Critically ill patients supported by extracorporeal membrane oxygenation (ECMO) may have significant fluid overload (FO) before or during their treatment.1 Illnesses resulting in acute severe respiratory failure can require large volumes of intravenous fluid administration during the initial resuscitation before commencing ECMO. Once extracorporeal support has been established, frequent blood transfusion and fluid loading may be required to maintain adequate ECMO blood flow. Fluid overload results in noncardiogenic pulmonary edema, exacerbating lung or other organ injury, and is associated with increased mortality in patients with acute respiratory distress syndrome.2–5
In critically ill patients requiring ECMO support, acute kidney injury (AKI) and FO have been shown to correlate with higher mortality.6–8 A recent study in adult patients supported by ECMO showed that a positive fluid balance at ECMO day 3 was an independent predictor of 90 day mortality.1 The Extracorporeal Life Support Organization (ELSO) suggests the goal of fluid management in patients supported by ECMO is to return the extracellular fluid volume to normal (dry weight) and maintain it there. The ELSO guidelines recommend diuretic treatment as an initial option for fluid removal, and for those with a poor response to diuretics or those with overt renal failure, continuous renal replacement therapy (CRRT) is recommended.9
There is no evidence to suggest that one technique of fluid removal is superior to others. A multicenter international survey showed that management of FO in patients supported by ECMO was highly variable10 and that using CRRT at an early stage in ECMO treatment to prevent or treat FO is a relatively common practice. The use of CRRT may allow the administration of necessary therapies such as nutrition, medications, and blood products in addition to treating any coexisting AKI.11–13
Despite the common practice of fluid removal in patients supported by ECMO, no guidance currently exists on the optimal timing of fluid removal, the rate of fluid removal, or the total amount of fluid to remove. Furthermore, it is not known how to monitor the clinical consequences of fluid removal during ECMO and how continued fluid removal might affect native lung function or patient outcome. We sought to determine whether early fluid removal during combined CRRT and venovenous extracorporeal membrane oxygenation (VV-ECMO) treatment in adult patients is associated with improved native lung function and survival.
Materials and Methods
Our study was conducted in an 18-bed, medical surgical intensive care unit (ICU) of an inner city, university-affiliated teaching hospital. The 570-bed hospital is a tertiary referral center for cardiothoracic surgery, including heart and lung transplantation and is also the national referral center for adult extracorporeal life support (ECLS). The ECLS service is run primarily by the critical care medical and nursing team, with additional input from the cardiothoracic surgery and perfusion services. The ICU operates a national ECMO retrieval service, transferring patients on ECMO from other centers when required. Ethical approval for the study was sought and approved by the local ethical committee.
Our ICU ECMO database was accessed to identify all adult patients from 2010 to 2015 who received VV-ECMO for acute severe respiratory failure. Patients who were supported by VV-ECMO and who had concomitant CRRT for >24 hours during the ECMO run were included. Patients requiring massive transfusion of blood products (>5 units of red blood cells in 24 hours), requiring major surgery during the course of their ECMO run, or not surviving beyond 72 hours of ECMO run were excluded. Those treated with ECMO in a different institution before transfer were also excluded, because of insufficient data being available for analysis.
Description of Extracorporeal Membrane Oxygenation and Mechanical Ventilator Settings
The decision to commence ECMO was made on a case-by-case basis by the treating intensivist according to established guidelines.9
Following an unfractionated heparin bolus of 5,000 international units, percutaneous venovenous cannulation was performed using Seldinger technique with ultrasound guidance. Patients were cannulated using either bifemoral, femoral–jugular, or internal jugular (bicaval dual lumen catheter) approach. Cannula positioning was confirmed using echocardiography or fluoroscopy. A centrifugal pump integrated with a polymethylpentene hollow fiber oxygenator (Cardiohelp, Maquet, Germany) was used with a heparin-coated ECMO circuit.
In the absence of any contraindication to anticoagulation, an infusion of unfractionated heparin was subsequently commenced to achieve a target activated clotting time between 180 and 220 seconds. For each patient, ECMO blood flow was adjusted to achieve arterial oxygenation saturation >85% with the least negative ECMO access pressure (usually about −50 mm Hg) and to maintain systemic lactate <2 mmol/L. Sweep gas flow rate to the oxygenator (using 100% O2) was adjusted according to arterial pH and carbon dioxide tension.
After starting ECMO, neuromuscular blockade was ceased and patients remained lightly sedated. Mechanical ventilator settings were reduced to rest settings (FiO2: 0.5; positive end-expiratory pressure [PEEP]: 10–15 cm H2O; respiratory rate: 12–16 breaths/minute, and pressure support: 10–12 cm H2O above PEEP) and remained unchanged throughout the period of ECMO support.
Hemodynamic and Fluid Management
Fluid administration was decided by the treating intensivist. During the study period, crystalloid (0.9% saline) solution was the standard maintenance fluid used, and fluid boluses were given using colloid solution (Gelofusine). Norepinephrine infusion was titrated to achieve mean arterial pressure of >65 mm Hg, with vasopressin used as second-line vasopressor agent when required.
Description of Extracorporeal Membrane Oxygenation–Continuous Renal Replacement Therapy Setup and Procedure
We connected the CRRT circuit (Prismaflex; Gambro Ltd, Peterborough, UK) directly to the ECMO circuit. The dialysis circuit was primed according to standard techniques, and the CRRT access line connected to the postoxygenator side of the ECMO circuit via a high-flow three-way tap. The CRRT return line was then connected to the preoxygenator side of the ECMO circuit, also via a high-flow three-way tap. Both CRRT connections were to the positive pressure side (post-blood pump) of the ECMO circuit, and a small volume of recirculation was expected.
Dialysis treatments were performed as continuous venovenous hemodiafiltration and ultrafiltration rates adjusted according to fluid removal goals as directed by the treating intensivist. Our standard practice was to commence CRRT soon after ECMO initiation in all patients with the aim of preventing or treating FO at an early stage, targeting return to patient “dry” weight as guided by a combination of clinical and hemodynamic parameters.
The ICU computerized clinical information system (ICIP; Philips, Eindhoven, the Netherlands) was accessed to gather standard patient demographics, ECMO and CRRT data, fluid balance, and details of blood and fluid administration. Indications for CRRT included FO, AKI, or a combination of both. Fluid overload was defined as the presence of positive fluid balance at initiation of ECMO support or as diagnosed clinically by the presence of tissue edema on clinical examination. Acute kidney injury was defined according to Kidney Disease Improving Global Outcomes (KIDGO) 2012 criteria.14
Daily and cumulative fluid balance was recorded for each patient up to the conclusion of the ECMO run or to a maximum of 14 ECMO days. Data for the first 14 days was included in this study because the median duration of ECMO support was 12.5 (7.8–19.3) days. It was also assumed that changes in fluid balance would have a more obvious impact in patients with shorter ECMO runs.
Daily fluid balance was calculated as the difference between all fluids in and out per day, and cumulative fluid balance was the sum of daily fluid balances for all ECMO days. Recovery of native lung function during each ECMO day was monitored by the changes in tidal volume and dynamic pulmonary compliance (defined as tidal volume in milliliters/[inspiratory pressure support − PEEP] in cmH2O). Mean daily compliance was calculated by averaging the hourly pulmonary compliance over each 24 hour period when the patient was supported by ECMO and CRRT. Native lung recovery was also monitored by changes in arterial partial pressure of oxygen (PaO2) and the PaO2 to mechanical ventilator FiO2 ratio.
Primary outcome studied was survival to hospital discharge. Secondary outcome measures included 28 day survival, recovery of native lung function, and need for renal replacement therapy (RRT) beyond hospital discharge.
Continuous variables were compared between survivors and nonsurvivors with Wilcoxon rank-sum tests given likely non-normal distributions, and categorical variables were compared with Fisher exact test. Statistical significance was determined at a p value of <0.05, with no adjustment for multiple comparisons. Logistic regression was performed to examine a relationship between each variable and survival to hospital discharge. Multivariate regression was then performed to adjust for measured covariates if considered likely to influence outcome or if p < 0.2 on univariate analysis. This was performed with a Bayesian logistic regression with a weakly informative Cauchy prior.
Finally, using the full dataset of daily measurements of fluid balance and compliance in each patient, a generalized estimating equations (GEE) approach with an “independence” working correlation structure was used to model the effect of daily fluid balance on compliance. To explore the temporal nature of this relationship, models were fitted with different “lags” between fluid balance and compliance that day (0 lag), 1 day, and 2 days later. There was no imputation for missing data at any stage of analysis, with case-complete data only used. Statistical analysis was performed using R Statistical Software 3.3.2 (Foundation for Statistical Computing, Vienna, Austria, 2016) and the arm, glmulti, and geepack libraries.
Thirty-one adult patients received VV-ECMO (and CRRT) for severe respiratory failure during the study period. Seven patients were excluded: two patients required major surgery for perforated colon and peritonitis during the course of ECMO, one patient had massive bleeding after single lung transplantation, one patient received ECMO in a different institution for 1 week before being transferred to our hospital, one patient who did not survive to 72 hours post initiation of ECMO support, and two patients who did not receive CRRT. The remaining 24 patients (total of 395 ECMO days) were included in the study.
Seventeen patients (71%) survived to 28 days, with 15 patients (63%) surviving to hospital discharge, with all achieving recovery of renal function. Baseline clinical and treatment-related variables are compared between survivors and nonsurvivors (see Table, Supplemental Digital Content, http://links.lww.com/ASAIO/A311). There were no statistically significant differences in distribution of diagnoses between survivors and nonsurvivors (Table 1) and CRRT treatment (Table 2). Survivors were found to be significantly younger (median age: 39 vs. 57 years; P = 0.010), have a higher noradrenaline dose at ECMO initiation (0.22 vs. 0.06 µg/kg/minute; P = 0.037), require a lower number of red blood cell transfusions, and have shorter ECMO duration.
Univariate logistic regression did not show a significant association between either outcome (in-hospital survival or 28 day survival) and mean fluid balance or cumulative fluid balance at day 3, day 7, or overall. Multivariate regression identified patient age, sex, noradrenaline dose at ECMO initiation, diagnosis, and CRRT before ECMO as potentially significant covariates. After stepwise model selection, age and noradrenaline dose remained, and with fluid balance, these generated a final model for in-hospital survival in Table 3. For each liter of cumulative positive fluid balance, there was an odds ratio of 0.80 (95% credible interval: 0.63–1.04) but with a p value that did not reach significance (P = 0.097).
Finally, in GEE modeling of daily data, changes in fluid balance were examined as predictors of pulmonary compliance. Two hundred sixty-three ECMO days were analyzed with no missing values. “Lags” of the daily fluid balance were used to examine for a delayed effect of cumulative fluid balance on compliance 1 or 2 days later (Figure 1), but these were not significant. Cumulative fluid balance and mean daily balance were found to be associated with indices of native lung recovery (Table 4). Increased daily fluid balance was more strongly associated with decreased compliance on that day, with an approximate −4.37 change in compliance for each liter increase in cumulative fluid balance (95% CI: −6.13 to −2.62; P < 0.001).
Plots of fluid balance and compliance for each patient showed decreasing median cumulative fluid balance with time in survivors compared with nonsurvivors (Figure 2). Although compliance improved in some nonsurvivors, it then deteriorated again, in contrast to a group of survivors who had rapidly negative cumulative fluid balance and improving compliance.
In adult patients supported by VV-ECMO and CRRT, we observed that both negative cumulative daily fluid balance and more negative mean daily fluid balance were strongly associated with an improvement in pulmonary compliance in our patient group. We also observed that those who achieved a cumulative negative fluid balance at day 3, day 7, and day 14 were more likely to survive to hospital discharge, although the effect did not reach statistical significance.
Comparison With Previous Studies
In critically ill adult patients with acute respiratory failure, use of a conservative fluid management strategy has previously been shown to improve lung function and shorten duration of mechanical ventilation.5 Fluid overload is common in those patients requiring VV-ECMO, and removing excess fluid has become a common practice in many centers. A recent study of adult patients supported by ECMO found those with a more negative fluid balance on ECMO day 3 had improved 90 day mortality.1 They included both patients with severe heart or lung failure supported with ECMO, the majority (60%) received RRT, and fluid removal was achieved using both loop diuretic therapy and CRRT. The increased recognition of the presence of a positive fluid balance was reflected by the same authors who noted that over the duration of their study (2006–2013), there was a progressive decrease in the positive fluid balance recorded during the first 3 ECMO days. Compared with the study by Schmidt et al., which included both patients supported by venoarterial and VV-ECMO, our study concentrated on a cohort of patients with respiratory failure requiring VV-ECMO, all of whom received fluid management using CRRT. Although our data did not reveal a significant mortality benefit, more negative cumulative balance was associated with a trend toward improved survival to hospital discharge. Interestingly, survivors presented with a higher dose of noradrenaline requirements before initiation of ECMO. The reasons for this are not clear and may warrant further exploration in future studies involving a larger cohort of patients.
It has become standard practice in our ICU to initiate CRRT for fluid removal and prevention of FO, as early as tolerated for all patients supported by VV-ECMO. However, international practices for fluid removal in patients supported by ECMO are variable. The ELSO recommends fluid removal by spontaneous diuresis, administration of loop diuretics, or using RRT9 but does not offer specific guidance on how, when, and for how long to use CRRT. A multinational survey conducted by Fleming et al.10 highlighted a marked variation in clinical practice on how centers deal with FO during ECMO. Of the centers responding, 23% reported not using any RRT during ECMO. Of those centers using RRT, only 51% reported using a CRRT machine in-line with the ECMO circuit, and 59% reported treatment or prevention of FO as the major indication for RRT. The majority of the responding centers (85%) were located in North America with data on CRRT practice in European centers remaining limited.
Literature to date on fluid removal with CRRT during ECMO has been limited to single-center studies and for the most part in pediatric patients.7,15–17 Use of CRRT has been shown to result in improved fluid balance and caloric intake with less diuretic use in pediatric patients requiring ECMO for respiratory failure.7
Furthermore, concerns around precipitating chronic renal failure when CRRT is used for fluid removal in patients supported by ECMO have not been substantiated.8 This is supported by our data, with all of the survivors having renal recovery at hospital discharge.
Effect of Fluid Balance on Pulmonary Compliance
The strong association of cumulative negative fluid balance with improved pulmonary compliance highlights the potential beneficial effect of fluid removal in this patient group. Significantly, the temporal correlation between cumulative fluid removal and compliance that day suggests a mechanistic link between reversal of pulmonary edema and lung recovery. The fact that some patients did not survive despite effective fluid removal likely reflects the complex patient- and disease-related factors that come in to play.
Kelly et al.18 found that in neonatal patients supported by ECMO for respiratory failure, total body water and mobilization of edema were important determinants of native lung recovery and ECMO duration. Native lung recovery was assessed by changes in appearance of pulmonary edema observed in plain chest radiography. To our knowledge, the link between fluid removal and native lung recovery has not previously been studied in adult patients supported by ECMO. Pulmonary compliance reflects just one aspect of native lung function and may be seen as a relatively crude measurement of recovering lung function. Other indices (such as oxygenation ability, as estimated by P/F ratio, and diffusion capabilities of the lungs) may also provide important indications of native lung performance. However, in patients supported by ECMO, measurements of native lung oxygenation function can be particularly difficult to capture accurately, with the effects of ventilator settings, FiO2, ECMO blood flow rate, plasma Hb level, and cardiac output all potentially affecting the patient’s arterial oxygen saturation.19 For these reasons, we felt that pulmonary compliance would provide a useful index to capture trends in native pulmonary function related to FO.
Strengths and Limitations
To the best of our knowledge, this is the first study to explore the relationship between fluid removal and outcome in adult patients receiving VV-ECMO and concomitant CRRT. In addition to supporting the safety and efficacy of fluid removal using in-line CRRT, our findings highlight the potential benefits of fluid removal in terms of survival and recovery of native pulmonary function.
Our study has some important limitations. It suffers from the inherent limitations of a retrospective observational design, and generalizability of the findings is limited by the single-center setting. Our sample size was relatively small at just 24 patients; however, this was reflective of the activity in our center over a 5 year period and included a significant total number of ECMO days, with 263 daily data points included in the analysis of compliance as a function of fluid removal. Differences in pre-ECMO fluid management may also have influenced our findings. Finally, the surrogate measures of native lung function we measured all have their shortcomings. However, we felt that these were practical indicators that can be assessed on a daily basis for all patients supported by VV-ECMO.
In summary, our observational study suggests that in adult patients requiring VV-ECMO for severe respiratory failure, early fluid removal using CRRT is associated with a trend toward improved survival. The positive effects of fluid removal in survivors is reflected by a stepwise improvement in native pulmonary compliance with fluid removal, and observing these changes may be helpful in tracking potential improvement in native lung function. Further work is required to define the potential impact of fluid removal using CRRT in adult patients supported by VV-ECMO.
1. Schmidt M, Bailey M, Kelly J, et al. Impact of fluid balance
on outcome of adult patients treated with extracorporeal membrane oxygenation
. Intensive Care Med 2014.40: 1256–1266.
2. Murphy CV, Schramm GE, Doherty JA, et al. The importance of fluid management in acute lung injury secondary to septic shock. Chest 2009.136: 102–109.
3. Rosenberg AL, Dechert RE, Park PK, Bartlett RH. Review of a large clinical series: association of cumulative fluid balance
on outcome in acute lung injury: a retrospective review of the ARDSnet tidal volume study cohort. J Intensive Care Med 2009.24: 35–46.
4. Simmons RS, Berdine GG, Seidenfeld JJ, et al. Fluid balance
and the adult respiratory distress syndrome. Am Rev Respir Dis 1987.135: 924–929.
5. Wiedemann HP, Wheeler AP, Bernard GR, et al. Comparison of two fluid-management strategies in acute lung injury. N Eng J Med 2006.354: 2564–2575.
6. Askenazi DJ, Selewski DT, Paden ML, et al. Renal replacement therapy in critically ill patients receiving extracorporeal membrane oxygenation
. Clin J Am Soc Nephrol 2012.7: 1328–1336.
7. Hoover NG, Heard M, Reid C, et al. Enhanced fluid management with continuous venovenous hemofiltration in pediatric respiratory failure patients receiving extracorporeal membrane oxygenation
support. Intensive Care Med 2008.34: 2241–2247.
8. Paden ML, Warshaw BL, Heard ML, Fortenberry JD. Recovery of renal function and survival after continuous renal replacement therapy
during extracorporeal membrane oxygenation
. Pediatr Crit Care Med 2011.12: 153–158.
9. Extracorporeal Life Support Organization: ELSO Guidelines for Cardiopulmonary Extracorporeal Life Support Extracorporeal Life Support Organization, Guidelines for Adult Respiratory Failure, Version 1.4. 2017. Ann Arbor, MI, Extracorporeal Life Support Organization, https://www.elso.org
10. Fleming GM, Askenazi DJ, Bridges BC, et al. A multicenter international survey of renal supportive therapy during ECMO: the Kidney Intervention During Extracorporeal Membrane Oxygenation
(KIDMO) group. ASAIO J 2012.58: 407–414.
11. Antonucci E, Lamanna I, Fagnoul D, Vincent JL, De Backer D, Silvio Taccone F. The impact of renal failure and renal replacement therapy on outcome during extracorporeal membrane oxygenation
therapy. Artif Organs 2016.40: 746–754.
12. Chen H, Yu RG, Yin NN, Zhou JX. Combination of extracorporeal membrane oxygenation
and continuous renal replacement therapy
in critically ill patients: a systematic review. Crit Care 2014.18: 675.
13. Seczyńska B, Królikowski W, Nowak I, Jankowski M, Szułdrzyński K, Szczeklik W. Continuous renal replacement therapy
during extracorporeal membrane oxygenation
in patients treated in medical intensive care unit: technical considerations. Ther Apher Dial 2014.18: 523–534.
14. Khwaja A. KDIGO clinical practice guidelines for acute kidney injury. Nephron Clin Pract 2012.120: c179–c184.
15. Cavagnaro F, Kattan J, Godoy L, et al. Continuous renal replacement therapy
in neonates and young infants during extracorporeal membrane oxygenation
. Int J Artif Organs 2007.30: 220–226.
16. Santiago MJ, Sánchez A, López-Herce J, et al. The use of continuous renal replacement therapy
in series with extracorporeal membrane oxygenation
. Kidney Int 2009.76: 1289–1292.
17. Selewski DT, Cornell TT, Blatt NB, et al. Fluid overload and fluid removal in pediatric patients on extracorporeal membrane oxygenation
requiring continuous renal replacement therapy
. Crit Care Med 2012.40: 2694–2699.
18. Kelly RE Jr, Phillips JD, Foglia RP, et al. Pulmonary edema and fluid mobilization as determinants of the duration of ECMO support. J Pediatr Surg 1991.26: 1016–1022.
19. Schmidt M, Tachon G, Devilliers C, et al. Blood oxygenation and decarboxylation determinants during venovenous ECMO for respiratory failure in adults. Intensive Care Med 2013.39: 838–846.