Nephrology Dialysis Transplantation

Nephrol Dial Transplant ( 1997) 1: 11 117 Original Article Nephrology Dialysis Transplantation Comparison of lactate and bicarbonate buered haemofiltration fluids: use in critically ill patients A. N. Thomas1, J. M. Guy, R. Kishen1, I. F. Geraghty1, B. J. M. Bowles1 and P. Vadgama Departments of 1Intensive Care and Chemical Pathology, Hope Hospital, Salford, UK Abstract Key words: acidosis; bicarbonate; critical illness; Objective. To compare acid base balance, lactate concentration, and haemodynamic and O transport vari- haemofiltration; lactate ables during haemofiltration with replacement fluid containing 44.5 mmol/l Na+ lactate or 40 mmol/l Na+ HCO and 3 mmol/l lactic acid. Introduction 3 Design. A prospective, randomized trial. Setting. A multidisciplinary, adult intensive care unit During continuous venovenous haemofiltration venous in a university hospital. blood is actively passed over a haemofilter membrane. Patients. Forty acidotic patients who required haemowith infused fluid before the blood returns to the Ultrafiltrate lost through the membrane is replaced filtration, were dependent on mechanical ventilation, and had PA catheters in situ. patient. Increasing this exchange of fluid to up to 40 Interventions. During haemofiltration patients received litres a day produces creatinine clearances of over lactate or bicarbonate replacement fluid at a mean rate 30 ml/min [1] and, more significantly, also leads to of 1.7 l/h (SD 0.3).Arterial blood gases, plasma lactate, improvements in survival []. The use of large volumes and haemodynamic and O transport variables were of replacement fluid makes the individual constituents measured before and after 1 and 4 h haemofiltration. of the fluid very important. One such constituent is Ultrafiltrate was collected for lactate estimation. sodium lactate; this is added to facilitate in vivo Measurements and main results. As means (SD). The buering of H+ ions. It is metabolized primarily in net gain of lactate was 63 mmol/h (1 mmol ) with the liver and skeletal muscle to produce bicarbonate; Na+ lactate and 0 mmol/h (0.3 mmol ) with Na+ hence its activity as a buer [3]. A previous study HCO. There was a significant increase in ph and suggested that the metabolism of this exogenous lactate 3 [ lactate] in both groups, but [lactate] was higher in is impaired in critical illness and that the accumulation patients receiving lactate. Twenty-one patients survived of infused lactate may even produce a paradoxical to ICU discharge, these patients were significantly less acidosis [4]. Animal studies also suggested that, inde- acidotic after filtration (lactate group: 0 h: ph 7.3 pendently of any associated acidosis, the lactate ion is (0.09), [lactate].4 mmol/l (1.7);1 h: ph 7.34 (0.09), itself a myocardial depressant [5]. It was therefore felt [ lactate] 4.7 mmol/l (.4); 4 h: ph 7.36 (0.07), [ lactcardiac output and organ perfusion and that this would 3 that the accumulation of lactate may have depressed ate] 4.7 mmol (.7). HCO group: 0 h: ph 7.3 (0.09), [ lactate].3 (1.3); 1 h: ph 7.3 (0.06), [ lactate] have caused the paradoxical metabolic acidosis..9 mmol/l (1.8); 4 h: ph 7.35 ( 0.08), [ lactate] In a previous study of haemofiltration with replace-.8 mmol/l (.0). Base deficit: survivors: 0 h: 9 mmol/l ment fluid containing 44.5 mmol/l lactate we showed (4); 1 h: mmol/l (3). Non-survivors: 0 h: 10 that after 4 h of haemofiltration non-survivors had mmol/l (3); 1 h: 6 mmol/l (3)). Haemodynamic and higher plasma lactate concentrations and were more O transport variables were not significantly aected acidotic than survivors [6]. We were unable to exclude by treatment group or outcome. the possibility that this was due to a partial failure of Conclusions. The degree of correction of acidosis during lactate metabolism in the non-survivors. This led us to the first 4 h of haemofiltration was determined by compare acid base balance and haemodynamic and patients outcome but was not aected by the substituwith filtration fluid containing either bicarbonate or oxygen transport variables after 4 h of haemofiltration tion of bicarbonate- for lactate-containing replacelactate. If patients had been unable to metabolize ment fluids. lactate normally then we would have expected to see a more rapid improvement in acid base balance in the Correspondence and oprint requests to: Dr A. N. Thomas. c/o Department of Anaesthesia, Hope Hospital, Stott Lane, Salford M6 patients receiving bicarbonate. We would have also 8HD, UK. have expected patients receiving bicarbonate to have 1997 European Renal Association European Dialysis and Transplant Association

Lactate and bicarbonate buered haemofiltration fluids in critically ill patients 113 had better cardiac function if the lactate ion was a myocardial depressant. Subjects and methods The study was approved by the hospital ethical committee. The need for informed consent was waived because it was anticipated that all patients entered into the study would be sedated and would therefore be unable to give or withhold consent. Patients contained 44.5 mmol/l of sodium lactate, with more than 98% being present as the L- isomer. The bicarbonate fluid, prepared by the same commercial source, had the same background solute composition except that it contained no sodium lactate; it was, however, necessary to add 3 mmol of lactic acid to each litre of the fluid during its manufacture to maintain the stability of the solution; 00 ml of 8.4% (w/v) sodium bicarbonate was then added to each 5-litre bag of this fluid on the intensive care unit prior to the fluid s use. Potassium chloride was also added to both fluids immediately before use to maintain the serum potassium between 4.5 and 6.0 mmol/l. The electrolyte concentrations of the two fluids are summarized in Table. Rates of fluid replacement and ultrafiltrate production were adjusted to patients individual needs. Heparin and/or prostacyclin were infused proximally to the filter to avoid clot formation within the extracorporeal circuit. Before entering the study patients had already undergone fluid resuscitation with blood and gelatin solutions. This resuscitation was guided by haemodynamic and oxygen transport measurements. Systemic and pulmonary artery pressures and cardiac output measurements were made using a Vygon 19 0 G femoral or radial artery cannula and a 7.5 F Opticath PA catheter attached to a Thermoset closed loop injectate system (Abbott Labs, Chicago, USA) and a Marquette Tramscope 1c monitoring system(marquette Electronics, Milwaukee, USA). In addition to this, 38 of the 40 patients were receiving positive inotropes to augment cardiac output, and 0 of these patients were also receiving noradrenaline to maintain an adequate mean arterial pressure. All of the patients also required mechanical ventilation at the time that they entered the study. Forty-one patients were randomly allocated to receive either sodium lactate or sodium bicarbonate replacement fluids during their first episode of haemofiltration following admis- sion to the intensive care unit. The patients had undergone a period of resuscitation but remained acidotic with a base deficit of more than 5 mmol/l in all cases. Details of the primary admission diagnoses of the patients receiving the two replacement fluids are shown in Table 1. Filtration was initiated either for the treatment of acute renal failure, for control of fluid balance, or for the treatment of severe sepsis or for a combination of these indications. Vascular access for filtration was obtained using a doublelumen catheter ( Dualyse-cath, Vygon, Ecouen, France.) placed in a central vein. Filtration was then established using a Hospal BSM haemofilter pump and a 1. m polyacry- lonitrate Filtral 1 AN69 hollow-fibre filter (Hospal, Basle, Switzerland ). The pump speed was set to produce flows of about 100 ml/min of blood through the filter. The circuit was adjusted to produce fluid exchange rates of between 1.5 Protocol and.0 l/min. Haemovex ( Ivex Pharmaceuticals, Larne, N. Ireland) was used as the lactate-containing fluid. This Immediately before haemofiltration, haemodynamic measurements were made and systemic and pulmonary artery Table 1. Patient s primary admission diagnosis. The Table shows the numbers in each diagnostic group in total and divided by the two replacement fluids used. Numbers shown in parentheses are the numbers in each group surviving to ICU discharge Primary diagnosis Na+ lactate Na+ HCO 3 Total blood was sampled for blood gas analysis as well as for haemoglobin and oxygen saturation measurement (oximeter 8 and blood gas analyser BGM 131, Instrumentation Laboratory, Lexington, MA, USA). Derived haemodynamic and oxygen transport data were calculated from standard formulae [7]. Further aliquots of systemic arterial blood group group were then sampled and placed in fluoride oxalate. These were initially stored at 5 C and subsequently centrifuged within 4 h. The plasma sample was then analysed for L-lactate Large-bowel perforation 3 () 6 ( 3) 9 (5) Perforation of stomach (1) ( 0) 4 (1) using a YSI 300 Stat Plus Lactate analyser ( Yellow Springs or duodenum Instrument Co Inc, OH, USA ). Systemic arterial blood was Hospital-acquired 3 () ( ) 5 (4) also sampled for estimation of plasma sodium, chloride and pneumonia Community-acquired () 1 ( 0) 3 () pneumonia Surgery for abdominal (1) ( 0) 4 (1) Table. Composition of sodium lactate and sodium bicarbonate solutions at the time of use (concentrations expressed as mmol/litre) aortic aneurysm Urinary tract infection 1 (1) ( ) 3 (3) Bowel necrosis 1 (0) 1 (1) (1) Heart failure () 0 (0) () Constituent Na+ lactate fluid Na+ HCO 3 Oesophageal leak (0) ( 0) 4 (0) Septic abortion 0 (0) 1 ( 0) 1 (0) Pancreatitis 1 (0) 0 (0) 1 (0) (mmol/l ) (Haemovex ) fluid Na+ Lactate 44.5 0 Rhabdomyolysis 0 (0) (0) (0) Lactic acid 0 3 with drug overdose Bicarbonate 0 40 Smoke inhalation 1 (0) 0 ( 0) 1 (0) Sodium 14 155 Total 0 (11) 1 (11) 41 ( ) Chloride 103 10 Calcium.0 1.8 Patient 38 was admitted with both urinary tract infection and Potassium 0 6 0 6 hospital-acquired pneumonia. She received bicarbonate fluid and Magnesium 0.75 0.77 survived.

114 A. N. Thomas et al. creatinine concentrations, liver function tests, prothrombin two periods of filtration, using the two replacement time, and haemoglobin, white cell, and platelet counts. fluids are shown in Table 3. The bicarbonate- Haemofiltration was then commenced as described above containing fluid produced no significant net loss or and measurements of lactate concentration, blood gases and gain of lactate, while the lactate-containing fluid prohaemodynamic and oxygen transport variables were then duced a net gain of 65 mmol of lactate per hour. In repeated as near to 1 h after the start of filtration as was clinically convenient. A third set of these measurements was addition to this infused lactate, three patients in the then obtained at 4 h, and plasma sodium and chloride were lactate group and two patients in the bicarbonate also estimated at this time. All patients then received conven- group also received citrated blood (range 1 4 units). tional sodium lactate replacement fluid for the remaining Standard assessments of liver function made prior to time that they required haemofiltration. In addition to the filtration showed no evidence of gross liver dysfunction blood samples described, samples of ultrafiltrate were also in any patient. The mean prothrombin ratio was 1.5 collected each hour into fluoride oxalate tubes. Samples were (range 1 3.8), the mean [bilirubin] was 17 mmol/l then mixed for each 1-h period, the volume used from each (range 43), the mean [alkaline phosphatase] was 84 sample was a constant percentage of the total volume of iu/l. (range 17 35) and the median [alanine transamiultrafiltrate produced in that hour. The L-lactate concentranase] was 3 iu/l. (range 3 195). High transaminase tion in these mixed samples was then measured and this was used, together with the recorded volume of ultrafiltrate, to concentrations resulted from rhabdomyolysis and not estimate the amount of lactate removed by the filter over from liver cell damage in this group of patients. The each time period. A record was also kept of the number of mean pulmonary artery temperature was 36.6 C (SD times filtration was interrupted by filters clotting, the volume 0.9) before filtration and was 36.1 C (SD 0.7) during of replacement fluid infused, the urine output, and the units filtration; lactate metabolism was therefore unlikely to of transfused citrated blood. Patients other fluid require- have been significantly aected by changes in body ments were met with saline, gelatin, and parenteral and temperature. A total of 4 haemofilter circuits had to enteral nutrition but not with additional lactate-containing be changed in 15 patients (6 patients in the bicarbonate fluids. The patients were subsequently divided into those group, 9 in the lactate group, range 1 3 filters per who survived to be discharged from the ICU and those who patient) during the 4-h study period. died before discharge. Changes in acid base balance and lactate concentration are shown in Figure 1. There was no significant Statistical analysis change in the pco in either group; there was, however, a significant increase in ph and a reduction in base Within-subject comparisons between before and after the deficit in both groups; there were no significant dierstart of filtration were made using a repeated measures ences between the two groups with respect to these analysis of variance. Between-subject comparisons were made variables. The plasma lactate concentration increased using chi-squared tests for categorical data and unpaired Student s t tests for normally distributed data. Where signimore marked in patients who received sodium lactate, significantly in both groups, but this increase was much ficant dierences were found for variables measured at one of the three time points, if the variable was measured at all so that the lactate concentration was significantly three points, then the dierence was only described as higher in these patients after both 1- and 4-h filtra- significant if the pooled values for all three points were also tion. Values for haemodynamic and oxygen transport significantly dierent between the groups. Results were con- variables are shown in Table 4; there was no significant sidered to be significant if the P value was less than 0.05. dierence in these variables between either of the two treatment groups or between measurements made before or after the start of haemofiltration. Results In addition to dierences in lactate concentrations Forty-one patients were enrolled into the study over a Table 3. Fluid and lactate gain and loss during haemofiltration with 1-month period (0 received bicarbonate and 1 bicarbonate and lactate replacement fluids received sodium lactate). There was no significant dierence between the two groups with respect to age, Variable Bicarbonate group Lactate group sex, or admission APACHE II score. ( Lactate group: age 66 (11) years, APACHE II score 0 (8), 11 males, 0 1 h 1 4 h 0 1 h 1 4 h 10 females. Bicarbonate group: age 6 ( 16) years, APACHE II score (8) 10 males 10 females.) One Ultra filtrate volume 0.9 0.1 1. 0. patient in the lactate group was withdrawn from the ( l/1 h) (3.5) ( 6.9) (3.6) ( 4.9) trial after 4 h because of a rising blood lactate and Replacement volume 1.1 0.1 0.5 19. increasing acidosis; she was then treated with bicarbon- (l/1 h) (3.) (5.7) (3.4) (4.3) Ultrafiltrate [ lactate].9 5.3 3.1 5.5 ate filtration but died 1 h later. Of the remaining 40 (mmol/l ) (.3) (3.3) (.0) (.6) patients one patient in each group had treatment Lactate infused 5.4 5.3 73.1 75.0 withdrawn after 1 h filtration, leaving 38 patients, 1 (mmol/h) (0.6) (0.7) (13.0) (1.3) of whom were eventually discharged from the intensive Lactate cleared in 5.1 5.8 8.0 10.0 ultrafiltrate (mmol/h) (3.3) (4.4) (4.) (5.1) care unit. The patients primary admission diagnoses Net gain of lactate 0.3 0.3 6.1 65.0 are shown in Table 1. (mmol/h) ( 3.4) ( 4.0) (18.7) ( 10.7) The fluid and lactate balances achieved during the

Lactate and bicarbonate buered haemofiltration fluids in critically ill patients 115 also went on to compare variables in patients who survived to be discharged from the ICU and in patients who died before discharge. The survivors were significantly younger than the non-survivors; however, there was no significant dierence in admission APACHE II scores [8] or sex distribution between the groups (survivors, mean age 57 years (SD 13 years), mean APACHE II score, 0 (SD 7) m/f 13/8; non-survivors, mean age 7 years (SD 7 years) mean APACHE II score, (SD 7) m/f 8/11). Changes in ph, base excess and lactate with haemofiltration are shown in survivors and non-survivors in Figure. ph and base deficit were not significantly dierent between the two groups before filtration; however, highly significant dierences had developed after 1 h filtration, and the survivors still had a significantly smaller base deficit after 4 h filtration. The lactate concentration was significantly greater in the non-survivors prior to filtration and after 1 h filtration. Haemodynamic and oxygen transport variables were also divided by outcome, of these variables only the LVSWI, after 4 h, showed any significant dierence between the survivors and the non-survivors (survivors, 44 (SD 16) g mm ; non-survivors, 34 (SD14) g mm ). Fig. 1. Plasma lactate concentration, ph, base deficit, and pco before haemofiltration and after 1 and 4 h haemofiltration in patients receiving Na+ lactate & or Na+ bicarbonate a replacement fluids. (*=significant change from prefiltration value; &=significant dierence between the two patient groups). Discussion Plasma lactate concentration is determined by the balance between the rate at which lactate enters the circulation and the rate that it is cleared, primarily as a result of hepatic and skeletal muscle metabolism. It is therefore not surprising that the net infusion of over 60 mmol/h of Na lactate should result in plasma lactate concentrations in the range of 3 5 mmol/l during filtration. Two previous studies have also shown similar lactate concentrations in critically ill patients on established haemofiltration [4,9]. The results of the first of these studies also suggested that this hyperlactaemia was associated with hypotension and a worsening acidosis; these findings were not, however, confirmed in the second study. Dierences between the results in the two studies could have been caused by the eects of abnormal liver function on lactate metabolism in patients in the first study or by a more aggressive approach taken to avoid covert hypoperfusion in patients in the second study. In a previous study of patients receiving sodium lactate we demonstrated that non-survivors remained more acidotic and had a higher plasma lactate concen- between the two fluids, the bicarbonate solution also contained more sodium and chloride at the time of use ( Table ). This resulted in a significantly increased concentration of these ions in patients receiving bicarbonate, so that their plasma concentration was significantly higher than in the lactate group after 4 h filtration. (Bicarbonate group: pre-filtration Na+, 136 mmol/l (SD 6); Cl, 107 mmol/l (SD 8); post- tration than the patients who survived [6 ]. We were filtration Na+, 140 mmol/l (SD ), Cl, 109 mmol/l unable to exclude the possibility that this was because (SD ). Lactate group: Na+ 135 mmol/l (SD6), Cl, these patients were unable to metabolize lactate to 107 mmol/l (SD 7); post-filtration Na+ 137 mmol/l bicarbonate. The results of this present study confirm (SD 3), Cl, 104 mmol/l (SD ).) The concentration the importance of a continuing acidosis during the of these ions was higher in the bicarbonate fluid initial 4 h of filtration as a predictor of outcome. The because it was originally envisaged that only 30 mmol/l results do not, however, suggest that the substitution sodium bicarbonate would be added to this fluid. of bicarbonate for lactate results in any improvement Although the primary aim of this study was to in acid base balance or haemodynamic or oxygen compare the eects of the two replacement fluids, we transport variables. It is therefore likely that the

116 A. N. Thomas et al. Table 4. Cardiac index, SVR, LVSWI, oxygen delivery and consumption together with the dose of dobutamine before and after 1 and 4 h haemofiltration with lactate- or bicarbonate-containing replacement fluids Variable Prefiltration Post 1 h filtration Post 4 h filtration Lactate HCO 3 Lactate HCO 3 Lactate HCO 3 fluid fluid fluid fluid fluid fluid Cardiac index 4.0 4.3 3.6 4.0 3.8 4.1 (l/m/min) (1.) (1.4) (0.9) (1.0) (1.) (1.) SVR 77 78 946 8 943 758 (dyne/s/cm5) (31) (37) (83) (0) (400) (307) LVSWI 31 37 37 37 38 40 (g/m/m) (11) (16) (14) (15 ) (15) (17) Dobutamine 7.0 3.6 8. 3.4 7.5 4.1 (mg/kg/min) (6.7) (5.0) (7.7) (4.4) (7.7) (5.1) Oxygen delivery 610 651 556 593 556 651 (ml/m/min) (154) (08) (14) (150 ) (158) (09) O consumption 116 130 13 116 1 138 (ml/m/min) (41) (41) (59) (3 ) (37) (55) when compared with lactate solutions. This is because of the time taken and potential risk in having to add the bicarbonate immediately prior to use. It seems unlikely that the dierences in sodium and chloride concentrations could have masked such a benefit. Although we anticipated the increased lactate concentration in patients who received sodium lactate, we did not anticipate the increase seen in patients who received bicarbonate. This small increase in lactate concentration with bicarbonate could be explained by the process of haemofiltration either increasing the anaerobic production of lactate or reducing liver and skeletal blood flow to reduce the clearance of lactate. However, as we found no reduction in oxygen delivery or change in the oxygen extraction ratio, a more likely explanation is that the increase in blood ph during haemofiltration inhibited the ph-sensitive active trans- port of lactate into hepatocytes [10]. The relative inability of oxygen transport variables to predict outcome when compared with acid base abnormalities was probably caused by several factors. Firstly several inotropes as well as fluid challenges and adjustments in ventilation were used to try to meet oxygen transport goals, so that although we did not show significant changes to one treatment, it was probably true that oxygen transport variables were maintained in the non-survivors as a result of increas- ing several treatments. Secondly oxygen transport vari- ables are measured at one time point, whereas blood gas variables reflect the patients condition over the preceding few hours before the sample is taken. The dierences in haemodynamic and oxygen transport variables between survivors and non-survivors in this study are very similar to those seen in a recent study of patients in the early stages of critical illness where LVSWI was also shown to be the only variable that predicted outcome [11]. In summary, we studied a group of critically ill, acidotic patients who did not have overt hepatic failure. In these patients the rate of correction of acid base abnormalities during the first 4 h of filtration was not Fig.. Base deficit, ph and plasma [lactate] before haemofiltration and after 1 and 4 h haemofiltration in all patients who died & or survived to ICU discharge a (&=significant dierence between survivors and non-survivors). reduced ability of non-survivors to correct a metabolic acidosis during haemofiltration is caused by a process other than a defect in lactate metabolism. The fact that bicarbonate did not significantly improve acid base balance also suggests that the changes seen in liver and muscle blood flow and metabolism in critical illness are not sucient to aect the conversion of lactate to bicarbonate. These observations should not be extrapolated to patients who are in liver failure as these patients would be expected to have additional problems in handling exogenous lactate; none of the patients that we studied were in liver failure. We also cannot exclude the possibility that the choice of buer solution may have important clinical eects when used for periods in excess of 4 h. Dierences in the sodium and chloride concentrations in the two fluids produced small but significant dierences in the plasma concentrations of these ions after 4 h; however, it seems unlikely that this would have significantly aected acid base balance. For bicarbonate solutions to be used during haemofiltration they would have to show very clear practical benefits

Lactate and bicarbonate buered haemofiltration fluids in critically ill patients 117 aected by the substitution of sodium bicarbonate for 3. Cohen RD. Roles of the liver and kidney in acid-base regulation and its disorders. Br J Anaesth 1991; 67: 154 164 sodium lactate in the haemofiltration replacement fluid. 4. Davenport A, Will EJ, Davidson AM. Hyperlactataemia and A delay in the reversal of a metabolic acidosis in these metabolic acidosis during haemofiltration using lactate buered patients is, however, a useful predictor of outcome. fluids. Nephron 1991; 59: 461 465 5. Yatani A, Fujino T, Kinoshitu K, Goto M. Excess lactate modulates ionic currents and tension components in frog atrial Acknowledgements. The authors would like to thank Ivex muscle. J Mol Cell Cardiol 1981; 13: 147 161 Pharmaceuticals for their help in supplying the lactate-free fluid. 6. Thomas AN, Guy JM, Kishen R, Bowles BJM, Vadgama P. They would also like to thank the nursing sta for collecting the Lactate concentration and acid-base balance during haemofilultrafiltrate samples, Ms Fiona Campbell for statistical advice, and tration with lactate buered fluid. Clin Intensive Care 1995; Professor R. A. Little for reviewing the paper. 6: 78 8 7. Nightingale P. Practical points in the application of oxygen transport principles. Intensive Care Med 1990; 16: 5173 517 8. Knaus WA, Draper EA, Wagner DP, Zimmerman JE. APACHE References II: a severity of disease classification system. Crit Care Med 1985: 13: 818 89 9. Nimmo GR, Mackenzie SJ, Walker S, Grant IS. Acid-base 1. Brocklehurst I, Thomas AN, Guy JM, Kishen R, Creatinine responses to high volume haemofiltration in the critically ill and urea clearance during continuous veno-venous haemofiltra- patient. Nephrol Dial Transplant 1993; 8: 854 857 tion in 16 critically ill patients. Anaesthesia 1996 10. Edlund GL, Halestrap AP. The kinetics of transport of lactate. Stock M, Hartl WH, Zimmerer E, Inthorn D. Comparison of and pyruvate into rat hepatocytes. Biochem J 1988; 49: 117 33 pump driven and spontaneous haemofiltration in post-operative 11. Metrangolo L, Fiorillo M, Friedman G et al. Early hemodynamic acute renal failure. Lancet 1991; 337: 45 455 course in septic shock. Crit Care Med 1995; 3: 1971 1976 Received for publication: 16.8.96 Accepted in revised form: 30.1.96