Abstract
Patients undergoing extracorporeal membrane oxygenation (ECMO) support frequently develop renal failure requiring renal replacement therapy (RRT). RRT may be performed via a dialysis catheter based approach or via the ECMO circuit. We describe our experience with both techniques. A total of 68 patients undergoing ECMO support at our institution were retrospectively analyzed. Predictors of renal failure requiring RRT were determined. Patients undergoing RRT via a dialysis catheter were compared with those undergoing RRT via the ECMO circuit. 10 of the 68 patients required RRT support prior to ECMO. Of the remaining 58 patients, 25 (43%) required new RRT support on ECMO. Lower albumin levels and postcardiotomy shock were predictive of new renal failure requiring RRT on ECMO. RRT performed via the ECMO circuit demonstrated similar efficacy as via a dialysis catheter. Outcomes were much worse for patients requiring new RRT on ECMO support, with a doubling of the length of ECMO support and less that one-third the survival rate of patients not requiring RRT on ECMO support. New renal failure requiring RRT occurs in nearly one-half of patients on ECMO support, with poor outcomes. RRT may be performed via the ECMO circuit with similar efficacy as via a dialysis catheter.
Keywords
Introduction
In patients undergoing venoarterial extracorporeal membrane oxygenation (VA ECMO) support, renal failure is a frequent consequence of the preexisting shock state.1–5 Renal replacement therapy (RRT) is frequently indicated and may be performed via a dialysis catheter, although this may be challenging with limited access sites due to ECMO cannulae and the risks of catheter placement in the setting of anticoagulation. In such scenarios, the ECMO circuit may serve as an RRT access option. Various strategies for connecting an RRT hemofilter or circuit to an ECMO circuit have been described, each with their potential benefits and drawbacks.5–13 We describe our experience with RRT in patients on VA ECMO support with the goal of assessing predictors of renal failure requiring new RRT while on ECMO in patients not previously requiring RRT prior to institution of ECMO support, and to compare the adequacy of RRT when performed via a standard dialysis catheter approach versus via the ECMO circuit.
Methods
Data collection and plan
This was a retrospective cohort study of all patients undergoing VA ECMO placement at our institution from 2010 to 2019. Patients underwent VA ECMO for cardiogenic shock or cardiac arrest in the setting of postcardiotomy shock, acute myocardial infarction, nonischemic cardiomyopathy, or rarely pulmonary or other etiologies. Data regarding pre-ECMO clinical parameters, cannulation strategies, and outcomes including renal failure as defined by the requirement of RRT, weaning and survival were collected. Pre-ECMO hemodynamic and laboratory variables collected were the last value available within 24 h of ECMO placement. A creatinine level of 4 mg/dL was assigned to any patient on RRT at the time of ECMO cannulation. Indications for RRT support on ECMO included refractory volume overload as determined by hemodynamics on right heart catheterization, pulmonary edema on chest radiograph, and physical exam findings of edema not responsive to diuretics, refractory metabolic acidosis, hyperkalemia, and uremia. Continuous venovenous hemodialysis was the type of RRT performed on VA ECMO patients. This has been consistent over the 9 years of the study as were the physicians involved (surgeons, intensivists, nephrologists). Details of RRT support including access strategy, access, effluent, and return line pressures, RRT and ECMO blood flow rates, RRT efficacy, efficiency, and circuit patency were collected. The study was approved by the New York Methodist Hospital Institutional Review Board (Board Ref # 1382246; date of approval 2/5/2019).
Inclusion and exclusion criteria
All patients over 18 years of age undergoing VA ECMO institution at our institution from 2010 to 2019 were included. Pediatric patients were excluded.
Treatment delivered
The VA ECMO circuit consisted of either the Biomedicus (Medtronic, Minneapolis, MN) or the Centrimag (Levitronix, Waltham, MA) pump depending on the time period and a Quadrox D oxygenator (Maquet, Wayne, NJ). Femoral, axillary, and central aortic arterial cannulation and femoral, internal jugular, and central right atrial venous cannulation strategies were utilized depending the clinical scenario and surgeon preference. Patients were heparinized to an activated clotting time of 250 s for cannulation and maintained at a partial thromboplastin time of 45–60–s during the course of VA ECMO support unless contraindicated. Management of ECMO patients with the exception of RRT access (see below) has been consistent over the 9 years of the study as were the physicians involved (surgeons, intensivists, nephrologists).
Prior to 2014, RRT on ECMO was performed via a dialysis catheter. Starting in 2014, a protocol was established to perform RRT via the ECMO circuit. After this time, RRT was performed utilizing a dialysis catheter or via the ECMO circuit at the discretion of the treating physicians. RRT was performed via the NxStage continuous venovenous hemofiltration circuit (NxStage Medical, Inc., Lawrence MA). When performed via the ECMO circuit, both access and return lines were connected to the ECMO circuit post-pump and pre-oxygenator as shown in Figure 1. Due to constant alarms on the CVVH circuit from high access line pressures (caused by post-pump access), the access line pressure monitor is turned off when RRT is performed via the ECMO circuit. An RRT run was defined as a period of continuous RRT in a given patient via one of the two abovementioned techniques. Ultrafiltration efficiency was defined as the amount of fluid actually removed relative to the amount set to be removed. RRT circuit patency was defined as the number of RRT circuit cartridge changes per run. RRT efficacy was defined as the percent decrease in creatinine levels. Comparisons of the adequacy of RRT was were based on the ultrafiltration efficiency and efficacy.
Renal replacement therapy circuit spliced into extracorporeal membrane oxygenation circuit post-pump/preoxygenator.
Primary response variables
The first goal was to assess predictors of renal failure requiring new RRT while on ECMO in patients not previously requiring RRT prior to institution of ECMO support. RRT was performed for the indications mentioned above, including volume overload, metabolic acidosis, hyperkalemia, and uremia. Patients not on RRT prior to ECMO support were categorized into two groups based on whether RRT after ECMO was instituted. Pre-ECMO clinical parameters were assessed for their ability to predict need for RRT after ECMO was instituted.
The second goal was to compare the adequacy of RRT when performed via a standard dialysis catheter approach versus via the ECMO circuit
Statistical analysis
Continuous variables were described as the mean and standard deviation and compared with the students t-test. Categorical variables were described as frequency and percentages and compared with the chi squared test or fisher exact test. Multivariable logistic regression analysis was used to determine independent of renal failure requiring new RRT while on ECMO in patients not previously requiring RRT prior to institution of ECMO support. All variables with a p-value of less than .05 on univariable analysis were entered into the multivariable model. A p value of .05 or less was considered significant. Analyses were performed with Stata 13 (StataCorp LLC College Station, TX).
Results
From 2010 to 2019 68 patients were supported with VA ECMO at our institution. 10 of these were already on RRT support prior to institution of ECMO (three chronic dialysis patients, and seven patients with acute renal failure prior to institution of ECMO support). Of the remaining 58 patients, 25 (43%) required new RRT support on ECMO (Figure 2). These 25 patients were compared to the 33 patients not requiring RRT on ECMO support to assess for predictors of renal failure requiring RRT on ECMO.
Breakdown of patients.
Baseline characteristics are listed in Table 1, with a variety of clinical paramaters associated with renal failure requiring new RRT on ECMO support. These factors were selected for analysis as they are either markers of pre-existing chronic comorbidities (age, albumin, and platelet count) or markers of the degree of shock and time in shock (extracorporeal cardiopulmonary resuscitation [eCPR], time on ventilator, and laboratory parameters of end organ function). On multivariable analysis, lower albumin levels and postcardiotomy shock were predictors of renal failure requiring new RRT on ECMO support (Table 2). Outcomes were predictably worse in patients requiring new RRT on ECMO support, with significantly longer duration of ECMO support, decreased rates of weaning from ECMO support, and higher mortality (Table 3).
BLE 1: Baseline characteristics.
eCPR: extracorporeal cardiopulmonary resuscitation.
Independent predictors of the need for new RRT on ECMO support.
RRT: renal replacement therapy; ECMO: extracorporeal membrane oxygenation; eCPR: extracorporeal cardiopulmonary resuscitation.
Outcomes.
RRT: renal replacement therapy.
Prior to 2014, RRT on ECMO was performed via a dialysis catheter as in non-ECMO patients at our institution. Starting in 2014, a protocol was established to perform RRT via the ECMO circuit. After this time, RRT was performed via a dialysis catheter or via the ECMO circuit at the discretion of the treating physicians.
Thirty-five patients underwent RRT on ECMO support (25 requiring new RRT on ECMO plus ten patients requiring RRT prior to ECMO support). Twenty-two runs of RRT (total of 128.6 h) were performed on ECMO support via a dialysis catheter. Twenty-two runs of RRT (total 117.6 h) were performed on ECMO support via the ECMO circuit itself. Access line pressures were not recorded during runs via the ECMO circuit as continuous alarming from the high pressure generated by the ECMO pump necessitated turning this monitor off. Return line and effluent pressures were significantly higher as predicted during runs via the ECMO circuit as blood is returned to the high-pressure post-pump portion of the ECMO circuit rather than the low-pressure native venous system during runs via a dialysis catheter. There were no significant differences between the two techniques with regards to RRT blood flow rates, ECMO flow rates, ultrafiltration efficiency (amount of fluid actually removed relative to the amount set to be removed), RRT circuit patency (number of RRT circuit cartridge changes per run), and RRT efficacy (percent decrease in creatinine levels) (Table 4). Two patients were unable to be maintained on RRT via the ECMO circuit due to low RRT blood flow rates and circuit clotting and required conversion to a dialysis catheter based approach. One of these patients also was unable to be maintained on RRT via a dialysis catheter and underwent multiple conversions from one approach to another, as well as a hybrid approach utilizing the ECMO circuit for RRT access and a dialysis catheter for RRT return, again with little success.
Details of RRT support.
RRT: renal replacement therapy; ECMO: extracorporeal membrane oxygenation.
Discussion
In the current study, approximately half of all patients treated with VA ECMO required RRT support. Lower albumin levels and postcardiotomy shock were predictors of new renal failure requiring RRT on ECMO. RRT was performed via a dialysis catheter or via the ECMO circuit, with access and return performed post-pump and pre-oxygenator. Although RRT performed via the ECMO circuit was associated with higher return pressures as expected, no differences between the two techniques were noted with regards to RRT or ECMO flow rates, ultrafiltration efficiency, RRT circuit patency, or RRT efficacy. Outcomes were much worse for patients requiring new RRT on ECMO support, with a doubling of the length of ECMO support required and less than one-third the survival rate compared to patients not requiring RRT on ECMO support.
Renal failure requiring RRT in patients on VA ECMO support is not uncommon, occurring in 20%–59% of cases. 5 In addition to preexisting renal insufficiency and cardiogenic shock, circuit related factors such as nonpulsatile flow dynamics, hemolysis, and the systemic inflammatory response also play significant roles.5,8,11 Early RRT may help avoid hypervolemia, metabolic acidosis, and has been suggested to promote recovery from the systemic inflammatory response by clearance of inflammatory cytokines.5,8
Aggressive volume management is a cruicial part of successful outcomes after VA ECMO. Based on the current study, when performing RRT on ECMO we were able to remove ~90% of the desired volume set to be removed with RRT via either a catheter based approach or via the ECMO circuit (ultrafiltration efficiency). We therefore believe that RRT on ECMO by either method is able to yield negative fluid balance as needed in these frequently volume overloaded patients. In the setting where access is an issue and ultrafiltration is all that is needed, aquapheresis is another reasonable option.
RRT in patients on ECMO support may be performed via a standard dialysis catheter or via the ECMO circuit. Utilizing the ECMO circuit for RRT obviates the need for dialysis catheter placement in the setting of anticoagulation and limited access sites due to cannulation needs. A hemofilter may be connected in-line with the ECMO circuit (uncommon), or an external RRT circuit (commonly used in most intensive care units) may be connected to the ECMO circuit. Several strategies for connecting an RRT circuit to an ECMO circuit have been described, each with their theoretical advantages and drawbacks.5–12
A typical ECMO circuit consists of a venous drainage cannula, a pump, an oxygenator, and an arterial return cannula. A hemofilter may be connected in-line with the ECMO circuit with the filter inlet connected post-pump/pre-oxygenator to allow the ECMO pump to push blood through the filter. The hemofilter outlet is reconnected pre-pump or post-pump/pre-oxygenator as well.5,6,11 The former strategy results in a shunt as blood passing through the hemofilter travels “backwards” along the ECMO circuit and therefore the ECMO flow probe should be placed at a point on the ECMO circuit proximal or distal to the hemofilter connections in order to avoid overestimation of ECMO flows. Benefits include a smaller priming volume as an RRT circuit is not utilized. However, in the absence of an RRT circuit, separate infusion pumps to deliver fluids and dialysate are needed. Furthermore, the inability to precisely control ultrafiltration is a significant drawback.5,7,11 Discrepancies in ultrafiltration volumes of up to 800cc/day have been reported with this method.5,11,13 More accurate methods of controlling ultrafiltration volume by periodically weighing the ultrafiltrate are possible but extremely cumbersome for nursing staff.5,6 An international survey of 65 centers performing RRT via an ECMO circuit demonstrated that 21.5% utilize a hemofilter in-line with the ECMO circuit. 14
More commonly, an RRT circuit is connected to the ECMO circuit. RRT access and return lines may be connected in a variety of combinations, including pre-pump, post-pump/pre-oxygenator, and post-oxygenator, again each with their own benefits and drawbacks. The RRT circuit measures pressures at both access and return sites. Unusually low or high detected pressures by the RRT circuit may occur based on what segment of the ECMO circuit the RRT line is attached to. Poor RRT blood flow rates result from poor inflow or high resistance to outflow. Subsequent alarms result interruption of flow, which can’t be overridden even when recognized to be benign without silencing the alarms altogether.5,7,9–11 This is suboptimal as it may lead to a loss of information regarding circuit function and represents a drawback to certain combinations of connections. 9 Poor RRT blood flow rates and flow interruption has been associated with RRT circuit thrombosis and reduced RRT circuit life expectancy and remains the main theoretical drawback to connecting an RRT circuit to the ECMO circuit.11,15
When performing RRT via a dialysis catheter, the RRT access line typically sees pressures in the range of 0–20mmHg (central venous pressure). The pre-pump venous segment of the ECMO circuit experiences the lowest pressures due to the suctioning effect of modern centrifugal pumps. Connecting the RRT access line to the ECMO circuit pre-pump (Figure 3) will result in detected pressures of −20 to −100 mmHg. RRT access line pressure alarms typically tolerate a pressure range of −250 to +200 mmHg. 15 Depending on cannulae size and position, ECMO pump speed and flows, and transient changes in preload (i.e. hypovolemia, patient coughing), severe drops in access line pressures may occur, which can lead to alarms and interruption of flow.5,7,9–11,15 A risk of air entrainment during manipulation of circuit attachment components is also possible given the very negative pressures of the venous line of the ECMO circuit.5–6,9–10 This can lead to pump malfunction as entrained air passes through the ECMO pump or even systemic embolization in severe cases.9,11
Renal replacement therapy circuit spliced into extracorporeal membrane oxygenation circuit with access line pre-pump and return line post-pump/preoxygenator.
Similarly, connecting the RRT return line pre-pump will also result in low pressures alarms (Figure 4). 9 RRT return line pressure alarms typically tolerate a range of −50 to +350 mmHg. 15 ECMO venous line pressures frequently can drop below this as described above, leading to alarms and interruption of flow. 11 However, the low resistance of the pre-pump segment of the ECMO circuit allows for optimal RRT return flow.10–11 The risk of air entrainment during manipulation of circuit attachment components is again a consideration in any strategy utilizing the pre-pump segment of the ECMO circuit.
Renal replacement therapy circuit spliced into extracorporeal membrane oxygenation circuit with access line post-pump/preoxygenator and return line pre-pump.
The post-pump/pre-oxygenator segment of the ECMO circuit as in our experience sees the highest pressure due to the pressure generated by the pump and the resistance of the oxygenator (Figure 1). 10 Connection to the post-pump segment of the ECMO circuit will typically result in high detected pressures of approximately 150–320 mmHg with persistent alarms and interruption of flows.5,9,11 In such cases, the alarm may be deactivated altogether. Connection to the post-pump/pre-oxygenator segment of the ECMO circuit eliminates the risk of pump failure in the setting of inadvertent air entrainment, and allows the oxygenator to filter such air. 11 However, given the high pressure in the post-pump components of the ECMO circuit, inadvertent air entrainment is unlikely. 5 Connection of the RRT access line in particular to the ECMO circuit post-pump/pre-oxygenator has the advantage of the high pressure-head generated by the ECMO pump which allows for optimal flow into the RRT circuit.10–11
Connection to the post-oxygenator segment of the ECMO circuit has the significant disadvantage of systemic embolization in the setting of inadvertent air entrainment as the oxygenator serves as a filter. 11 However, given the high pressure in the post-pump components of the ECMO circuit, inadvertent air entrainment is again unlikely. 5 When connecting the RRT return line to the post-oxygenator segment of the ECMO circuit, the risk of embolization of debris/thrombus generated by the RRT circuit exists. 6 The RRT return line in our opinion should always be connected to the ECMO circuit pre-oxygenator to filter such debris/thrombus (Figure 5).
Renal replacement therapy circuit spliced into extracorporeal membrane oxygenation circuit with access line post-pump/preoxygenator and return line postoxygenator.
Quadrox oxygenators have Leur-lock ports on the oxygenator inlet and outlet, and some groups have advocated utilizing these access sites for the RRT circuit.9,11 One group described connecting the RRT access line pre-oxygenator and the return line post-oxygenator, with a filter spliced in to capture air and microemboli from the RRT circuit. In addition to systemic embolization, another potential consequence of such a configuration is shunting as the blood passing through the RRT circuit bypasses the oxygenator alltogether, although this had minimal effect on systemic oxygenation as the ratio of RRT to ECMO blood flow was less than 0.1. 9
Tymowski et. al. utilize an initial connection strategy identical to ours, with RRT access and return lines connected to the ECMO circuit post-pump/pre-oxygenator (Figure 1). They proposed a standardized protocol for relocating access and return lines to a pre-pump location in the setting of high-pressure alarms, which obviated the need to silence alarms as we had to do in our experience. A drawback of this again is the risk of inadvertent air entrainment during manipulation of circuit attachment components, although this did not occur. 15 They noted higher RRT blood flow rates, prolonged circuit life with less circuit clotting, and similar efficacy (similarly defined by creatinine levels) in patients subjected this protocol. No adverse events occurred. 12
We adopted a policy of connecting both the RRT access and return lines post-pump/pre-oxygenator, first and foremost, for safety reasons (Figure 1). Such a strategy minimizes the air entrainment and systemic embolization risks associated with pre-pump and post-oxygenator configurations, respectively. Furthermore, we hoped to optimize flow into the RRT circuit with the pressure head generated by the pump. The main drawback is the frequent alarms with flow interruption, and the resulting need to silence the access line high pressure alarm. However, by all clinical parameters assessed, RRT was performed successfully with no differences in clinical outcomes compared with RRT performed via a catheter. In two patients, intractable flow interruptions led us to revert to a dialysis catheter-based approach, which was similarly problematic in one of the two patients. In this case we utilized a novel hybrid configuration utilizing the ECMO circuit for RRT access and a dialysis catheter for RRT return. Such an approach maximizes inflow to the RRT circuit with the pressure-head generated by the ECMO circuit and maximizes return to the low-pressure native venous system of the patient via a dialysis catheter. However, regardless of the configuration of RRT connections, RRT circuit thrombosis recurred for unclear reasons.
We noted a very high mortality for ECMO patients requiring RRT (84%), similar to that reported in the literature (71%–80%).8,9 In a metanalysis looking at outcomes of patients requiring ECMO and RRT support, need for RRT support carried an odds ratio for in-hospital mortality of 5.89 (p < 0.00001). 6 Hypoalbuminemia was a powerful independent predictor of new renal failure requiring RRT on ECMO support, possibly representing a marker of chronic illness. Postcardiotomy shock was similarly a powerful independent predictor of new renal failure requiring RRT on ECMO, and is consistent with the poorer outcomes noted in these highly complicated and very sick patients.
Limitations of this study include those inherent to a retrospective analysis utilizing chart review including incomplete data, potential inaccuracies in data, and potential for selection bias. A major limitation was the small sample size, and the inhomogenous nature of the patient cohort. In addition, due to differences in clinical practice across centers, extrapolation of results may be of limited value. Effluent flow rate and dose delivered relative to that prescribed is an important measure of RRT adequacy. Unfortunately the delivered effluent dose was not recorded in our electronic medical record system and therefore data regarding this parameter could not be analyzed.
In conclusion, renal failure requiring RRT is very common in patients on ECMO support. RRT in patients on ECMO support may be performed via a standard dialysis catheter, or via the ECMO circuit. When utilizing the ECMO circuit for RRT, we propose connection to the post-pump/pre-oxygenator component. Such a strategy minimizes the risks of air entrainment associated with manipulation of the low-pressure pre-pump component of the ECMO circuit and the risk of systemic embolization associated with manipulating the post-oxygenator components. Furthermore, it takes advantage of the high pressures in this segment to optimize inflow to the RRT circuit, although pressure alarms may need to be silenced. Further studies are needed to make stronger recommendations.
Footnotes
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
