EXTRACORPOREAL MEMBRANE OXYGENATION

Transcription

1 Outcome in Patients Who Require Venoarterial Extracorporeal Membrane Oxygenation Support After Cardiac Surgery Hesham A. Elsharkawy, MD, MSc,* Liang Li, PhD, Wael Ali Sakr Esa, MD, Daniel I. Sessler, MD, and C. Allen Bashour, MD* Objective: The authors analyzed hospital mortality in adult cardiac surgery patients who required postoperative venoarterial extracorporeal membrane oxygenation (ECMO) support for circulatory failure and identified perioperative patient variables associated with hospital mortality in these patients. Design: A retrospective study. Setting: A single institution, tertiary academic center. Participants: Adult patients requiring venoarterial ECMO support after cardiac surgery from January 1995 to December 2005 were identified from the Anesthesiology Institute Patient Registry. Twenty-two preselected patient variables were entered into a logistic regression model of hospital death. Interventions: None. Results: Two hundred thirty-three of 40,116 (0.58%) adult cardiac surgery patients required postoperative venoarterial ECMO, and among these, 149 (64%) died in the hospital. In an unadjusted analysis, older age, higher preoperative albumin, diabetes history, coronary artery bypass graft surgery, EXTRACORPOREAL MEMBRANE OXYGENATION (ECMO) is a means of providing temporary support in cases of severe cardiopulmonary failure refractory to medical therapy. 1-7 Venoarterial ECMO (VAE) is used in hemodynamically unstable patients for circulatory support and to improve gas exchange in patients with severe respiratory failure. Although the indications for VAE are better understood now than when it was first introduced more than 40 years ago, a clear classification of patients who are most likely to benefit from postoperative VAE support has been elusive; thus, there remains wide variability in its application. In fact, only 1 prospective VAE trial in adults has been reported. 5 Mortality remains high among patients who require VAE after cardiac surgery, and perioperative variables that may be useful to predict outcome remain poorly defined. This investigation was undertaken to analyze morbidity and mortality in adult cardiac surgery patients who required postoperative VAE support and to characterize a patient group in which postoperative VAE conferred benefit. MATERIALS AND METHODS With approval from the Institutional Review Board and waiver of the requirement for individual patient consent, adult patients who received VAE support after cardiac surgery during an 11-year period beginning From the *Department of Cardiothoracic Anesthesia, Quantitative Health Sciences, Anesthesiology Institute, and Outcomes Research, Cleveland Clinic, Cleveland, OH. Address reprint requests to Hesham Elsharkawy, MD, MSc, Departments of General Anesthesia, Anesthesiology Institute, Cleveland Clinic, 9500 Euclid Avenue/E-31, Cleveland, OH elsharh@ccf.org Published by Elsevier Inc / $36.00/0 doi: /j.jvca and longer total cardiopulmonary bypass (CPB) time were associated with increased hospital mortality, and a history of cardiogenic shock was associated with decreased mortality. In an adjusted logistic regression analysis, a history of cardiogenic shock and younger age were associated with decreased hospital mortality. The overall use of postoperative venoarterial ECMO in this patient population decreased since its peak in Conclusion: Venoarterial ECMO support after cardiac surgery was required in a small fraction of patients and was associated with very high hospital mortality; but among those requiring ECMO, mortality in these patients was lower in younger, nondiabetic patients with cardiogenic shock who had shorter CPB times. The mortality associated patient variables identified are not easily modifiable and do not appear sufficiently robust to define which patients should be selected for this potentially life-saving therapy. Published by Elsevier Inc. KEY WORDS: cardiac surgery, outcome, venoarterial extracorporeal membrane oxygenation, ECMO, anesthesia, mortality in January 1995 were identified by using the Anesthesiology Institute Patient Registry. Cardiac surgery patients who required postoperative venovenous ECMO support only for respiratory failure were excluded to maintain a homogenous study cohort. The Anesthesiology Institute s Registry contains demographic and clinical information on patients undergoing cardiac surgery. Information is obtained through a prospective daily review of the medical record, anesthesia records, and clinical care notes and by physical assessment. This information is collected at the patient s bedside in the cardiovascular intensive care unit (ICU). Supplemental demographic and clinical data available in other institutional databases are imported into the registry through manual and electronic interfaces. Data validations are built into the registry, and additional mechanized validations are performed quarterly to identify any quality issues that may not have been identified by the built-in validations. The decision to initiate VAE was made by surgical and ICU attending staff. The following indications for VAE were the most common: cardiac arrest, difficulty weaning from cardiopulmonary bypass (CPB) in the setting of maximum inotropic and intra-aortic balloon pump (IABP) support, refractory cardiogenic shock (systemic arterial pressure 85 mmhg or cardiac index 1.5 L/min/m 2 with adequate preload) with maximum inotropic support, and right-heart failure with or without concomitant left-heart failure. Contraindications included prolonged mechanical ventilation, sepsis syndrome, multisystem organ failure, and contraindication to systemic anticoagulation. VAE insertion occurred either during the initial operation or afterward in the cardiovascular ICU. Although VAE insertion timing may affect outcome, this variable was not analyzed because these data were not available during the entire investigation period. Perioperative patient variables (Table 1) were analyzed for their association with hospital mortality using the Student t test, Wilcoxon rank sum test, chi-square test, or Fisher exact test as appropriate. These variables were entered into a logistic regression model of hospital death. Important predictors were identified through a stepwise model selection procedure. The potential nonlinear effect of continuous covariates was investigated by nonparametric logistic regression. The statistical significance level was set at Journal of Cardiothoracic and Vascular Anesthesia, Vol 24, No 6 (December), 2010: pp

2 OUTCOME AFTER EXTRACORPOREAL MEMBRANE OXYGENATION 947 Table 1. Comparison Between Nonsurvivors and Survivors Variable Nonsurvivor (n 149) Survivor (n 84) p Value Age (y) Median Interquartile range Creatinine (mg/dl) 0.43 Median Interquartile range Albumin (mg/dl)/no of patients* Median Interquartile range Body mass index (kg/m 2 ) 0.18 Median Interquartile range Blood urine nitrogen (mg/dl) 0.37 Median Interquartile range Total CPB time (min) Median Interquartile range Female, no (%) 51 (34.2) 25 (29.8) 0.55 COPD/asthma, no (%) 13 (8.7) 7 (8.3) 0.99 History of lung surgery, no (%) 2 (1.3) 1 (1.2) 0.99 Pulmonary hypertension, no (%) 17 (11.4) 14 (16.7) 0.31 Smoking, no (%) 80 (53.7) 44 (52.4) 0.89 History of CHF, no (%) 81 (54.4) 42 (50.0) 0.57 Cardiogenic shock (new onset), no (%) 39 (26.2) 37 (44.0) Diabetes, no (%) 38 (25.5) 12 (14.3) Normal LVF, no (%) 34 (22.8) 12 (14.3) 0.13 Respiratory failure, no (%) 7 (4.7) 2 (2.4) 0.49 Circulatory failure, no (%) 147 (98.7) 84 (100) 0.54 IABP postbypass, no (%) 15 (10.1) 7 (8.3) 0.82 Redo, no (%) 71 (47.7) 45 (53.6) 0.42 Emergency surgery, no (%) 49 (32.9) 35 (41.7) 0.21 AV replace, no (%) 30 (20.1) 9 (10.7) 0.07 AV repair, no (%) 2 (1.3) 0 (0) 0.54 MV replace, no (%) 14 (9.4) 6 (7.1) 0.63 MV repair, no (%) 16 (10.7) 6 (7.1) 0.49 TV replace/repair, no (%) 10 (6.7) 6 (7.1) 0.99 Any CABG, no (%) 66 (44.3) 20 (23.8) Any valve problem, no (%) 50 (33.6) 19 (22.6) 0.1 Abbreviations: COPD, chronic obstructive pulmonary disease; CHF, congestive heart failure; LVF, left ventricular function; AV, arteriovenous; MV, mitral valve; PV, pulmonary valve; TV, tricuspid valve. *Eighteen missing data were excluded. Patient had respiratory failure before the insertion of ECMO. Patient had only circulatory failure before the ECMO insertion. RESULTS Among 40,116 patients who had cardiac surgery during the investigation period, 233 (0.58%) required postoperative VAE support for hemodynamic instability with or without concomitant respiratory failure. Forty-three patients who received venovenous ECMO solely for respiratory failure were excluded. Among the 233 patients who received postoperative VAE support, 149 (64%) died in the hospital. Patients who had coronary artery bypass graft (CABG) surgery and/or aortic valve surgery had the highest mortality. Twenty-five of the patients supported with VAE subsequently underwent cardiac transplantation. Peripheral cannulation (common femoral artery and vein) was used in 67% of the patients, and central cannulation (ascending aorta and right atrium or common femoral vein) in 33%. Venoarterial ECMO support was converted to an implantable left ventricular assist device in 28 patients. The decision to use VAE versus left and/or right ventricular assist device support was made by surgery staff. Thirty-three percent of the patients had either intraoperative or postoperative cardiogenic shock. Table 1 reports the univariate association between hospital mortality and preoperative and perioperative patient variables. Older age, higher preoperative albumin, diabetes history, CABG surgery, and longer total CPB time were associated with increased hospital mortality. New-onset cardiogenic shock was associated with decreased hospital mortality. The survivors and

3 948 ELSHARKAWY ET AL Table 2. The Final Selected Logistic Regression Model for The Probability of Hospital Death Intercept Age (10 years) Cardiogenic Shock Coefficient SE p Value Odds ratio CI lower CI upper Abbreviations: SE, standard error of the coefficient estimates; CI lower, the lower end of the confidence interval of the odds ratio; CI upper, the upper end of the confidence interval of the odds ratio. nonsurvivors did not differ in terms of sex, body mass index, left ventricular function, blood urea nitrogen, creatinine, past medical history of congestive heart failure, chronic obstructive pulmonary disease/asthma, smoking, pulmonary hypertension, history of lung surgery, reoperation, emergency surgery, or insertion of IABP after CPB. In an adjusted analysis, the authors fit a logistic regression model of hospital mortality using the variables shown in Table 1. These variables were prespecified before the data pull and statistical analysis. Stepwise variable selection was used to select variables that significantly predicted the risk of death in this adjusted analysis. The final selected model (Table 2) includes 2 variables: age (odds ratio 1.52 per decade, p 0.001) and cardiogenic shock (odds ratio 0.52, p 0.028). A nomogram (Fig 1) that shows the relationship among age, cardiogenic shock, and probability of hospital death created from this model, which shows a nearly linear increase in mortality risk with age. The presence of cardiogenic shock at the time of VAE insertion was associated with reduced risk of hospital death at all ages. A nonparametric logistic regression analysis with age confirmed that no age cut point could be identified at which the mortality increased significantly in a nonlinear fashion. In the adjusted analysis, all variables were available for each patient except albumin, which was missing in 18 patients. These missing values were imputed with the predicted mean value in a linear regression of albumin on all other predictors in the logistic regression. Of the 18 post-vae placement outcome variables analyzed, global neurologic deficit, renal failure, multiple system organ failure, and septic shock were associated with hospital mortality (Table 3). The overall VAE use decreased during the investigation period (Fig 2: number of patients who required VAE support compared with total cardiac surgery cases over time). Because this is a retrospective observational study, the univariate association is described here individually without adjusting for multiple comparisons. The median ages of the surviving and nonsurviving patients were 54 and 60 years, respectively (p 0.001). Although the histogram of age by vital status (Fig 3) shows that the nonsurvivors generally were older than the survivors, the statistically significant 6-year difference may be less clinically notable. The number of patients who received VAE support after cardiac surgery decreased during the investigation period (Fig 2). The year of the surgery was added into the logistic regression model for hospital death (the other 2 covariates were patient age and cardiogenic shock). The year of the surgery had a positive coefficient (p ), suggesting that over the investigation period the survival rate for these patients decreased (Table 4: hospital deaths v year of operation). DISCUSSION Although VAE rarely was required after cardiac surgery, the associated mortality in these patients was considerable. The 36% survival rate reported here is comparable to other rates reported including the Extracorporeal Life Support Organization Registry (33%), 8 Magovern et al (52% and 36%), 9,10 and Wang et al (33%). 4 In this investigation, mortality was lowest in younger, nondiabetic patients with reversible cardiogenic shock who had shorter CPB times. Among the ECMO-weaned patients, dialysis for acute renal failure was a significant factor in reducing the chance of survival. The present authors observed a near-linear relationship between age and mortality, with 50% survival at 53 years. Thus, patient selection can be guided by age, diabetic history, CPB time, and cardiogenic shock, but none of these parameters seems sufficient to define which patients should and should not be placed on VAE support. There was a greater than 10% relative mortality increase at all ages for patients not in cardiogenic shock who received VAE; thus, it was most beneficial in patients with reversible cardiogenic shock. Ko et al 3 reported a hospital survival of 30% in a retrospective review of 76 adult cardiac surgery patients who required postoperative ECMO support. These authors emphasized the importance of initiating ECMO support early in the setting of cardiogenic shock that is refractory to maximum inotropic and pressor support and/or IABP counterpulsation. Combes et al 11 reported a 40% survival rate in 81 mixed Fig 1. A nomogram predicting the probability of hospital death among extracorporeal membrane oxygenation patients; the solid line denotes patients with cardiogenic shock, and the dashed line denotes patients without cardiogenic shock.

4 OUTCOME AFTER EXTRACORPOREAL MEMBRANE OXYGENATION 949 Table 3. Post-ECMO Variables: Comparison Between Nonsurvivors and Survivors Variable Nonsurvivor (n 149) Survivor (n 84) p Value ARDS, no (%) 36 (24.2) 13 (15.5) 0.12 Dialysis start ICU, no (%) 79 (53.0) 22 (26.2) MSOF, no. (%) 62 (41.6) 13 (15.5) Sepsis syndrome, no (%) 35 (23.5) 13 (15.5) 0.17 Aspiratory pneumonia, no (%) 2 (1.3) 4 (4.8) 0.19 Bacteremia, no (%) 30 (20.1) 19 (22.6) 0.74 Cholecystitis, no (%) 4 (2.7) 4 (4.8) 0.46 Focal deficit, no (%) 7 (4.7) 3 (3.6) 0.99 Fungemia, no (%) 8 (5.4) 12 (14.3) GI bleed, no (%) 10 (6.7) 5 (6.0) 0.99 Global deficit, no (%) 20 (13.4) 3 (3.6) 0.02 Line sepsis, no (%) 7 (4.7) 8 (9.5) 0.17 Mediastinitis, no (%) 2 (1.3) 4 (4.8) 0.19 Pancreatitis, no (%) 5 (3.4) 0 (0) 0.16 Pneumonia, no (%) 32 (21.5) 27 (32.1) Seizure, no (%) 10 (6.7) 5 (6.0) 0.99 Septic shock, no (%) 18 (12.1) 2 (2.4) Urinary, no (%) 7 (4.7) 6 (7.1) 0.55 Abbreviations: ARDS, acute respiratory distress syndrome; ICU, intensive care unit; MSOF, multiple system organ failure; GI bleed, gastrointestinal bleed. medical and surgical patients who required VAE for cardiogenic shock. Venoarterial ECMO provided circulatory support and hemodynamic stability in these patients while cardiac and end-organ function improved. The benefit of VAE in older patients is less clear. Although this investigation showed that mortality increased nearly linearly with age, no upper age limit for VAE support in this setting could be established. Although 2 reports recommend excluding patients who are 75 years and older, 12,13 another investigation by Saito et al 14 reported comparable survival in older and younger ECMO patients. Although it is well established that preoperative malnutrition is associated with worse surgical outcomes, in this investigation the nonsurviving patients had higher preoperative serum albumin levels. 15,16 It is possible that the association between low preoperative albumin and a poor outcome in these patients diminishes over their prolonged postoperative course because malnutrition usually is treated aggressively by enteral or parenteral nutritional support and becomes more of a marker for the severity of illness over time. Operation length and mortality are associated, 17,18 and, thus, it is not surprising that the total CPB time was greater in the nonsurviving group. Fig 2. The percentage of patients having cardiac surgery who required extracorporeal membrane oxygenation. Fig 3. A histogram of age by vital status.

5 950 ELSHARKAWY ET AL Table 4. Year of Surgery and Hospital Deaths Year Total Death (%) (46.43) (47.50) (67.57) (60.87) (85.17) (46.15) (86.96) (75.00) (75.00) (66.67) (66.67) Total Although it is well established that preoperative renal insufficiency increases mortality in patients undergoing cardiac surgery, 19,20 in this investigation baseline renal function was not associated with outcome. A proportion of patients with renal dysfunction on VAE received renal replacement therapy at the same time. In these patients, it is likely that the insult that led to severe cardiac dysfunction and VAE also caused renal dysfunction and thus would not necessarily be associated with preoperative renal function or patient outcome. The proportion of patients requiring VAE after cardiac surgery decreased most strikingly during the investigation period from 1996 to 2000 (Fig 2). Although a cause-andeffect relationship cannot be established by this investigation, it is notable that the fraction of perioperative echocardiography examinations at the authors institution increased significantly during the same period. Intraoperative echocardiography allows for the precise evaluation of cardiac function and guides implementation of less invasive support, which often precludes the need for ECMO. The 2 trends coincide, but causality cannot be established from observational data sets such as these. Similarly, the ability to preclude ECMO support in some patients because of the newer technology available later in the investigation period could account for the yearly increased mortality in that ECMO may have been used only in the very highest acuity patients who had no other options. Early in the investigation, when fewer alternative therapies were available, ECMO likely was used more often in patients who were relatively less critically ill and more likely to survive. Additionally, overall patient acuity increased over the investigation period. This investigation was limited because it was a retrospective review of a single institution s experience. Not all information was available for review in every patient, and follow-up after hospital discharge was not conducted. Hemodynamic parameters at the time of ECMO initiation were not analyzed and most likely affected survival. The investigation was observational and did not allow for clinical control or randomization. Thus, it remains likely that factors other than age and cardiogenic shock played a role in determining mortality. In summary, mortality in patients who required postoperative VAE support was lower in younger patients, patients without diabetes, those with cardiogenic shock, and patients who had shorter CPB times. None of these variables can be modified easily; thus, the present results do not suggest any easy way to reduce the high mortality associated with VAE in patients recovering from cardiac surgery. It is unclear whether these criteria are sufficient to define which patients should be excluded from this potentially life-saving therapy. 1. Morris AH, Wallace CJ, Menlove RL, et al: Randomized clinical trial of pressure-controlled inverse ratio ventilation and extracorporeal CO2 removal for adult respiratory distress syndrome. Am J Respir Crit Care Med 149: , Doll N, Fabricius A, Borger MA, et al: Temporary extracorporeal membrane oxygenation in patients with refractory postoperative cardiogenic shock A single center experience. J Card Surg 18: , Ko WJ, Lin CY, Chen RJ, et al: Extracorporeal membrane oxygenation support for adult postcardiotomy cardiogenic shock. Ann Thorac Surg 73: , Wang SS, Chen YS, Ko WJ, et al: Extracorporeal membrane oxygenation support for postcardiotomy cardiogenic shock. Artif Organs 20: , Zapol WM, Snider MT, Hill JD, et al: Extracorporeal membrane oxygenation in severe acute respiratory failure. A randomized prospective study. JAMA 242: , Magovern GJ, Park SB, Maher TD: Use of a centrifugal pump without anticoagulants for postoperative left ventricular assist. World J Surg 9:25-36, Smedira NG, Moazami N, Golding CM, et al: Clinical experience with 202 adults receiving extracorporeal membrane oxygenation for cardiac failure: survival at five years. J Thorac Cardiovasc Surg 122: , Extracorporeal Life Support Organization (ELSO): ECMO Registry of the Extracorporeal Life Support. Ann Arbor, MI, 2004 REFERENCES 9. Magovern GJ Jr, Magovern JA, Benckart DH, et al: Extracorporeal membrane oxygenation: Preliminary results in patients with postcardiotomy cardiogenic shock. Ann Thorac Surg 57: , Magovern GJ Jr, Simpson KA: Extracorporeal membrane oxygenation for adult cardiac support: The Allegheny experience. Ann Thorac Surg 68: , Combes A, Leprince P, Luyt CE, et al: Outcomes and long-term quality-of-life of patients supported by extracorporeal membrane oxygenation for refractory cardiogenic shock. Crit Care Med 36: , Smith C, Bellomo R, Raman JS, et al: An extracorporeal membrane oxygenation-based approach to cardiogenic shock in an older population. Ann Thorac Surg 71: , Adamson RM, Dembitsky WP, Reichman RT, et al: Mechanical support: Assist or nemesis? J Thorac Cardiovasc Surg 98: , Saito S, Nakatani T, Kobayashi J, et al: Is extracorporeal life support contraindicated in elderly patients? Ann Thorac Surg 83: , Engelman DT, Adams DH, Byrne JG, et al: Impact of body mass index and albumin on morbidity and mortality after cardiac surgery. J Thorac Cardiovasc Surg 118: , Reeves BC, Ascione R, Chamberlain MH, et al: Effect of body mass index on early outcomes in patients undergoing coronary artery bypass surgery. J Am Coll Cardiol 42: , 2003

6 OUTCOME AFTER EXTRACORPOREAL MEMBRANE OXYGENATION Rady MY, Ryan T, Starr NJ: Perioperative determinants of morbidity and mortality in elderly patients undergoing cardiac surgery. Crit Care Med 26: , Holmes JH, Connolly NC, Paull DL, et al: Magnitude of the inflammatory response to cardiopulmonary bypass and its relation to adverse clinical outcomes. Inflamm Res 51: , Abramov D, Tamariz M, Fremes S, et al: Impact of preoperative renal dysfunction on cardiac surgery results. Asian Cardiovasc Thorac Ann 11:42-47, Albert A, Walter J, Hassanein W, et al: The impact of renal dysfunction on early mortality after cardiac surgery: Evaluating the threshold for an unfavorable creatinine clearance and the role of comorbidities. Clin Res Cardiol 2:S22-S28, 2007