hemofiltration; N, number of patients; HF, hemofiltration; Q b, blood flow; HP-HF, high permeability hemofiltratio; nr, not reported; ARF, acute renal failure; UF, ultrafiltrate; CP, cuprophane; PS, polysulfone; PAN, polyacrilonitrile; PA, polyamide; CTA, cellulose triacetate; nr, not reported. 143

145 Chapter 9 Very high-volume (VHV-HF) hemofiltration (> 50 ml/kg/h) Animal studies Animal studies in pigs [11,52,53] and dogs [54,55] with endotoxin-induced shock using VHV-HF showed beneficial effects on physiological endpoints [11,52-55] and survival time [52]. In contrast to ultrafiltrate from healthy donor animals, infusion of ultrafiltrate from septic donor animals into healthy acceptor animals decreased cardiac output and MAP [55,56] and death [52], indicating that septic ultrafiltrate contains toxic mediators. In the experiments of Rogiers et al. the observed beneficial effects were not due to TNF removal, because hemofiltration did not influence plasma TNF levels and ultrafiltrate TNF levels were very low throughout the experiment [55]. Interestingly, in the study of Grootendorst et al. cardiac performance was lower in the sham hemofiltration group (clamped ultrafiltrate line) than in the control group, suggesting negative effects of the extracorporeal system itself [11]. The effects of VHV-HF were further evaluated in two other sepsis models in pigs: clamping of the superior mesenteric artery [57] and bile induced pancreatitis [58,59]. In the gut ischemia and reperfusion model, VHV-HF did not only improve hemodynamics, but also reduced macroscopic gut damage and improved 24-h survival [57]. In the pancreatitis models, VHF-HF improved hemodynamics [58,59] reversed sepsis induced immunoparalysis resulting in considerable sepsis protection and 60-h survival [59]. In addition, the pancreatitis studies showed that VHV-HF was more efficient than LV-HF. [59] (Table 1). Of interest frequent filter replacement (every 12 h versus every 60 h) also improved the efficacy of hemofiltration, but to a lesser extend than increasing the ultrafiltration rate, suggesting that convection was more important than adsorption [59]. Rogiers et al. studied the effect of membrane characteristics during VHV-HF in dogs with endotoxin-induced shock [13]. They compared the polysulfone filter to the more adsorptive polyacrilonitrile filter [17,26]. Compared with no HF, pulmonary vasoconstriction decreased in both hemofiltration groups; however, myocardial contractility increased only in the polyacrilonitrile group. The more beneficial effects of the polyacrilonitrile group were only temporary and after 5 h, the effects in both hemofiltration groups were comparable, suggesting saturation of the polyacrilonitrile membrane. Adsorption and subsequent saturation to the polyacrilonitrile membrane was also suggested in the study of Bellomo et al. in dogs with endotoxin-induced shock [54]. The effect on blood pressure was early and only temporary, and the convective removal of the measured mediators (endothelin 1, TNF, prostaglandins, endotoxin) was low. 144

146 Hemofiltration & sepsis In contrast to clinical practice, animals in the experimental sepsis studies did not receive antibiotics, except in one study [60]. In that study, in dogs with bacterial peritonitis treated with antibiotics, VHV-HF had no significant effect on hemodynamics or 7-day survival. It should be noted that in this study, ultrafiltration rates were somewhat lower (60 ml/kg/h) than in the earlier very aggressive (>100 ml/kg/h) hemofiltration studies [11,13,55,57-59] (Table 1). Human studies Only one larger and two small RCTs studied the effects of VHV-HF [61-63]. The largest randomized study applied the highest ultrafiltration rate (200 ml/kg/h) for a duration of 8 hours, comparable to the Grootendorst experiments, and was performed in 61 patients following out of hospital cardiac arrest [63]. Patients, after cardiac arrest, exhibit a sepsislike syndrome combined with increased endotoxin levels [64]. Cardiac arrest patients were randomized to either control, VHV-HF or VHV-HF combined with hypothermia. Hemofiltration (with or without hypothermia) was associated with improved 6-month survival and a decreased risk of death from early intractable shock. No significant effect of hemofiltration on IL-6 or anaphylatoxins was found. In a small crossover study in 11 patients with septic shock and ARF, Cole et al. measured the hemodynamics, serum cytokines and complement concentration [61]. Patients were randomly assigned to VHV- HF (80 ml/kg/h for 8 h) or to LV-HF (13 ml/kg/h for 8 h). Less norepinephrine was required to maintain target MAP (> 70 mm Hg) during VHV-HF than during LV-HF (proportional decrease 68% versus 7% over 8 hours, p = 0.02), and this was not caused by an effect on the fluid balance. Other hemodynamic parameters did not change significantly over time during either therapy. Very high-volume hemofiltration caused a greater reduction in the level of anaphylatoxins than LV-HF. The concentration of mediators in the ultrafiltrate was negligible suggesting adsorption as the major mechanism for mediator removal with this polyacrilonitrile filter. According to the authors adsorption was higher during VHV- HF because of higher transmembrane pressures and because a larger filter was used. In another very small RCT in 37 patients with severe pancreatitis, hemodynamics and short-time survival rate were significantly better during VHV-HF (70 ml/kg/h) than in patients receiving LV-HF (18 ml/kg/h) [62] (Table 2). Several observational studies showed beneficial effects of VHV-HF on survival [65-68], hemodynamics [65,66,68,69] and oxygenation [69] (Table 3). This studies, however, because these studies did not include control groups and therefore, it is not clear if the observed effects are hemofiltration related. In a large observational study in septic patients (n = 91) receiving at least 24 h of VHV-HF (63 ml/kg/h for at least 24 h) mortality was lower (33%) than predicted by the APACHE II (76%) and SAPS II (71%) 145

147 Chapter 9 illness severity scores [67]. In a smaller (n = 24) observational study in septic patients, VHV-HF (40-60 ml/kg/h for 96 h) improved hemodynamics and decreased the need for catecholamine support [66]. The latter study also reported reduced observed 28- day mortality compared with what was predicted; however, the scores used to predict mortality were not developed to predict 28-day mortality. Concerns have been expressed about the feasibility and costs of continuous VHV-HF resulting in the introduction of short-term and pulse VHV-HF [68,70]. Encouraging results of short-term VHV-HF (116 ml/kg/h for 4 h) were reported in a small prospective uncontrolled study in patients with refractory septic shock [70]. Predetermined hemodynamic goals were achieved in 11 of the 20 patients. Of these 11 responders, 9 patients were still alive on day 28. All nonresponders died within the first 24 hours. The observed mortality (55%) was significantly lower than the APACHE II and SAPS II predicted mortality (79%). Retrospective analysis showed that in the responders, ultrafiltrate volume was higher and hemofiltration was started earlier than in the nonresponders, suggesting that a high ultrafiltrate dose and an early start of hemofiltration could be important factors. In another observational study in patients with sepsis, pulse VHV-HF (85 ml/kg/h for 6-8 h followed by 16 to 18 h of 35 ml/kg/h) improved hemodynamics, allowing a reduction in norepinephrine dose [68]. Moreover, the observed 28-day mortality was 47%, whereas the predicted mortality rates were 72% (based on APACHE II) and 68% (based on SAPS II). As mentioned by the authors the use of activated protein C in approximately 50% of the patients might have contributed to the improved outcome. Conclusion Animal studies report minor improvement of cardiac function and oxygenation during LV-HF but not during HV-HF. The survival benefit reported during HV-HF is hampered by the fact that hemofiltration is only compared to sham hemofiltration. In contrast, most of the VHV-HF studies report both beneficial physiological (in particular cardiovascular function) combined with clinical effects (short-term survival, organ damage and sepsisinduced immunoparalysis). The only negative VHV-HF study applied less aggressive ultrafiltration rates (60 ml/kg/h versus > 100 ml/kg/h), and all animals received antimicrobial treatment. In patients with sepsis and ARF, large randomized studies report that LV-HF is superior to peritoneal dialysis and HV-HF is superior to LV-HF, suggesting that a minimal renal replacement dose is necessary to improve survival (35 ml/kg/h). Patients with sepsis without ARF do not benefit from LV-HF. The VHV-HF studies report both physiological (hemodynamics, oxygenation) and clinical (survival) effects in patients with or without ARF; however, the studies are underpowered for survival. 146

148 Hemofiltration & sepsis Most of the observational LV-HF and HV-HF studies report beneficial effects on hemodynamics and/or gas exchange, but do not evaluate clinical endpoints. In contrast, observational VHV-HF report both improved survival and hemodynamics. Initiation time of hemofiltration In most of the experimental studies, hemofiltration started before or shortly after the microbial challenge, whereas in clinical practice, hemofiltration is generally started when shock and organ failure are already established. Animal studies Animal studies in pigs with pneumosepsis [71] or pancreatitis [14,72] showed that the beneficial effects of hemofiltration on physiological [14,71,72] and clinical endpoints [14,72] were more pronounced when hemofiltration was started early. In the animal studies early hemofiltration started simultaneously [14,72] or 2 h [71] after the insult, but before MAP fell, whereas late hemofiltration started when MAP started to decrease. Human studies The previously mentioned RCT of Jiang et al. in patients with pancreatitis found that in addition to a higher dose, survival was also significantly better in patients receiving early hemofiltration (within 48 h after onset of abdominal pain) than in the group with late hemofiltration (96 h after onset abdominal pain); however, the study was underpowered for survival [62]. In the cohort study by Honore and Matson in patients with severe septic shock, post hoc analysis showed an association between a better survival and an earlier start of hemofiltration [70]. Conclusion Early hemofiltration is more beneficial than late hemofiltration in animal studies, but it is important to realize that early hemofiltration as defined in animal studies is hardly feasible in clinical practice. In humans two studies show beneficial effects on survival when hemofiltration is started early, but these studies are either underpowered or retrospective. 147

149 Chapter 9 Pore size of the membrane In contrast to the use of conventional membranes, a substantial removal of middlemolecular-weight solutes can be achieved by increasing pore size of the membrane [21,22,25]. Animal studies VHV-HF and a high cutoff membrane significantly increased survival time (103 versus 56 h) in pigs with Staphylococcus Aureus sepsis compared with VHV-HF and a conventional membrane [12]. Ultrafiltrate analysis revealed an 8-fold higher protein concentration in the high permeability group compared to the conventional group and did not contain a high percentage of albumin (69 kda) suggesting that convective removal of mediators was improved with the high cutoff filter. Beneficial effects were not reported in another study in ponies with endotoxin-induced shock receiving VH-VHF (120 ml/kg/h) and a high cutoff membrane or no HF [73]. However in this small study, the endotoxin dose was probably too low to show a beneficial effect of hemofiltration (Nine of the 10 ponies recovered well). Human studies In a small observational pilot study on safety, intermittent high cutoff hemofiltration (13 ml/kg/h for 12 h, using a polysulfone membrane with an in vitro cutoff of 100 kda) led to a slight improvement of organ failure in 16 patients with septic shock and ARF combined with high IL-6 removal [74]. Despite a cumulative protein loss of almost 8 g/d, treatment appeared to be well tolerated hemodynamically and the coagulation system was not affected. The same author performed a small RCT in 24 patients with sepsis comparing the same high cutoff filter in the dialysis mode and in the HF mode [23]. Convection was more effective in the clearance of IL-1RA than diffusion, but was associated with greater protein loss especially when ultrafiltration volume was increased from 1 L/h to 2.5 L/h. No differences in clinical course and outcome were seen among the groups, but the number of patients per group (n = 6) was very small. In the human studies, the renal replacement dose in the 1-L/h group was lower than the suggested minimal renal dose of 35 ml/kg/h [51] and much lower than in the VHV-HF studies. The combination of high cutoff hemofiltration and very high ultrafiltration rates might be an option to increase cytokine removal further. The main drawback of the method is the loss of proteins other than cytokines, such as albumin. However, in the in vitro study of Uchino et al. using a large pore size filter (100kDa) clearances of cytokines were significantly improved by an increase of ultrafiltration rate from 1 to 6 L/h, whereas albumin clearance was not increased [75]. 148

150 Hemofiltration & sepsis Conclusion Although the number of high cutoff membrane studies is small, they show an effective removal of inflammatory mediators and suggest that the method is tolerated well. The issue of pore size and ultrafiltration rate needs further evaluation. General discussion To determine the role of ultrafiltration dose, timing of hemofiltration, and membrane characteristics (pore size and adsorption) on physiologic and clinical endpoints of hemofiltration in sepsis and SIRS, we summarized the available data in animals and humans. Ultrafiltration dose appears to be an important item. In animals, the negative studies are among those applying low ultrafiltration rates, and the most prominent effects are seen when very high ultrafiltrate rates are applied ( 100 ml/kg/h), showing beneficial effects not only on hemodynamics and gas exchange, but also on sepsis inducedimmunoparalysis, histological damage and short-term survival. However, animal studies poorly mimic the clinical situation as hemofiltration is started before or very early after the insult and sepsis is not treated with antimicrobial agents. The only negative VHV-HF study, is done in septic pigs also receiving antibiotic treatment, although it can be argued that the ultrafiltration rate in that study was lower (60 ml/kg/h) than in the other VHV-HF studies. Unfortunately, adequately sized RCT trials are not available. Evidence for a beneficial clinical effect of VHV-HF in humans is suggested by two RCT using a dose of 200 ml/kg/h in post cardiac-arrest patients and 72 ml/kg/h in patients with pancreatitis, both underpowered for survival as an eindpoint and from uncontrolled observational clinical studies using volumes ranging from 50 to 115 ml/kg/ h. There is reasonable evidence that hemofiltration with lower volumes has no benefit in patients with sepsis without ARF and that a dose of 35 to 45 ml/kg/h improves survival in critically ill patients with ARF as compared to 20 ml/kg/h. Although VHV-HF is associated with an increased loss of beneficial water-soluble substances (amino acids, vitamins, micronutrients), limitations of the venous access, and a higher tendency to filter clotting, human studies have not reported deleterious clinical effects. It should be noted that the highest ultrafiltration rates (> 85 ml/kg/h) were never applied for more than 8 hours. Timing is probably another important issue; however, the number of studies is smaller than the number of dose studies, and adequately sized RCT do not exist. In animals, 149

151 Chapter 9 studies on hemofiltration dose, started before or shortly after the insult, and this suggest that early initiation of hemofiltration is important. Indeed, in animal studies, late hemofiltration is less effective compared with early hemofiltration. Unfortunately, early hemofiltration as defined in animals is not feasible in clinical practice. One small RCT in patients with pancreatitis report improved survival when hemofiltration is started early (within 48 hours of the onset of abdominal pain), but the study is underpowered. The only other human study is observational and shows an association between survival and early start of VHV-HF in refractory septic shock. The first observations of high cutoff hemofiltration show a high removal potency of inflammatory mediators; however, studies to date are scarce and not powered for clinical endpoints. Human studies only applied high cutoff hemofiltration in combination with low ultrafiltration rates and for a short period. Treatment was tolerated well. The main drawback is the loss of other proteins. However, albumin loss declines when hemofiltration dose is increased. Although there is a physiologic rationale to use highly adsorptive membranes, early membrane saturation limits its practical use. Conclusion At this moment, clinical evidence is too low to recommend the application of hemofiltration as adjunctive therapy for sepsis without ARF. However, the findings so far justify suitably powered RCTs on the effects of early-short term or pulse hemofiltration using high ultrafiltrate volumes. The same applies to high cutoff hemofiltration as soon as more data on clinical endpoints and safety are available. Regarding the many uncertainties about optimal volume (high or very high), and type of membrane, clinical studies should first focus on endpoints as recovery from organ failure and length of treatment, before survival studies are started. A suitably powered RCT in humans should further be preceded by a critical analysis of why this has not yet been done. High costs, technical uncertainty, wide clinical variability, absence of a standard of concomitant treatment may have hampered the initiation of such studies. 150

152 Hemofiltration & sepsis References Bellomo R, Ronco C. Continuous renal replacement therapy: continuous blood purification in the intensive care unit. Ann Acad Med Singapore 1998;27: Gotloib L, Barzilay E, Shustak A, et al. Sequential hemofiltration in nonoliguric high capillary permeability pulmonary edema of severe sepsis: preliminary report. Crit Care Med 1984;12: De Vriese AS, Colardyn FA, Philippe JJ, et al. Cytokine removal during continuous hemofiltration in septic patients. J Am Soc Nephrol 1999;10: Bone RC. Important new findings in sepsis. JAMA 1997;278:249. Hotchkiss RS, Karl IE. The pathophysiology and treatment of sepsis. N Engl J Med 2003;348: Ronco C, Ricci Z, Bellomo R. Importance of increased ultrafiltration volume and impact on mortality: sepsis and cytokine story and the role of continuous veno-venous haemofiltration. Curr Opin Nephrol Hypertens 2001;10: Reiter K, Bellomo R, Ronco C, et al. Pro/con clinical debate: is high-volume hemofiltration beneficial in the treatment of septic shock? Crit Care 2002;6: Bellomo R, Honore PM, Matson J, et al. Extracorporeal blood treatment (EBT) methods in SIRS/Sepsis. Int J Artif Organs 2005;28: De Vriese AS, Vanholder RC, Pascual M, et al. Can inflammatory cytokines be removed efficiently by continuous renal replacement therapies? Intensive Care Med 1999;25: Venkataraman R, Kellum JA, Palevsky P. Dosing patterns for continuous renal replacement therapy at a large academic medical center in the United States. J Crit Care 2002;17: Grootendorst AF, van Bommel EF, van der Hoven B, et al. High volume hemofiltration improves right ventricular function in endotoxin-induced shock in the pig. Intensive Care Med 1992;18: Lee PA, Weger GW, Pryor RW, et al. Effects of filter pore size on efficacy of continuous arteriovenous hemofiltration therapy for Staphylococcus aureus-induced septicemia in immature swine. Crit Care Med 1998;26: Rogiers P, Zhang H, Pauwels D, et al. Comparison of polyacrylonitrile (AN69) and polysulphone membrane during hemofiltration in canine endotoxic shock. Crit Care Med 2003;31: Yekebas EF, Strate T, Zolmajd S, et al. Impact of different modalities of continuous venovenous hemofiltration on sepsis-induced alterations in experimental pancreatitis. Kidney Int 2002;62:

153 Chapter Brunet S, Leblanc M, Geadah D, et al. Diffusive and convective solute clearances during continuous renal replacement therapy at various dialysate and ultrafiltration flow rates. Am J Kidney Dis 1999;34: Kellum JA, Johnson JP, Kramer D, et al. Diffusive vs. convective therapy: effects on mediators of inflammation in patient with severe systemic inflammatory response syndrome. Crit Care Med 1998;26: Bouman CS, van Olden RW, Stoutenbeek CP. Cytokine filtration and adsorption during pre- and postdilution hemofiltration in four different membranes. Blood Purif 1998;16: Clark WR, Hamburger RJ, Lysaght MJ. Effect of membrane composition and structure on solute removal and biocompatibility in hemodialysis. Kidney Int 1999;56: Clark WR, Gao D. Low-molecular weight proteins in end-stage renal disease: potential toxicity and dialytic removal mechanisms. J Am Soc Nephrol 2002;13 Suppl 1:S41- S47. Morgera s. Management of acute renal failure with high permeability haemofiltration in sepsis: practical aspects. International Journal of Intensive Care 2006;autumn: Mariano F, Tetta C, Guida G, et al. Hemofiltration reduces the serum priming activity on neutrophil chemiluminescence in septic patients. Kidney Int 2001;60: Morgera S, Haase M, Rocktaschel J, et al. High permeability haemofiltration improves peripheral blood mononuclear cell proliferation in septic patients with acute renal failure. Nephrol Dial Transplant 2003;18: Morgera S, Klonower D, Rocktaschel J, et al. TNF-alpha elimination with high cut-off haemofilters: a feasible clinical modality for septic patients? Nephrol Dial Transplant 2003;18: Morgera S, Slowinski T, Melzer C, et al. Renal replacement therapy with high-cutoff hemofilters: Impact of convection and diffusion on cytokine clearances and protein status. Am J Kidney Dis 2004;43: Uchino S, Bellomo R, Morimatsu H, et al. Cytokine dialysis: an ex vivo study. ASAIO J 2002;48: Kellum JA, Dishart MK. Effect of hemofiltration filter adsorption on circulating IL-6 levels in septic rats. Crit Care 2002;6: van Bommel EF, Hesse CJ, Jutte NH, et al. Impact of continuous hemofiltration on cytokines and cytokine inhibitors in oliguric patients suffering from systemic inflammatory response syndrome. Ren Fail 1997;19: Schulman G. A review of the concept of biocompatibility. Kidney Int Suppl 1993;41: S209-S

154 Hemofiltration & sepsis Gomez A, Wang R, Unruh H, et al. hemofiltration reverses left ventricular dysfunction during sepsis in dogs. Anesthesiology 1990;73: Mink SN, Jha P, Wang R, et al. Effect of continuous arteriovenous hemofiltration combined with systemic vasopressor therapy on depressed left ventricular contractility and tissue oxygen delivery in canine Escherichia coli sepsis. Anesthesiology 1995;83: DiScipio AW, Burchard KW. Continuous arteriovenous hemofiltration attenuates polymorphonuclear leukocyte phagocytosis in porcine intra-abdominal sepsis. Am J Surg 1997;173: Stein B, Pfenninger E, Grunert A, et al. Influence of continuous haemofiltration on haemodynamics and central blood volume in experimental endotoxic shock. Intensive Care Med 1990;16: Stein B, Pfenninger E, Grunert A, et al. The consequences of continuous haemofiltration on lung mechanics and extravascular lung water in a porcine endotoxic shock model. Intensive Care Med 1991;17: Cole L, Bellomo R, Hart G, et al. A phase II randomized, controlled trial of continuous hemofiltration in sepsis. Crit Care Med 2002;30: John S, Griesbach D, Baumgartel M, et al. Effects of continuous haemofiltration vs intermittent haemodialysis on systemic haemodynamics and splanchnic regional perfusion in septic shock patients: a prospective, randomized clinical trial. Nephrol Dial Transplant 2001;16: Phu NH, Hien TT, Mai NT, et al. Hemofiltration and peritoneal dialysis in infectionassociated acute renal failure in Vietnam. N Engl J Med 2002;347: Sander A, Armbruster W, Sander B, et al. Hemofiltration increases IL-6 clearance in early systemic inflammatory response syndrome but does not alter IL-6 and TNF alpha plasma concentrations. Intensive Care Med 1997;23: Heering P, Morgera S, Schmitz FJ, et al. Cytokine removal and cardiovascular hemodynamics in septic patients with continuous venovenous hemofiltration. Intensive Care Med 1997;23: Hoffmann JN, Hartl WH, Deppisch R, et al. Effect of hemofiltration on hemodynamics and systemic concentrations of anaphylatoxins and cytokines in human sepsis. Intensive Care Med 1996;22: Klouche K, Cavadore P, Portales P, et al. Continuous veno-venous hemofiltration improves hemodynamics in septic shock with acute renal failure without modifying TNFalpha and IL6 plasma concentrations. J Nephrol 2002;15: Kruger I, Jacobi C, Landwehr P. Effects of continuous venovenous hemofiltration on pulmonary function and hemodynamics in postoperative septic multiorgan failure. Contrib Nephrol 1995;116:

155 Chapter Level C, Chauveau P, Guisset O, et al. Mass transfer, clearance and plasma concentration of procalcitonin during continuous venovenous hemofiltration in patients with septic shock and acute oliguric renal failure. Crit Care 2003;7:R160-R166. Meloni C, Morosetti M, Turani F, et al. Cardiac function and oxygen balance in septic patients during continuous hemofiltration. Blood Purif 1998;16: Ossenkoppele GJ, van der MJ, Bronsveld W, et al. Continuous arteriovenous hemofiltration as an adjunctive therapy for septic shock. Crit Care Med 1985;13: Blinzler L, Hausser J, Bodeker H, et al. Conservative treatment of severe necrotizing pancreatitis using early continuous venovenous hemofiltration. Contrib Nephrol 1991;93: Cosentino F, Paganini F, Lockrem J, et al. Continuous arteriovenous hemofiltration in the adult respiratory distress syndrome. Contrib Nephrol 1991;93: Gotloib L, Barzilay E, Shustak A, et al. Hemofiltration in septic ARDS. The artificial kidney as an artificial endocrine lung. Resuscitation 1986;13: Matamis D, Tsagourias M, Koletsos K, et al. Influence of continuous haemofiltrationrelated hypothermia on haemodynamic variables and gas exchange in septic patients. Intensive Care Med 1994;20: Heidemann SM, Ofenstein JP, Sarnaik AP. Efficacy of continuous arteriovenous hemofiltration in endotoxic shock. Circ Shock 1994;44: Murphey ED, Fessler JF, Bottoms GD, et al. Effects of continuous venovenous hemofiltration on cardiopulmonary function in a porcine model of endotoxin-induced shock. Am J Vet Res 1997;58: Ronco C, Bellomo R, Homel P, et al. Effects of different doses in continuous venovenous haemofiltration on outcomes of acute renal failure: a prospective randomised trial. Lancet 2000;356: Lee PA, Matson JR, Pryor RW, et al. Continuous arteriovenous hemofiltration therapy for Staphylococcus aureus-induced septicemia in immature swine. Crit Care Med 1993;21: Ullrich R, Roeder G, Lorber C, et al. Continuous venovenous hemofiltration improves arterial oxygenation in endotoxin-induced lung injury in pigs. Anesthesiology 2001;95: Bellomo R, Kellum JA, Gandhi CR, et al. The effect of intensive plasma water exchange by hemofiltration on hemodynamics and soluble mediators in canine endotoxemia. Am J Respir Crit Care Med 2000;161: Rogiers P, Zhang H, Smail N, et al. Continuous venovenous hemofiltration improves cardiac performance by mechanisms other than tumor necrosis factor-alpha attenuation during endotoxic shock. Crit Care Med 1999;27:

156 Hemofiltration & sepsis Grootendorst AF, van Bommel EF, van Leengoed LA, et al. Infusion of ultrafiltrate from endotoxemic pigs depresses myocardial performance in normal pigs. J Crit Care 1993;8: Grootendorst AF, van Bommel EF, van Leengoed LA, et al. High volume hemofiltration improves hemodynamics and survival of pigs exposed to gut ischemia and reperfusion. Shock 1994;2: Wang H, Zhang ZH, Yan XW, et al. Amelioration of hemodynamics and oxygen metabolism by continuous venovenous hemofiltration in experimental porcine pancreatitis. World J Gastroenterol 2005;11: Yekebas EF, Eisenberger CF, Ohnesorge H, et al. Attenuation of sepsis-related immunoparalysis by continuous veno-venous hemofiltration in experimental porcine pancreatitis. Crit Care Med 2001;29: Freeman BD, Yatsiv I, Natanson C, et al. Continuous arteriovenous hemofiltration does not improve survival in a canine model of septic shock. J Am Coll Surg 1995;180: Cole L, Bellomo R, Journois D, et al. High-volume haemofiltration in human septic shock. Intensive Care Med 2001;27: Jiang HL, Xue WJ, Li DQ, et al. Influence of continuous veno-venous hemofiltration on the course of acute pancreatitis. World J Gastroenterol 2005;11: Laurent I, Adrie C, Vinsonneau C, et al. High-volume hemofiltration after out-ofhospital cardiac arrest: a randomized study. J Am Coll Cardiol 2005;46: Adrie C, Adib-Conquy M, Laurent I, et al. Successful cardiopulmonary resuscitation after cardiac arrest as a «sepsis-like» syndrome. Circulation 2002;106: Honore PM, Jamez J, Wauthier M, et al. Prospective evaluation of short-term, highvolume isovolemic hemofiltration on the hemodynamic course and outcome in patients with intractable circulatory failure resulting from septic shock. Crit Care Med 2000;28: Joannes-Boyau O, Rapaport S, Bazin R, et al. Impact of high volume hemofiltration on hemodynamic disturbance and outcome during septic shock. ASAIO J 2004;50: Oudemans-van Straaten HM, Bosman RJ, van der Spoel JI, et al. Outcome of critically ill patients treated with intermittent high-volume haemofiltration: a prospective cohort analysis. Intensive Care Med 1999;25: Ratanarat R, Brendolan A, Piccinni P, et al. Pulse high-volume haemofiltration for treatment of severe sepsis: effects on hemodynamics and survival. Crit Care 2005;9: R294-R

157 Chapter Wang H, Li WQ, Zhou W, et al. Clinical effects of continuous high volume hemofiltration on severe acute pancreatitis complicated with multiple organ dysfunction syndrome. World J Gastroenterol 2003;9: Honore PM, Matson JR. Short-term high-volume hemofiltration in sepsis: perhaps the right way is to start with. Crit Care Med 2002;30: Mink SN, Li X, Bose D, et al. Early but not delayed continuous arteriovenous hemofiltration improves cardiovascular function in sepsis in dogs. Intensive Care Med 1999;25: Yekebas EF, Treede H, Knoefel WT, et al. Influence of zero-balanced hemofiltration on the course of severe experimental pancreatitis in pigs. Ann Surg 1999;229: Veenman JN, Dujardint CL, Hoek A, et al. High volume continuous venovenous haemofiltration (HV-CVVH) in an equine endotoxaemic shock model. Equine Vet J 2002;34: Morgera S, Rocktaschel J, Haase M, et al. Intermittent high permeability hemofiltration in septic patients with acute renal failure. Intensive Care Med 2003;29: Uchino S, Bellomo R, Goldsmith D, et al. Super high flux hemofiltration: a new technique for cytokine removal. Intensive Care Med 2002;28:

158 Chapter 10 Summary

159 Chapter 10 CVVH is a continuous form of RRT and is predominantly used in the ICU for the critically ill patient with ARF. This thesis addresses several issues of CVVH. A brief introduction to these issues is given in Chapter 1. ARF in the ICU is a serious complication of critical illness and carries a high morbidity and mortality. Critical illness-related ARF rarely presents as an isolated organ dysfunction, but rather as a component of MODS. Increasing evidence suggests that ARF independently contributes to mortality in the critically ill. Unfortunately, there is no consensus on the optimal management of ARF. The optimal mode, dose and timing of RRT are a matter of debate. Chapter 2 reviews the available literature on CRRT and provides evidencebased recommendations for timing, dose and mode of RRT in critically ill patients with ARF. With the exception of dosis, recommendation grades are low because of the quality of the studies. Comparison among the studies is complicated by the use of various definitions of ARF. Furthermore, strategies of timing, dose and RRT mode are likely to interact. However, most of the studies only investigate one of these issues and do not report on the others, making it difficult to draw firm conclusions. In Chapter 3 we describe the results of a randomized controlled two-center study on the effects of three different modes of CVVH (early high-volume, early low-volume and late low-volume) on clinical outcomes in ventilated patients with circulatory insufficiency developing early ARF. No significant differences were observed among the three groups for survival (28- day, ICU- and hospital survival), recovery of renal function and duration of treatment. Of note, compared to literature, survival was favorable in all groups. The efficacy of CVVH depends on the running time of the system and therefore prevention of clotting of the extracorporeal system is an important issue. Understanding the mechanisms involved in premature clotting of the filtration circuit is useful to optimize anticoagulant strategies and maintain filter patency, well minimizing the risk of bleeding complications. Chapter 4 reports the early effects of predilution CVVH without therapeutic anticoagulation on systemic markers of coagulation activation and fibrinolysis. Early thrombin generation was detected in 40% of patients. CVVH without anticoagulation did not change the systemic concentrations of markers of the contact factor pathway or the tissue factor pathway of coagulation, nor did CVVH affect systemic fibrinolysis markers. Therefore, it is conceivable that alternative pathways of thrombin generation are responsible for filter clotting, including the direct activation of factor X, either on the surface of activated platelets, or by the integrin receptor MAC-1 on leukocytes. In Chapter 5 the effects of predilution CVVH on circuit thrombogenesis are compared with postdilution CVVH using the low molecular weight heparin nadroparin as an anticoagulant. At the predeterrmined sample times no signs of platelet activation 158

160 Summary or increased thrombin generation were found during either mode. During postdilution, baseline platelet count and maximal prefilter pressure were inversely related with circuit survival time. Many critically ill patients have serious infections and require treatment with one or more antimicrobial drugs. Drug dosing during CVVH is partly based on the CVVH clearance for the drug of interest, but unfortunately clinical data on this subject are scarce. In Chapter 6 we compare the observed CVVH clearance (calculated from measured data) with the predicted CVVH clearance (calculated from the unbound protein fraction of a drug as reported in the literature) for seven antimicrobial drugs frequently used in the ICU. In addition, we investigated and concluded that antibiotic dosing according to the predicted CVVH removal provides and as reliable estimate than that according to the observed CVVH removal, except for those antibiotics that have both a narrow therapeutic index and a predominantly renal clearance. At present there is no measure of residual renal function in a patient on CVVH, other than diuresis, because urea and creatinine in serum and urine are influenced by the technique itself. Cystatin C, a new marker of glomerular filtration, may be a reliable marker of residual renal function in patients receiving CVVH, however, little is known about its possible extracorporeal elimination. In Chapter 7 we demonstrated that during predilution CVVH (2 L/h) the removed quantity of CysC is less than 30% of its production and that CysC serum concentrations do not change rapidly. These observations suggest that serum CysC can be used to monitor residual renal function during predilution CVVH (2 L/h). CVVH is also applied for some non-renal indications, including sepsis and SIRS, but these indications are far less established than ARF. The mechanism by which hemofiltration in sepsis could be beneficial is through the removal of proinflammatory and antiinflammatory mediators, reducing their toxic effect. The application of hemofiltration as adjunctive therapy in sepsis is, however, still a subject of much debate. In Chapter 8 we investigated the convective and adsorptive removal of cytokines during in vitro hemofiltration using four different membranes both in the predilution mode and the postdilution mode. Reduction of TNF, IL-6 and IL-8 was most impressive with the polyacrilonitrile membrane after four hours of hemofiltration and was largely due to adsorption. Adsorption of TNF, IL-6 and IL-8 was also seen with the other membranes. None of the membranes filtered TNF. Filtration of IL-6 and IL-8 was low with all membranes with only marginal differences among membranes or between filtration modes. In contrast, some of the low-molecularweight inflammatory mediators are, however, readily removed with hemofiltration. 159

161 Chapter 10 In Chapter 9, we reviewed the available data in animals and humans to determine the role of hemofiltration on physiological and clinical effects in sepsis and SIRS, focusing on the role of ultafiltration dose timing and membrane characteristics. Indeed dose and timing appear to be important items but at this moment there is not enough clinical evidence to recommend the application of hemofiltration as adjunctive therapy for sepsis without ARF. The findings so far, however, justify suitably powered RCTs on the effects of early, short-term or pulse hemofiltration using high ultrafiltrate volumes. The same applies to high cutoff hemofiltration as soon as more data on clinical endpoints and safety are available. Regarding the many uncertainties about the optimal volume and type of membrane, clinical studies should first focus on endpoints such as recovery from organ failure and length of treatment, before survival studies are started. Closing remarks Management of ARF in the critically ill is extremely variable and there are no published standards for the provision of renal replacement therapy. Many questions remain to be answered, for example, what is the optimal dose, initiation time, discontinuation time, and anticoagulation strategy of renal replacement therapy? So far, studies on RRT are hampered by the fact that over 35 definitions for ARF are used in the literature, making it impossible to compare studies and to draw firm conclusions. An important step forward in the definition and management of ARF was achieved with the introduction of ADQI. ADQI represents a process of international consensus and evidence-based statements in the definition and management of ARF. In 2002 the first consensus definition for ARF was published (RIFLE criteria). The terminology Acute Kidney Injury has been put forth as the preferred nomenclature to replace ARF with the understanding that the spectrum of AKI is broad and includes different degrees of severity. Since then several clinical studies have shown that the proposed RIFLE classification is suitable for definition of AKI in the ICU and is correlated with hospital mortality. In 2005 the AKIN was introduced, modeled after the ADQI to develop uniform standards for defining and classifying AKI and to establish a forum for multidisciplinary interaction to improve care for patients with, or at risk for AKI. AKIN represents over 20 key societies in critical care and nephrology all over the world, along with additional experts in adult and pediatric AKI. This year the key objective for the AKIN conference was to develop a research agenda in three key areas: epidemiology, initial management and renal replacement therapy in AKI. The issue of dose is being addressed in three large multicenter RCTs: the ATN study in the United States, the RENAL study in Australia/ 160

162 Summary New Zealand and the YVOIRE study in Europe. The results of these studies should be available in 2008 and it seems prudent to wait for the publication of their findings before starting new clinical studies on this subject. Another important issue is whether there is only one optimal dose or if different subgroups of patients need different doses (for example higher doses in sepsis patients). The optimal anticoagulation strategy for the extracorporeal circuit is also not defined and needs further attention. Timing is another important issue, but in order to design future studies on timing it is important to perform studies on the epidemiology of AKI first. The development and introduction of various biomarkers for AKI are also an exciting new research area. Biomarkers could be helpful to define AKI but also to recognize damage to the kidney in an early stage and to evaluate preventive strategies. Moreover, functional biomarkers could be helpful in deciding when to start and stop RRT. The expanding body of research in AKI and the continued emergence of ideas for new research give us hope that better outcomes are in store for the patient of the future. 161

163

164 Chapter 11 Samenvatting

165 Chapter 11 CVVH is een continue vorm van nierfunctievervangende therapie. CVVH wordt voornamelijk toegepast op de ICU bij ernstig zieke patiënten met acuut nierfalen. In dit proefschrift worden verschillende aspecten van CVVH belicht. In Hoofdstuk 1 wordt een korte inleiding gegeven op de in dit proefschrift bestudeerde CVVH onderwerpen. Het ontstaan van acuut nierfalen bij de ICU patiënt is een ernstige complicatie die de morbiditeit en mortaliteit doet toenemen. Acuut nierfalen op de ICU presenteert zich vrijwel altijd in het kader van multipel orgaanfalen en zelden als enkel orgaanfalen. Studies hebben aangetoond dat bij patiënten met multipel orgaanfalen het ontstaan van acuut nierfalen onafhankelijk bijdraagt aan een hogere sterfte. Helaas bestaat er voor acuut nierfalen op de ICU geen consensus ten aanzien van preventie- en behandelingsstrategieën, inclusief nierfunctievervangende therapie. In Hoofdstuk 2 worden evidence-based aanbevelingen gegeven voor timing, dosis en modus van nierfunctievervangende therapie bij de ernstig zieke ICU patiënt, gebaseerd op de huidige literatuur. Behalve voor dosering, is het niveau van aanbevelingen ten gevolge van de kwaliteit van de studies laag. Het onderling vergelijken van studies wordt bemoeilijkt door de grote verscheidenheid aan definities voor acuut nierfalen. Een ander probleem is dat timing, dosis en modus elkaar waarschijnlijk beïnvloeden, terwijl de meeste studies slechts één onderwerp bestuderen en niet de andere aspecten belichten, waardoor het moeilijk wordt om harde conclusies te trekken. Hoofdstuk 3 beschrijft een prospectieve gerandomiseerde studie over de toepassing van drie verschillende CVVH modi (vroeg hoog-volume, vroeg laag-volume en laat laag-volume CVVH) bij beademde patiënten met vroeg acuut nierfalen. Gekeken werd naar de effecten op overleving (28-dagen, ICU en ziekenhuis overleving), herstel van nierfunctie en duur van de CVVH-, IC- en ziekenhuisbehanding. Er waren geen significante verschillen tussen de drie groepen. In vergelijking met eerder gepubliceerde studies was de overleving in de drie groepen opmerkelijk hoger. Preventie van vroegtijdige stolling van het extracorporele circuit is essentieel voor een efficiënte CVVH behandeling. Met behulp van (systemische) antistolling streeft men naar een maximale filteroverleving enerzijds en minimale bloedingcomplicaties anderzijds. Voor een optimaal beleid in antistolling is het belangrijk om inzicht te hebben in de mechanismen die ten grondslag liggen aan vroegtijdige filterstolling. Hoofdstuk 4 beschrijft de vroege effecten van predilutie CVVH op systemische stolling- en fibrinolyse markers. Tijdens de studie werd geen systemische antistolling gegeven omdat dit de concentratie van de onderzochte markers zou kunnen beïnvloeden. Vroege thrombinegeneratie werd gezien bij 40% van de patiënten. CVVH had geen effect op de systemische markers van intrinsieke (via contactactivatie) en extrinsieke (via weefselfactor) stolling, noch 164

166 Samenvatting op de systemische markers van fibrinolyse. Mogelijk wordt thrombine gegenereerd via alternatieve routes, bijvoorbeeld via directe activatie van factor X op geactiveerde thrombocyten, of via de MAC-1 receptor op leukocyten. In Hoofdstuk 5 werd stolling in het extracorporele circuit tijdens postdilutie CVVH vergeleken met stolling tijdens predilutie CVVH. De patiënten werden behandeld met een standaard dosis laag moleculair heparine (nadroparine). Bloedmonsters werden afgenomen op vooraf vastgestelde tijden en toonden geen tekenen van thrombocytenactivatie noch van thrombinegeneratie, in beide CVVH opstellingen. De overlevingsduur van het circuit in de postdilutie opstelling was omgekeerd evenredig aan het aantal thrombocyten voor aanvang van de procedure en aan de maximale prefilterdruk. De ziekte van ICU patiënten wordt vaak gecompliceerd door infecties waarvoor behandeling met één of meerdere antibiotica noodzakelijk is. Bij het doseren van antibiotica tijdens CVVH moet rekening worden gehouden met de CVVH klaring van het betrokken geneesmiddel. Helaas zijn klinische data over dit onderwerp schaars. In Hoofdstuk 6 wordt voor zeven antimicrobiële middelen die frequent op de IC gebruikt worden, de waargenomen CVVH klaring (berekend met behulp van waargenomen data) vergeleken met de voorspelde CVVH klaring (berekend met behulp van de in de literatuur opgegeven niet gebonden eiwitfractie van een geneesmiddel). Vervolgens is de dosisaanpassing die gebaseerd is op de voorspelde CVVH klaring vergeleken met die gebaseerd op de waargenomen CVVH klaring. In deze studie werd aangetoond dat dosisaanpassing gebaseerd op de voorspelde CVVH klaring een voldoende betrouwbare inschatting geeft om toegepast te kunnen worden in de dagelijkse praktijk, met uitzondering van die antibiotica met een smalle therapeutische breedte en overwegend renale klaring. Tijdens behandeling met CVVH bestaat er geen goede maat voor de restfunctie van de nier, anders dan de diurese, aangezien het kreatinine en ureum gehalte in bloed en urine worden beïnvloed door CVVH. Cystatine C in het bloed is een nieuwe marker voor glomerulaire filtratie en zou wellicht een goede maat zijn voor rest nierfunctie tijdens CVVH. Tot op heden, is er weinig bekend over de mogelijke eliminatie van cystatine C tijdens CVVH. In Hoofdstuk 7 tonen we aan dat tijdens predilutie CVVH (2 L/h) de verwijderde hoeveelheid cystatine C minder dan 30% van de productie is en dat de cystatine C concentratie in het bloed niet snel verandert. Deze waarnemingen suggereren dat tijdens predilutie CVVH (2 L/h) cystatine C gebruikt kan worden om de restfunctie van de nieren te monitoren. 165

167 Chapter 11 Een enkele keer wordt CVVH op de ICU toegepast voor niet-renale indicaties, in het bijzonder sepsis en SIRS. Het toepassen van CVVH bij sepsis is gebaseerd op de gedachte dat CVVH pro-inflammatoire en anti-inflammatoire mediatoren verwijdert waardoor het toxische effect van deze mediatoren wordt gereduceerd. Echter, CVVH voor sepsis zonder acuut nierfalen is een onderwerp van veel discussie. In Hoofdstuk 8 wordt de eliminatie van cytokinen onderzocht in een in vitro hemofiltratie opstelling. Vier verschillende hemofilters werden onderzocht, zowel in de predilutie opstelling als ook in de postdilutie opstelling. De grootste afname in TNF, IL-6 en IL-8 concentratie na vier uur hemofiltratie werd gezien met gebruik van het polyacrilonitriel membraan. Deze afname was grotendeels het gevolg van membraanadsorptie. Adsorptie van TNF, IL-6 en IL-8 werd ook gezien met de andere membranen. Geen enkel filter liet filtratie van TNF zien. De vier hemofilters toonden wel geringe filtratie van IL-6 en IL-8, met slechts geringe verschillen tussen de filters onderling en tussen de predilutie en postdilutie opstelling. Echter, kleinere ontstekingsmediatoren worden wel gemakkelijk middels filtratie verwijderd. Hoofdstuk 9 is een systematische literatuuranalyse van humaneen dierstudies en beschrijft de effecten van hemofiltratie als behandeling voor sepsis/ SIRS op fysiologische en klinische eindpunten. Daarbij is speciaal gelet is op de rol van dosis, timing en membraankarakteristieken. Ultrafiltraatdosis en timing blijken belangrijke onderwerpen te zijn, maar op dit moment is het klinische bewijs te laag om hemofiltratie aan te bevelen als behandeling voor sepsis zonder acuut nierfalen. Echter, de huidige bevindingen rechtvaardigen een gerandomiseerde studie met voldoende statistische power naar het effect van vroege, kortdurende hemofiltratie met hoge ultrafiltraat flows. Voorafgaande aan het opzetten van survival studies is het aan te bevelen eerst klinische studies te doen met als eindpunten herstel van orgaan falen en behandelingsduur, aangezien er nog veel onzekerheid is over de optimale dosis (hoog of zeer hoog) en membraankarakteristieken. Met betrekking tot de hoog doorlaatbare (100 kda) membranen, kan gesteld worden dat er behoefte is aan meer studies gericht op enerzijds klinische eindpunten en anderzijds mogelijke complicaties (bv uitspoelen van eiwitten en stollingsfactoren). Tot slot De behandeling van acuut nierfalen bij ernstige zieke IC patiënten varieert enorm en er bestaat geen consensus voor het toepassen van nierfunctievervangende therapie, waaronder CVVH. Veel vraagstukken zijn onbeantwoord, bijvoorbeeld wat is de optimale dosis, timing en antistolling voor nierfunctievervangende therapie. De studies tot dusver worden belemmerd door het toepassen van meer dan 35 verschillende definities voor 166

168 Samenvatting acuut nierfalen, waardoor het onmogelijk wordt studies onderling te vergelijken en harde conclusies te trekken. De introductie van ADQI was een belangrijke ontwikkeling voor de definiëring van acuut nierfalen. ADQI vertegenwoordigt een werkwijze van internationale consensus en evidence-based verklaringen voor de definitie en behandeling van acuut nierfalen. In 2002 publiceerde ADQI de eerste consensus definitie van acuut nierfalen (de RIFLE criteria). De term acute nierschade (ANS) werd geïntroduceerd in plaats van acuut nierfalen. De term ANS onderkent dat het spectrum van deze aandoening breed is en verschillende gradaties omvat. Verscheidene studies hebben de werkzaamheid van de RIFLE criteria op de ICU aangetoond en ook de correlatie van deze criteria met ziekenhuismortaliteit. In 2005 werd AKIN geïntroduceerd als opvolger van ADQI. AKIN vertegenwoordigt meer dan 20 intensive care en nefrologieverenigingen als ook experts op het gebied van acute nierschade. Het doel van AKIN is niet alleen om te komen tot consensus op het gebied van nomenclatuur, classificatie en behandeling van ANS, maar ook om de belangrijkste vraagstellingen te formuleren en te vertalen in studies. De huidige speerpunten zijn de epidemiologie van ANS, de initiële behandeling van ANS en de toepassing van nierfunctievervangende therapie voor ANS. Op dit moment lopen er drie grote trials die zich richten op de dosis van nierfunctie vervangende therapie bij ANS: De ATN studie in de Verenigde Staten, de Renal studie in Australië/Nieuw Zeeland en de YVOIRE studie in Europa. Het lijkt verstandig om de resultaten van deze studies af te wachten alvorens nieuwe studies op te zetten. Een andere belangrijke vraag is of er slechts één optimale dosis is, of dat er subgroepen bestaan die baat hebben bij een andere dosis (bv. hogere dosis in patiënten met sepsis en ANS). De streefdosis en overleving van het extracorporele circuit zijn nauw met elkaar verbonden. Stolling van het extracorporele circuit is de achillespees van de continue nierfunctievervangende behandelingen en ook op dit gebied is er behoefte aan meer onderzoek. Naast dosis is ook timing een belangrijk onderwerp. Toch lijkt het verstandig om eerst studies naar de epidemiologie van ANS te doen alvorens studies op te zetten naar de timing van nierfunctie vervangende therapie. Op dit moment komen er steeds meer biomarkers op de markt voor ANS en dit opent vele nieuwe mogelijkheden voor onderzoek. Biomarkers kunnen helpen om ANS in een vroeg stadium op te sporen, maar ook om ANS te definiëren of om de behandeling te monitoren. De ontwikkelingen op het gebied van ANS zijn hoopgevend voor een betere uitkomst in de toekomst. 167

169

170 Dankwoord

171 Dankwoord Tot slot een woord van dank. Dit proefschrift zou er nooit zijn gekomen zonder de hulp en steun van velen. Allen wil ik hiervoor bedanken, maar een aantal mensen wil ik speciaal noemen. Allereerst mijn promotor, prof. dr M.B. Vroom. Beste Margreeth, onze samenwerking begint in 2001 toen jij het hoofd van onze afdeling werd. Je hebt me altijd aangemoedigd om mijn onderzoek te continueren, en je hebt het ook mogelijk gemaakt om promotie en klinisch werk te combineren. Ik ben je zeer dankbaar dat je hebt toegezien op het welzijn van het boekje, maar ook oog had voor dat van mij en mijn gezin. Ook mijn co-promotores dr H.M. Oudemans-van Straaten en dr M.J. Schultz ben ik zeer erkentelijk. Beste Heleen, met heel veel plezier kijk ik terug op onze samenwerking. Jouw steun en kennis gingen veel verder dan CVVH en daar ben ik je heel dankbaar voor. Ik hoop dat onze samenwerking en vriendschap zich nog vele jaren zal voortzetten. Beste Marcus, jouw enthousiasme en ambities op het gebied van onderzoek werkten aanstekelijk en waren voor mij een enorme steun. Gelukkig stopt onze samenwerking niet bij dit proefschrift en verheug ik me op onze nieuwe onderzoeksprojecten. De hooggeleerden prof. dr A.B.J. Groeneveld, prof. dr J. Kesecioglu, prof. dr R de Haan, prof. dr R.T. Krediet, prof. dr M. Levi, prof. dr T. van der Poll wil ik danken voor hun bereidheid plaats te nemen in mijn promotiecommissie en mijn proefschrift kritisch te beoordelen. Beste Josef, ik heb je leren kennen toen ik als IC fellow in het AMC begon. Ik bewaar zeer goede herinneringen aan deze tijd en ik vind het een hele eer om je nu weer terug te zien in de commissie. Niet alleen dank, maar ook veel respect wil ik uitspreken over de patiënten en hun familieleden die aan de onderzoeken hebben meegewerkt in een voor hun vaak zo moeilijke tijd. Uiteraard ben ik ook veel dank verschuldigd aan al mijn co-auteurs. Zij hebben geholpen met het opzetten van studies, het bewerken van bloedmonsters, het interpreteren van resultaten en het kritisch doorlezen van manuscripten. Voor sommige een extra woord. Mijn collega Evert de Jonge heeft mij enorm geholpen met de interpretatie van de resultaten van het stollingsactivatie onderzoek. Beste Evert, daarnaast ben je natuurlijk altijd een gezellige collga en ook als dorpsgenoot kan ik altijd op je rekenen. Ook de input van Joost Meijers, hoofd van het Laboratorium Experimentele Vasculaire Geneeskunde en van Kamran Bakhtiari zijn onontbeerlijk geweest voor het tot stand komen van het stollingsactivatie onderzoek. Onze ziekenhuisapotheker Erik van Kan 170

172 Dankwoord ben ik zeer erkentelijk voor de hulp in de antibiotica klaringsstudie en het opzetten van nieuwe essays. Veel dank ook aan de mensen die mij met de statistiek hebben geholpen, in het bijzonder prof dr J.G.P Tijssen en dr. J.C. Korevaar. Beste Joke, de afgelopen twee jaar heb ik veel van je geleerd, vooral dat je hulp onmisbaar is. Als laatste een speciaal woord van dank aan Marije Baas en Frans Hoek voor hun hulp in het cystatine C onderzoek. Beste Marije, ik hoop dat we onze plezierige samenwerking kunnen voortzetten tijdens jouw promotieonderzoek. Uiteraard gaat mijn dank ook uit naar de mensen met wie ik dagelijks werk, in het bijzonder mijn IC-collegae en de afdelingssecretaresse Mary-Anne Simons. Lieve Mary- Anne, je bent onze steun en toeverlaat. Je hebt altijd een luisterend oor en een vriendelijk woord of het nu over werk of privé gaat. Een bijzonder woord van dank ook aan alle IC-verpleekundigen. Door jullie inzet is CVVH een succes geworden en uiteraard dragen jullie bij aan de dagelijkse werkvreugde. Een bijzonder woord van dank voor mijn opleider prof. dr Chris Stoutenbeek die tot groot verdriet van velen in 1998 veel te vroeg overleden is. Toch ben ik dankbaar voor de twee jaar die ik met hem mocht samenwerken. Hij was voor mij een groot voorbeeld en mede door zijn inzet is de dosisstudie (hoofdstuk 3) en de in vitro studie (hoofdstuk 8) gerealiseerd. Veel dank ook aan mijn lieve paranimfen Claire Bouman en Anne-Cornelie de Pont. Lieve Claire, ik kan me geen lievere zus wensen. Eigenlijk ben je heel mijn leven al mijn paranimf. Ik hou van je en ben trots op je. Lieve AC, al jaren breng je letterlijk en perikelen van de dag en lachen we samen heel wat af. Papa et maman cette thèse est dédie a vous. Merci pour votre éternel amour et soutien. Last but by no means least, I thank the four most important men in my life. Every day they give me the greatest gift: to love and be loved in return. Nick, Chris en Luke, being you until the moon and back. Dear Art, the day we met on the tennis court you turned my life upside down. Thank you for everything we share. My heart belongs to you. 171

173

174 Curriculum vitae