Venoarterial (VA) extracorporeal membrane oxygenation (ECMO) has proven to be a useful tool in stabilizing critically ill patients who require cardiopulmonary support.
Given the severity of illness and complicated clinical picture seen in patients supported by ECMO, it is necessary to study predictors of successful weaning and survival. Serum lactic acid level has been identified as one such predictor as elevations are seen in periods of increased metabolic stress and global tissue ischemia.
Lactate is produced via an anaerobic process utilizing pyruvate at the end of glycolysis. Daily production of lactate is approximately 1,500 mmol/L with most being metabolized by the liver and kidney. Normal physiologic levels are accepted at 1 mEq/L. Lactate production occurs when there is an accumulation of pyruvate within the cytosol overwhelming the ability for utilization in the Krebs’s cycle. This can occur during periods of elevated physiologic demand for energy such as periods of prolonged exercise. During physiologic stress and tissue hypoxia, lactate production is increased due to the inability of pyruvate to be used for the aerobic process of mitochondrial oxidative phosphorylation. Lactate is energetically favorable given its utilization to oxidize NADH → NAD+ for further anaerobic glycolysis.1
Serum lactate therefore represents a useful indicator of tissue hypoxia and metabolic stress. Lactate has long been used as a prognostic indicator of treatment success in critically ill patients. Previous studies have shown that regardless of inciting cause down-trending lactate values are positive prognostic indicator. The degree of change in a given time period is variable given the cause but is generally accepted that decreased lactate is representative of improvement in critically ill patients in the intensive care unit (ICU).2
Patients receiving ECMO are critically ill, and serum lactate represents a useful tool in monitoring response to and effectiveness of treatment. In the past, baseline serum lactate was identified as the strongest predictor of outcome in patients on ECMO support.3,4 The data on lactate clearance as a prognostic marker are conflicting.4,5
Our aim was to study absolute lactic acid values at different time points as well as changes from the baseline values during treatment to determine the point of time at which lactate level is most predictive of the outcome.
Methods
Patient Population and Data Collection
This was an observational study of patients who received ECMO support at the University of Kentucky from 2014 to 2018. The Institutional Review Board approved the study and waived the need for patient consent. We retrospectively analyzed clinical and laboratory data of consecutive patients who received VA ECMO support for various cardiac indications. These included variables related to patients’ demographics, comorbidities, indication for ECMO support, ECMO characteristics, laboratory variables, and outcomes. Among the laboratory tests, we collected the lactic acid values before initiating ECMO and then on days 1, 3, 5, and 10 while on ECMO support. At our institution, normal reference range for lactate is 0.5–2.2 mmol/L.
The primary objective of this analysis was to study the predictive utility of various lactic acid parameters and association with outcomes. The two study endpoints were successful weaning from ECMO support and survival to discharge from the hospital.
Statistical Analysis
Primary analysis compared lactic acid levels among patients successfully weaned from ECMO and those not successfully weaned, as well as patients who survived till hospital discharge and those who died during hospitalization. Continuous variables were tested for normality of distribution using the Shapiro–Wilk test. Due to non-normality of distribution of most variables including lactic acid level, we have compared continuous variables by the Mann–Whitney U test and listed them as median and interquartile range (IQR). Kruskal–Wallis test was used to compare continuous variables in more than two groups. Categorical variables were compared with the χ2 test and expressed as counts and percentages. The ability of various lactic acid parameters to predict successful weaning from ECMO and survival to discharge was assessed by calculating the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Comparison of these lactic acid parameters to identify which one is superior in predicting study endpoints was performed using the Hanley and McNeil method as implemented in the MedCalc software. Multivariate logistic regression analysis was used to investigate if lactic acid independently predicted hospital survival. Variables included in the model were those that achieved significance at p < 0.05 level on univariate analysis. A p value of less than 0.05 was considered statistically significant. All statistical significance was assessed using a two-sided p values. Data were analyzed using IBM SPSS 21.0 statistical software (IBM SPSS Version 21.0., Armonk, NY).
Results
Patient Characteristics
A total of 238 patients received ECMO support for various cardiac indications including cardiac arrest 48 (20.2%), myocardial infarction 30 (12.6%), heart failure 50 (21%), pulmonary embolism 38 (16%), right ventricular failure 22 (9.2%), postcardiotomy syndrome 31 (13%), and primary graft failure post heart transplant 14 (5.9%). The mean age of study population was 53.4 ± 14 years. One hundred fifty-two (63.9%) patients received ECMO for left ventricular failure, 72 (30.3%) for right ventricular failure, and 14 (5.9%) for biventricular failure. Seventy-four percent of the patients received peripheral ECMO cannulation, while 26% received central cannulation. The mean duration of ECMO support was 5 ± 6.7 days, and length of hospital stay was 31 ± 45.7 days. Among the 238 patients included in the analysis, 129 (54.2%) were successfully weaned from ECMO and 98 (41.2%) were discharged alive. The baseline characteristics of survivors and nonsurvivors are presented in Table 1.
Table 1. -
Baseline Characteristics of Hospital Survivors and Nonsurvivors Among 238 Patients Receiving ECMO Support
| Variables |
Hospital Survival (n = 98) |
Hospital Death (n = 140) |
p value |
| Demographics and comorbidities |
| Age, years, median (IQR) |
52.2 (39.7–62.4) |
57.1 (49.6–64.0) |
0.019 |
| Male sex, % (n) |
66.3 (65/98) |
72.1 (101/140) |
0.336 |
| Dialysis, % (n) |
15.3 (15/98) |
17.9 (25/140) |
0.604 |
| ECMO indication, % (n) |
| Arrhythmia |
8.2 (8/98) |
28.6 (40/140) |
<0.001 |
| MI |
8.2 (8/98) |
15.7 (22/140) |
0.084 |
| HF |
29.6 (29/98) |
15 (21/140) |
0.007 |
| APE |
21.4 (21/98) |
12.1 (17/140) |
0.054 |
| Post cardiotomy |
9.2 (9/98) |
15.7 (22/140) |
0.141 |
| Cannulation provider, % (n) |
| Cannulation via CTS |
66.3 (65/98) |
72.9 (102/140) |
0.278 |
| Cannulation via cardiologist |
33.7 (33/98) |
27.1 (38/140) |
0.278 |
| Failing ventricle, % (n) |
| LVF |
56.1 (55/98) |
69.3 (97/140) |
0.037 |
| RVF |
35.7 (35/98) |
26.4 (37/140) |
0.125 |
| Biventricular failure |
8.2 (8/98) |
4.3 (6/140) |
0.211 |
| Management post-ECMO, % (n) |
| CTS |
29.8 (28/94) |
23.4 (32/137) |
0.274 |
| Anesthesia |
22.3 (21/94) |
28.5 (39/137) |
0.297 |
| CCU |
47.9 (45/94) |
48.2 (66/137) |
0.964 |
| ECMO characteristics, % (n) |
| Peripheral cannulation |
75.5 (74/98) |
73 (100/137) |
0.664 |
| Central cannulation |
24.5 (24/98) |
27 (37/137) |
0.664 |
| Laboratory variables, median (IQR) |
| Na (meq/L) |
140 (136–146) |
143 (137–149) |
0.067 |
| Creatinine (mg/dl) |
1.5 (1.1–2.4) |
1.7 (1.2–2.4) |
0.246 |
| ALT (IU/L) |
77 (25–217) |
81 (34–387) |
0.362 |
| Bilirubin (mg/dl) |
1.1 (0.6–2.2) |
1.3 (0.6–2.6) |
0.396 |
| Albumin (g/dl) |
2.5 (2.1–2.8) |
2.4 (2.0–2.8) |
0.446 |
| Platelet (per cmm3) |
159 (104–224) |
136 (83–202) |
0.034 |
| Pre-ECMO LA (mmol/L) |
4 (1.8–7.8) |
6.9 (2.5–12.1) |
0.007 |
| LA day 1 (mmol/L) |
1.7 (1.0–4.0) |
4.3 (1.8–8.8) |
<0.001 |
| LA day 3 (mmol/L) |
1.2 (0.8–1.6) |
1.8 (1.1–2.8) |
0.001 |
| LA day 5 (mmol/L) |
0.9 (0.7–1.6) |
1.6 (1.0–2.8) |
0.001 |
| LA day 10 (mmol/L) |
1 (0.7–1.6) |
1.5 (1.1–2.1) |
0.062 |
| Outcomes |
| Successful ECMO weaning, % (n) |
100 (98/98) |
22.1 (31/140) |
<0.001 |
| Length of hospitalization (days), median (IQR) |
33 (19–64) |
8 (3–18) |
<0.001 |
| Days from admission to ECMO (days), median (IQR) |
1.5 (0.6–7.3) |
1 (0.4–6.4) |
0.207 |
| Duration of ECMO (days), median (IQR) |
3 (2–6) |
2 (1–6) |
0.031 |
ALT, alanine transferase; APE, acute pulmonary embolism; CCU, coronary care unit; CTS, cardiothoracic surgery; ECMO, extracorporeal membrane oxygenation; HF, heart failure; IQR, interquartile range; LA, lactic acid; LVF, left ventricular failure; MI, myocardial infarction; RVF, right ventricular failure.
Patient who survived hospitalization were younger (p = 0.019), had ECMO support for cardiac arrhythmias (p < 0.001) or heart failure (p = 0.007), had ECMO support for a failing left ventricle (p = 0.037), and had a higher platelet count (p = 0.034). We found no difference in hospital survival according to which ventricle is failing: left ventricle versus right ventricle versus biventricular failure (p = 0.096). There was no significant difference in frequency of successful ECMO weaning according to the provider performing the cannulation, whether cardiologist or cardiothoracic surgeon (p = 0.474), nor there was a difference in hospital survival (p = 0.278) according to these providers. There was no difference in successful ECMO weaning (p = 0.446) and hospital survival (p = 0.429) according to post-ECMO service whether it was provided by cardiothoracic surgery, anesthesia or coronary care unit. Also, there was no difference in frequency of hemodialysis between hospital survivors and nonsurvivors (15.3% vs. 17.9%, in survivors and nonsurvivors, respectively; p = 0.604). There was a significant difference in pre-ECMO lactic acid level according to underlying etiology whether cardiac arrest (8.1 ± 7.1 mmol/dl) versus myocardial infarction (8.6 ± 7.1 mmol/dl) versus heart failure (5.0 ± 4.6 mmol/dl) versus pulmonary embolism (6.6 ± 5.6 mmol/dl), p across groups 0.038, but there was no difference in day 1 lactic acid (p = 0.141), day 3 lactic acid (p = 0.656), and day 5 lactic acid (p = 0.465). There was no difference in lactic acid according to central versus peripheral cannulation at baseline pre-ECMO (p = 0.953), day 1 (p = 0.208), day 3 (p = 0.481), and day 5 (p = 0.087).
Lactic Acid and Study Outcomes
Lactic acid results were available at 5 time points: pre-ECMO (7.2 ± 6.4 mmol/L), and post ECMO days 1 (4.7 ± 5.0 mmol/L), day 3 (2.2 ± 2.6 mmol/L), day 5 and day 10 (1.9 ± 2.2 mmol/L) for 226, 202, 110, 79 and 26 patients, respectively. Patient successfully weaned from ECMO had a significantly lower lactic acid level pre-ECMO (6.0 vs. 8.6 mmol/L; p = 0.001), at day 1 (2.8 vs. 7.1 mmol/L; p < 0.001), day 3 (1.4 vs. 3.5 mmol/L; p < 0.001), and day 5 (1.3 vs. 2.7 mmol/L; p = 0.001), compared with those who were not successfully weaned (Figure 1). Lactic acid had a significant AUC for predicting successful weaning when checked before ECMO (AUC, 0.630; 95% confidence interval [CI], 0.563–0.693; p = 0.0006), at day 1 (AUC, 0.753; 95% CI, 0.687–0.811; p < 0.0001), and day 3 (AUC, 0.733; 95% CI, 0.641–0.813; p < 0.0001). Patients who were discharged alive had significantly lower lactic acid pre-ECMO (5.8 vs. 8.2 mmol/L; p = 0.007), at day 1 (2.7 vs. 6.2 mmol/L; p < 0.001), day 3 (1.3 vs. 3.0 mmol/L; p = 0.001), and day 5 (1.2 vs. 2.4 mmol/L; p = 0.001), compared with those who died in-hospital (Figure 2). Lactic acid also had a significant AUC for predicting hospital survival when checked before ECMO (AUC, 0.606; 95% CI, 0.539–0.670; p = 0.005), at day 1 (AUC, 0.705; 95% CI, 0.637–0.767; p < 0.0001), and day 3 (AUC, 0.684; 95% CI, 0.589–0.770; p = 0.0003).
;)
Figure 1.: Longitudinal comparison of LA level among patients successfully weaned from ECMO and those who were not weaned. Bars represent the mean and error bars represent standard error of mean. p value is obtained via Mann–Whitney U test. ECMO, extracorporeal membrane oxygenation; LA, lactic acid.
Figure 2.: Longitudinal comparison of LA level among patients who were discharged alive vs. those who died in-hospital. Bars represent the mean and error bars represent standard error of mean. p value is obtained via Mann–Whitney U test. ECMO, extracorporeal membrane oxygenation; LA, lactic acid.
Lactic Acid Change and Study Outcomes
We have investigated the association of magnitude in reduction of lactic acid from pre-ECMO to day 1 and pre-ECMO to day 3 on study outcomes. There was no association between the magnitude of reduction of lactate from pre-ECMO to day 1 on successful ECMO weaning (p = 0.121) and hospital survival (p = 0.127). Similarly, there was no association between the magnitude of reduction of lactate from pre-ECMO to day 3 on successful ECMO weaning (p = 0.418) and hospital survival (p = 0.671).
Comparison of Lactic Acid Parameters and Study Outcomes
Using Hanley and McNeil method to compare areas under the curve, we compared the association of various lactic acid values and study outcomes. With regard to successful ECMO weaning, we noted that day 3 lactic acid is superior to pre-ECMO lactic acid (p = 0.007), lactic acid on the first day of ECMO support, reduction from pre-ECMO to day 1 (p = 0.0125), and from pre-ECMO to day 3 (p = 0.0313). There was a trend that lactic acid at day 1 is superior to pre-ECMO lactic acid (p = 0.0536). The reduction of lactic acid from pre-ECMO to day 1, and pre-ECMO to day 3 is not superior to any absolute value at either pre-ECMO, days 1 or 3 (Figure 3).
Figure 3.: Comparison of area under the curve of various LA parameters in predicting successful ECMO weaning. ECMO, extracorporeal membrane oxygenation; LA, lactic acid.
With regard to hospital survival, lactic acid at ECMO day 3 was superior to pre-ECMO lactic acid (p = 0.0385), lactic acid on day 1, lactic acid reduction from pre-ECMO to day 1 (p = 0.0177) and from pre-ECMO to day 3 (p = 0.0361) (Figure 4). Thus, lactic acid day 3 best-predicted hospital survival and a cutoff value of ≤ 1.7 meq/L had the highest combined sensitivity (87.3%) and specificity (50.9%) in predicting this outcome.
Figure 4.: Comparison of area under the curve of various LA parameters in predicting hospital survival. ECMO, extracorporeal membrane oxygenation; LA, lactic acid.
Multivariable Analysis
A total of 98 (41.2%) patients survived hospitalization. In the multivariate logistic regression model (Nagelkerke R2 = 0.252, p for Hosmer–Lemeshow = 0.298), day 3 lactic acid was an independent predictor of hospital survival after covariate adjustment (odds ratio [OR], 0.505; 95% CI, 0.290–0.880; p = 0.016). Table 2 lists variables included in the multivariate analysis.
Table 2. -
Multivariate Analysis of Predictors of Survival
| Variables |
OR (95% CI) |
p
|
| Age |
0.983 (0.954–1.013) |
0.276 |
| ECMO indication-heart failure |
1.190 (0.819–1.727) |
0.361 |
| Thrombocytopenia |
1.002 (0.997–1.007) |
0.392 |
| ECMO duration |
0.971 (0.907–1.040) |
0.398 |
| Lactic acid day 3 |
0.505 (0.290–0.880) |
0.016 |
CI, confidence interval; ECMO, extracorporeal membrane oxygenation; OR, odds ratio.
Discussion
Our data indicate an association between pre- and post-ECMO lactic acid values and both successful weaning and survival to discharge. We demonstrated that lactate on VA ECMO support is more important indicator of the prognosis than lactate before placing the VA ECMO and on day one on VA ECMO. We studied the magnitude of reduction of lactic acid levels while on ECMO support and found them to be weaker predictors of the outcomes than absolute levels. These data suggest that while pre-ECMO lactic acid can be used as a prognostic indicator, lactic acid several days into treatment may be superior in guiding treatment and assessing response. Lactic acid values on day 5 and day 10 lose their discriminative ability when compared using AUC, likely due to a smaller number of values for these days. It could also be that given our average ECMO duration of 5 ± 6.7 days, those who remained on ECMO until day 10 were unlikely to survive to decannulation. We showed, for the first time, that the pre-ECMO lactate differs by etiology of cardiogenic shock, with heart failure and pulmonary embolism demonstrating the lowest lactate values.
While there is an agreement that lactate level before the ECMO support is a predictor of mortality, data on lactic acid levels on support are sparse and conflicting. Some studies report pre-ECMO levels only6–9 or the levels soon after ECMO initiation10 Both baseline lactate and the level after 6 hours of support were found to be associated with poor outcomes by Guenther et al.11 Per Slottosch et al.12 lactate before ECMO and peak lactate level during ECMO support showed no significant connection to mortality, while lactate and lactate clearance at 24 hours were predictive for 30 day mortality. On the other hand, Mungan et al.5 found that changes in lactate levels after ECMO implantation are superior to single lactate measurements as a prognostic sign of mortality.
Very few studies addressed the question of comparative significance of lactate at different times of ECMO support. Formica et al.13 demonstrated that blood lactate levels 48 and 72 hours after the initiation of ECMO predicted outcomes, while the pre-ECMO lactate did not. However, they did not compare the prognostic value of absolute levels versus the magnitude of lactate level change on support. Similarly, Loforte et al.14 found the blood lactate level 72 hours after ECMO initiation can help identify survivors and nonsurvivors on ECMO, but the changes in absolute levels were not included into the analysis.
The dynamic changes of lactate within 6 hours and 12 hours after the beginning of the ECMO support were reported by Li et al.15 The mean lactate concentration and especially early lactate clearance 12 hours after the initiation of ECMO provided best prognostic guidance. The authors did not include later time points.
Our study, therefore, represents the most complete analysis of lactate at different time points on VA ECMO support in comparison with lactate clearance. Our limitations include collecting retrospective data from a single center. Further study is needed regarding outcome predictors for ECMO use with larger study populations.
In practice, there is so much emphasis, including the attempts to make a clinical prognosis, based on a pre-ECMO lactate, that we felt it was interesting when we found out that the lactate well into ECMO support was in fact more meaningful for prognosis. As patients on ECMO are usually also on the ventilator/sedation, with unclear neurologic status, and unable to provide any subjective input, we think that any extra piece of information indicating the prognosis is useful from a clinical standpoint.
Conclusions
For patients with cardiogenic shock on VA ECMO support, absolute levels of lactic acid, and not lactate clearance, remain a most useful tool in assessing response to ECMO therapy and predicting outcomes. Lactic acid as a prognostic indicator becomes more valuable while on ECMO support that before placement on ECMO. Furthermore, day 3 lactate was superior to the value obtained before ECMO or on the first day of support.
References
1. Levy B: Lactate and shock state: The metabolic view. Curr Opin Crit Care. 12: 315–321, 2006.
2. Vincent JL, Quintairos E Silva A, Couto L Jr, Taccone FS: The value of blood lactate kinetics in critically ill patients: A systematic review. Crit Care. 20: 257, 2016.
3. Fux T, Holm M, Corbascio M, Lund LH, van der Linden J: Venoarterial extracorporeal membrane oxygenation for postcardiotomy shock: Risk factors for mortality. J Thorac Cardiovasc Surg. 156: 1894–1902.e3, 2018.
4. Jung C, Janssen K, Kaluza M, et al.: Outcome predictors in cardiopulmonary resuscitation facilitated by extracorporeal membrane oxygenation. Clin Res Cardiol. 105: 196–205, 2016.
5. Mungan İ, Kazanci D, Bektaş Ş, Ademoglu D, Turan S: Does lactate clearance prognosticates outcomes in ECMO therapy: A retrospective observational study. BMC Anesthesiol. 18: 152, 2018.
6. Blumenstein J, Leick J, Liebetrau C, et al.: Extracorporeal life support in cardiovascular patients with observed refractory in-hospital cardiac arrest is associated with favourable short and long-term outcomes: A propensity-matched analysis. Eur Heart J Acute Cardiovasc Care. 5: 13–22, 2016.
7. Burrell AJ, Pellegrino VA, Wolfe R, et al.: Long-term survival of adults with cardiogenic shock after venoarterial extracorporeal membrane oxygenation. J Crit Care. 30: 949–956, 2015.
8. Dennis M, McCanny P, D’Souza M, et al.; Sydney ECMO Research Interest Group: Extracorporeal cardiopulmonary resuscitation for refractory cardiac arrest: A multicentre experience. Int J Cardiol. 231: 131–136, 2017.
9. Flécher E, Anselmi A, Corbineau H, et al.: Current aspects of extracorporeal membrane oxygenation in a tertiary referral centre: Determinants of survival at follow-up. Eur J Cardiothorac Surg. 46: 665–671; discussion 671, 2014.
10. Carroll BJ, Shah RV, Murthy V, et al.: Clinical Features and outcomes in adults with cardiogenic shock supported by extracorporeal membrane oxygenation. Am J Cardiol. 116: 1624–1630, 2015.
11. Guenther SP, Brunner S, Born F, et al.: When all else fails: Extracorporeal life support in therapy-refractory cardiogenic shock. Eur J Cardiothorac Surg. 49: 802–809, 2016.
12. Slottosch I, Liakopoulos O, Kuhn E, et al.: Lactate and lactate clearance as valuable tool to evaluate ECMO therapy in cardiogenic shock. J Crit Care. 42: 35–41, 2017.
13. Formica F, Avalli L, Colagrande L, et al.: Extracorporeal membrane oxygenation to support adult patients with cardiac failure: Predictive factors of 30-day mortality. Interact Cardiovasc Thorac Surg. 10: 721–726, 2010.
14. Loforte A, Marinelli G, Musumeci F, et al.: Extracorporeal membrane oxygenation support in refractory cardiogenic shock: Treatment strategies and analysis of risk factors. Artif Organs. 38: E129–E141, 2014.
15. Li CL, Wang H, Jia M, Ma N, Meng X, Hou XT: The early dynamic behavior of lactate is linked to mortality in postcardiotomy patients with extracorporeal membrane oxygenation support: A retrospective observational study. J Thorac Cardiovasc Surg. 149: 1445–1450, 2015.
