Membrane Lung and Blood Pump Use during Prolonged Extracorporeal Membrane Oxygenation: Trends from 2002–2017

Abstract

Extracorporeal life support (ECLS) has grown in its application since its first clinical description in the 1970’s. The technology has been used to support a wide variety of mechanical support modalities and diseases including respiratory failure, cardiorespiratory failure, and cardiac failure. Over many decades and safety and efficacy studies, followed by randomized clinical trials and thousands of clinical uses, ECLS is considered an accepted treatment option for severe pulmonary and selected cardiovascular failure.

Extracorporeal life support involves the use of support artificial organs, including a membrane lung and blood pump. Over time, changes in the technology and the management of ECLS support devices have evolved.

This manuscript describes the use of membrane lungs and blood pumps used during ECLS support from 2002–2017 in over 65,000 patients reported to the Extracorporeal Life Support Organization (ELSO) Registry. Device longevity and complications associated with membrane lungs and blood pump are described and stratified by age group: neonates, pediatrics, and adults.

Introduction

Extracorporeal life support (ECLS) has grown in its utilization,¹ supporting evidence² and resultant clinical acceptance³ over the past 40 years. The term ECLS encompasses the family of extracorporeal support modalities for long-term support including respiratory failure, cardiorespiratory failure, cardiac failure and carbon dioxide retention.⁴ The evolution of ECLS is well documented in the Extracorporeal Life Support Organization (ELSO) Registry, an international database with more than 130,00 patients across more than three decades.⁵

Because the technology was used to support primary respiratory failure, the clinical application of ECLS was initially referred to as ECMO (extracorporeal membrane oxygenation). The first successful human case was described in 1972.⁶ ECMO first gained prominence for acute respiratory failure in neonates with safety and efficacy being demonstrated.^7–11 Randomized studies followed demonstrating a beneficial aspect to ECMO support as a treatment for neonatal respiratory failure.^2,12,13 With gradual acceptance of the technique, other centers began to apply treatment to neonates and expand use to older children and adults. Patients with certain aspects of cardiac failure were also supported.^14–16 However, prior to 1990, ECMO was mostly investigational; a focus at academic and research institutions which concentrated on neonatal extracorporeal research.⁵

In the 2000s expansion proliferated into widespread adult application. Adult ECMO trials showing better efficacy and survival, including the CESAR trial,¹⁷ H1N1 virus outbreak¹⁸ and EOLIA trial¹⁹ contributed to this growth. However, the longevity and durability of the support devices required for ECMO support have never been studied on a large scale. Over time, changes in the technology and the management of ECMO support devices evolved from roller pumps and silicone rubber gas exchange devices to the use of second-generation centrifugal pumps and artificial lungs constructed from polymethylpentene fibers. In addition, there has been a technological evolution of vascular access catheter development for venovenous support that included both dual lumen and bicaval catheters.

The purpose of this study is to describe the use of membrane lungs and blood pumps used during ECMO support from 2002–2017. Device longevity, as well as certain complications associated with membrane lungs and blood pump use, are described, and subdivided by neonates, pediatrics, and adults.

Methods

All ECMO runs reported to the ELSO Registry from 2002–2017 were included. Data were voluntarily submitted by individual ELSO member centers worldwide. These do not include data from other nonmember ELSO centers. All data were deidentified, with no specific patient or center information linked to an ECMO run. The University of Michigan Institutional Review Board (IRB) reviewed this study and determined that it did not meet criteria for IRB regulation (HUM00184999).

The definitions of the parameters described in this report (e.g., run characteristics, complications) are defined by ELSO. Although some definitions may be considered subjective, they are a consensus agreement of definitions agreed to by member centers. Data were reported for all patients and stratified into three subgroups by age: neonatal, pediatric, and adult. Neonatal patients are those 28 days or less; pediatric patients were defined as greater than 28 days and less than 18 years of age; adults were defined as 18 years of age and older.

The specific membrane lung and blood pump for each case is typically reported. Entries that were missing both the membrane lung and the blood pump were excluded from this analysis. The data on membrane lung use were categorized by the biomaterial of the device used for gas transfer: silicone rubber, polypropylene, or polymethylpentene. These various biomaterials were evaluated as they became available for clinical use. Blood pump data were categorized by pump type: roller pump, passive-filling roller pump, or centrifugal pump. The specific device manufacturers and models included in each category are listed in the Appendix.

The six most highly reported complications associated with use of a membrane lung or blood pump were described: hemolysis, membrane lung failure, pump failure, circuit clots, circuit air, and raceway rupture. Other complications, such as circuit change-out, connector cracks, heat exchanger malfunction, and non-roller-pump-related tubing rupture were reported at a very low incidence and were not included in this report. Complications were self-reported by the centers submitting data to ELSO. ELSO recommends that centers define mechanical complications as those complications that require intervention, such as a change in circuit components or equipment. The definition for hemolysis is a plasma free hemoglobin level > 50mg/dL sustained for at least two consecutive days. The incidence of a complication was defined as the reported number of each complication divided by the reported number of cases in a given time period. The exception to this was the complication of raceway rupture. As this is a complication specific to roller pumps, the incidence of raceway rupture was calculated as the number of reported raceway ruptures divided by the number of reported cases with a roller pump listed. The overall complication rate was calculated as the percent of cases with one or more complication.

The mode of support used in each case (e.g., veno-arterial, veno-venous) was not specified in these data; therefore, all modes are reported as a single group and not analyzed separately. The type of support (i.e., Cardiac, Pulmonary, ECPR) was reported but was not analyzed separately in this study. Data on overall patient outcomes (e.g., survival of ECLS, survival to discharge) were not reported in these data and therefore are not included in this study. In addition, other devices including vascular access catheters, hemofiltration devices and temperature control units were not described in this report.

Statistical Analysis

Data were described as median and interquartile range for continuous variables and number of observations and percent for categorical variables. Data were analyzed separately based on age subgroup (neonatal, pediatric, adult) and device type as indicated. Since the count of runtime hours was over dispersed, negative binomial regression analysis offset by the logarithm of the number of runs per year was used to assess the relationship between run times and study year. Differences in runtimes across years were compared using incident-rate ratios (IRR) produced by the negative binomial model, with 2009 selected as the reference year for comparison. The incident rate ratio is the ratio of the incident rate of any given study year to the incident rate of 2009 (the reference year). Parameter estimates when modeling counts are generally expressed in terms of incidence risk or rate ratios. Poisson regression analysis offset by the logarithm of the number of runs per year was used to examine complication rates over time for each age subgroup. Associations between pump type, membrane lung type, and individual complication rates (hemolysis, membrane lung failure, pump failure, circuit clots, circuit air, and raceway rupture) as well as all complications were examined via logistic regression analysis while controlling for number of hours on ECMO. Since roller pumps and silicone rubber oxygenators are considered the historical baseline pump and membrane lung types, respectively, they were used as reference categories for the regression analysis. Passive-filling roller pumps were included in the roller pump category for the above statistical analyses. Significance was set at p<0.05 and all analyses were performed using R software (R Core Team, Vienna, Austria).

Results

In the 15-year period from 2002–2017, 75,916 cases were reported. Figure 1 describes the number of reporting centers and cases per year broken down by each of the 3 age subgroups. There were 119 reporting centers in 2002 that nearly tripled up to 412 reporting centers in 2017. The number of neonatal cases over this time period has been relatively the same, with a small increase in pediatric cases. The largest growth has occurred in the adult population. From this dataset, 64,600 (85.1%) disclosed the membrane lung used and 65,429 (86.2%) disclosed the blood pump used. The cases used for analysis included 20,330 neonatal runs, 17,311 pediatric runs and 31,500 adult runs.

Figure 1:

ECLS runs per year stratified by age subgroup and total ELSO member centers reporting data per year. ECLS runs include those entries that report either the membrane lung and/or the blood pump used.

Runtime

Study year was significantly associated with runtime hours in the total sample (Chi-square=97.45; p<0.001) as well as each of the 3 age subgroups (neonatal: Chi-square=148.29, p<0.001; pediatric: Chi-square=39.39, p<0.001; adult: Chi-square=105.20, p<0.001). As illustrated in Figure 1, the predicted runtime for the neonatal group followed a consistent downward trend from 2009 to 2017. Runtimes were significantly lower in 2012–2017 compared to 2009, with the lowest runtimes seen in 2017 (164.14, 95% confidence interval [CI]=157.32–170.97). Runtimes in the neonatal subgroup did not significantly differ prior to 2009. The pediatric subgroup also saw runtimes trend generally downward from 2009 to 2017, though the trend was not as pronounced as in neonates. The lowest predicted runtime in pediatrics was in 2014 (172.77, CI=164.14–181.42), where the IRR was 0.85 (p<0.001) compared to 2009. Runtimes in 2012 (IRR=0.91, p=0.018) and 2017 (IRR=0.93, p=0.047) were also significantly lower than 2009. Pediatric runtimes in 2009 did not significantly differ from any prior year. Adult runtimes did not demonstrate a consistent trend. The predicted runtimes were lowest in 2008 (151.86, CI=137.39–166.32) and highest in 2013 (216.63, CI=208.12–225.14).

Membrane Lungs

A specific membrane lung was identified in 64,600 cases. Silicone rubber devices were used almost exclusively until about 2006, when polypropylene and polymethylpentene devices were gradually introduced (Figure 2). Polymethylpentene device use has increased since and has been the predominant membrane lung biomaterial used since 2009, as commercially produced silicone rubber devices were gradually discontinued.

Figure 2:

Median runtimes (in hours) for each year for all patients and for the 3 age subgroups. Error bars depict the standard error for the runtime for the given year and age subgroup. ref = reference year, 2009; * = significantly different from reference year (p<0.05).

Silicone rubber devices were used in neonates and pediatric patients a few years longer compared to adults (Figure 2). Adults ECLS runs have almost exclusively used polymethylpentene membrane lungs, as the increase in adult cases mirrors the increase in polymethylpentene membrane lung use.

Blood Pumps

A specific blood pump was identified in 65,437 cases. Roller pumps were used in the majority of all cases until about 2010 (Figure 3). A small user base utilized passive-filling roller pumps until the early 2010s, until those devices went out of production. Second-generation centrifugal pumps introduced in the mid-2000s became the predominant pump used which coincided with the expansion of use in adult cases. Gradually, centrifugal pumps expanded use in all age subgroups and became the most common pump type; however, roller-pump use is still common in neonatal and—to a slightly lesser degree—in pediatric applications (Figure 3). Roller-pump use remains uncommon in adults.

Figure 3:

Membrane lung use by membrane biomaterial and age subgroup. PMP = polymethylpentene, PP = polypropylene, SR = silicone rubber.

Complications

Circuit clots were the highest reported complication, followed by hemolysis and membrane lung failure (Table 1). Over time, the incidence rate of nearly all complications significantly decreased in each age subgroup (Table 1, Figure 4), which will be more closely described in subsequent sections. Longer runtimes were associated with increased odds of incurring complications in all age subgroups, with the exception of raceway rupture in adults. As an example, the odds of having any complication after 1 week on ECMO increased by a factor of 1.7, after 2 weeks by a factor of 2.7 and after 3 weeks by a factor of 4.5 (p<0.001).

Table 1.

Incidence of circuit-associated complications and relative change in incidence per year over the study period

	All Ages			Neonatal			Pediatric			Adult
	Incidence			Incidence			Incidence			Incidence
	‘02–09 (%)	‘10–17 (%)	Relative change per year (%)	‘02–09 (%)	‘10–17 (%)	Relative change per year (%)	‘02–09 (%)	‘10–17 (%)	Relative change per year (%)	‘02–09 (%)	‘10–17 (%)	Relative change per year (%)
Any Complication	37.0	25.3	−4.3 (−4.5, −4.0)	37.1	35.4	−0.4 (−0.9, 0.1)	36.8	30.2	−2.0 (−2.5, −1.4)	36.7	19.5	−7.1 (−7.7, −6.4)
Hemolysis	9.4	7.4	−2.6 (−3.2, −2.0)	8.3	12.1	5.1 (4.1, 6.1)	10.4	9.7	−0.5 (−1.5, 0.6)	11.3	4.7	−8.5 (−9.6, −7.3)
Membrane Lung Failure	10.1	4.5	−8.1 (−8.7, −7.5)	7.3	3.7	−7.0 (−8.3, −5.8)	10.1	4.3	−8.7 (−9.8, −7.5)	19.7	4.9	−14.3 (−15.2, −13.4)
Pump Failure	1.6	0.9	−6.2 (−7.6, −4.8)	1.7	1.2	−3.6 (−6.0, −1.1)	1.8	1.0	−6.1 (−8.8, −3.4)	1.0	0.8	−4.8 (−8.1, −1.2)
Circuit Clots	24.1	16.7	−4.2 (−4.6, −3.9)	27.0	24.8	−0.8 (−1.4, −0.2)	21.6	19.8	−0.6 (−1.3, 0.2)	19.5	12.4	−6.0 (−6.8, −5.2)
Circuit Air	3.3	2.2	−4.8 (−5.8, −3.8)	3.4	3.7	0.6 (−1.0, 2.3)	4.1	3.4	−2.2 (−3.9, −0.5)	1.6	1.2	−2.5 (−5.4, 0.6)
Raceway Rupture^	0.5	0.1	−3.1 (−7.2, 1.0)	0.3	0.2	−1.9 (−8.5, 5.0)	0.8	0.1	−4.8 (−10.4, 1.0)	0.1	0.0	−4.6 (−17.0, 9.0)

Note: “Incidence” columns indicate the average incidence across the 8-year periods of 2002–2009 (labeled ‘02–09) and 2010–2017 (labeled ‘10–17). The 95% confidence interval for the relative annual percent change in incidence is stated in parentheses. Relative change in complication incidence per year is based on Poisson regression analysis offset by the logarithm of the number of runs per year.

^{^}Raceway rupture incidence was calculated as a percent of runs using a roller pump.

Figure 4:

Blood pump use by pump design type and age subgroup.

Hemolysis

The only complication rate that has significantly increased since 2002 was hemolysis in neonatal runs (Table 1), with an overall incidence of 10.3% and an annual increase of 5.1% (CI=4.1–6.1%; p<0.001). Regression analysis of the association between hemolysis and the membrane lung biomaterial showed that the use of polymethylpentene membrane lungs in neonates was associated with lower odds of hemolysis compared to the historical reference category of silicone rubber membrane lungs (odds ratio [OR] = 0.68, CI=0.59–0.79) (Table 2). Regression analysis of the association between hemolysis and pump type demonstrated that the odds of hemolysis in neonates greatly increased with centrifugal pumps compared to the historical reference category of roller pumps (OR = 5.88, CI=5.17–6.71).

Table 2.

Results of logistic regression of circuit-associated complications versus membrane lung and pump types

			All		Neonatal		Pediatric		Adult
Complication	Device		OR	95% CI	OR	95% CI	OR	95% CI	OR	95% CI
All Complications
	ML	PMP	0.70*	(0.67, 0.75)	0.81*	(0.74, 0.87)	0.89*	(0.80, 0.98)	0.66*	(0.53, 0.82)
		PP	2.28*	(2.09, 2.48)	1.74*	(1.52, 1.99)	2.14*	(1.83, 2.50)	3.29*	(2.59, 4.19)
	Pump	Centrifugal	0.87*	(0.83, 0.91)	1.86*	(1.72, 2.01)	1.12*	(1.03, 1.21)	0.76*	(0.66, 0.87)
Hemolysis
	ML	PMP	0.53*	(0.48, 0.59)	0.68*	(0.59, 0.79)	0.67*	(0.57, 0.79)	0.45*	(0.31, 0.68)
		PP	1.56*	(1.37, 1.77)	1.17	(0.95, 1.43)	1.37*	(1.09, 1.70)	2.36*	(1.58, 3.61)
	Pump	Centrifugal	2.11*	(1.93, 2.31)	5.88*	(5.17, 6.71)	1.92*	(1.68, 2.20)	1.38*	(1.06, 1.82)
Membrane Lung Failure
	ML	PMP	0.37*	(0.33, 0.41)	0.39*	(0.33, 0.46)	0.36*	(0.30, 0.44)	0.35*	(0.25, 0.49)
		PP	4.56*	(4.04, 5.14)	4.12*	(3.40, 4.99)	4.59*	(3.71, 5.67)	5.31*	(3.79, 7.53)
	Pump	Centrifugal	1.25*	(1.14, 1.37)	1.09	(0.92, 1.29)	0.98	(0.84, 1.14)	0.79*	(0.64, 0.99)
Pump Failure
	ML	PMP	0.71*	(0.57, 0.89)	0.66*	(0.48, 0.89)	0.84	(0.56, 1.29)	0.97	(0.42, 2.61)
		PP	1.26	(0.90, 1.75)	1.47	(0.93, 2.24)	1.86*	(1.03, 3.26)	0.62	(0.18, 2.12)
	Pump	Centrifugal	0.76*	(0.63, 0.93)	1.13	(0.83, 1.53)	0.95	(0.69, 1.31)	0.60	(0.37, 1.04)
Circuit Clots
	ML	PMP	0.84*	(0.79, 0.90)	0.91*	(0.84, 0.99)	1.17*	(1.04, 1.31)	0.99	(0.76, 1.30)
		PP	1.55*	(1.41, 1.71)	1.34*	(1.16, 1.55)	1.24*	(1.03, 1.50)	2.99*	(2.23, 4.07)
	Pump	Centrifugal	0.71*	(0.67, 0.75)	1.31*	(1.20, 1.43)	0.99	(0.90, 1.09)	0.78*	(0.67, 0.92)
Circuit Air
	ML	PMP	1.09	(0.94, 1.25)	1.04	(0.85, 1.26)	1.28*	(1.01, 1.61)	0.95	(0.54, 1.77)
		PP	0.83	(0.62, 1.08)	0.87	(0.59, 1.25)	0.66	(0.39, 1.06)	0.75	(0.32, 1.69)
	Pump	Centrifugal	0.52*	(0.46, 0.58)	1.29*	(1.07, 1.56)	0.61*	(0.51, 0.74)	0.35*	(0.25, 0.50)
Raceway Rupture
	ML	PMP	0.42*	(0.27, 0.64)	0.54	(0.26, 1.05)	0.37*	(0.2, 0.64)	0.24	(0.04, 1.41)
		PP	1.17	(0.54, 2.22)	0.96	(0.23, 2.75)	0.61	(0.15, 1.69)	8.02*	(1.12, 52.58)

Note: OR = odds ratio, ML = membrane lung, PMP = polymethylpentene, PP = polypropylene. Regression controls for runtime. Silicone rubber (SR) was considered the reference category for membrane lungs (i.e., odds ratio [OR] = 1), and roller pumps were considered the reference category for pump types.

^*= p < 0.05

Hemolysis rates in pediatric patients were 9.9% and did not change significantly during the study period (Table 1). Adults had the lowest hemolysis rate at 5.1%, with a relative annual decrease of 8.5% per year (CI=7.3–9.6%; p<0.001). Regression analysis of hemolysis rates versus device types found that, like neonates, the odds of hemolysis decreased with polymethylpentene membrane lungs for pediatric (OR = 0.67, CI=0.57–0.79) and adult patients (OR = 0.45, CI=0.31–0.68), while the odds increased with centrifugal pump use in both groups (pediatrics: OR = 1.92, CI=1.68–2.20; adults: OR = 1.38, CI=1.06–1.82, respectively) (Table 2).

Membrane Lung Failure

Membrane lung failure occurred in 6% of all runs (Table 1). This rate decreased for all age groups across the study period, with adults experiencing the most rapid decrease of 14.3% annually (CI=13.4–15.2%; p<0.001) and neonates experiencing the slowest decline of 7.0% annually (CI=5.8–8.3%; p<0.001) (Table 1). In all age subgroups, PP MLs were associated with markedly increased odds of failure relative to SR (overall OR = 4.56, CI=4.04–5.14), whereas PMP MLs had lower odds of failure (overall OR = 0.37, CI=0.33–0.41) (Table 2).

Pump Failure

Pump failure was a rare complication, occurring in just over 1% of all runs. The rate of reported pump failure declined over time in each age subgroup: neonates at a rate of 3.6% per year (CI=1.1–6.0%; p=0.005), pediatrics at 6.1% per year (CI=3.4–8.8%; p<0.001), adults at 4.8% per year (CI=1.2–8.1%; p=0.008) (Table 1). Centrifugal pumps were associated with slightly lower odds of pump failure in all patients (OR=0.77, CI=0.63–0.93), though this relationship was not seen in any of the age subgroups when analyzed separately (Table 2).

Circuit Clots

Circuit clots were reported more commonly in neonates (25.8% of runs), followed by pediatrics (20.4%) and adults (13.0%). The neonatal subgroup saw a relative annual decrease in the incidence of circuit clots of 0.8% (CI=0.2–1.4%; p=0.007). Adult patients also saw an annual decrease in the incidence of circuit clots of 6.0% (CI=5.2–6.8%; p<0.001) (Table 1). There was no change over time in the incidence of circuit clots in pediatric patients. On regression analysis, the odds of circuit clots increased with MLs made of PP in neonates (OR = 1.34, CI=1.16–1.55), pediatrics (OR = 1.24, CI=1.03–1.50) and adults (OR = 3.00, CI=2.23–4.07) (Table 2). There were no consistent associations of pump types with the incidence of circuit clots.

Circuit Air

The incidence of circuit air was 2.5% for all patients across the study period. This incidence decreased by 4.8% per year (3.8–5.8%; p<0.001). On age subgroup analysis, pediatric patients were the only age subgroup that experienced a significant annual decrease in circuit air of 2.2% (0.5–3.9%; p=0.012) (Table 1). No consistent patterns arose in the associations between device use and the incidence of circuit air.

Raceway Rupture

Roller-pump raceway rupture is a rare complication with incidence never exceeding 1% of all roller-pump runs in any year. This incidence did not significantly change during the study period in any age subgroup (Table 1). The significant association between PP MLs and raceway rupture in adults was based on 7 reported episodes of raceway rupture in adult patients across the study period.

Discussion

The benefits of analyzing a large data set allows for the description of trends over time as well as reflection of changes in the state-of-the-art. These data describe three key areas: the changes of device use over time, the device longevity, and the incidence of circuit-associated complications. This analysis does not attempt to directly link any complication causation directly to any specific device or component. It is known and is expected that there are patient risks associated with and related to any form of extracorporeal support, such as bleeding, hemolysis and possible device malfunction/failure. When combined with circulatory support over several weeks or even months, the risk of complications would be anticipated to escalate.

Data from this study define two distinctive time related ECMO technology periods. The ECMO I technology era consisted primarily of using roller pumps and silicone rubber membrane lungs. This technology was used from the 1990s to about 2008. ECLS use was adapted from standard cardiopulmonary bypass devices used during cardiac surgery. All devices were used off-label because at that time, no pumps or membrane lungs were manufactured or cleared for use greater than 6 hours. Considering the ECMO circuit had no air interfaces, unlike the typical circuit used during cardiac surgery, the patient could be supported for days or weeks instead of a few hours. The gold standard membrane lung was the Kolobow silicone rubber device. The lung was characterized by a long blood path that generated a high-pressure gradient, often requiring daily platelet and red cell transfusions. The Kolobow lung was first used in the 1960s and was manufactured for nearly 50 years by 3 different companies until the device ceased production around 2012.²⁰

Roller pumps were commonly used during the ECMO I era but ran a risk for pump raceway tubing failure or for a tubing disruption or blowout occurring elsewhere in the circuit. The alternative would be a centrifugal pump, which in theory would provide a better safety advantage for extended support by providing better longevity. However, there was limitation in the duration of use of these first-generation centrifugal pumps, related to excessive heat production and seal failure, which resulted in hemolysis.^21–23 Although some centrifugal pumps were used successfully, widespread use slowly evolved over time and their use began to rise toward the end of the ECMO I era with the advent of newer centrifugal pump designs based on the Mendler concept in which the stagnant zones of the pump continuously flushed and washed to prevent overheating.²⁴ Thus, any comparison related to improvements in ECLS technology over time relates to the initial use of a silicone rubber membrane lung and a roller pump.

Trends in equipment use started to show a change during the latter part of the 2000s, which led to what may be referred to as the ECMO II technology era. Neonatal cases, which formed the foundation of ECMO support in the 1990s and early 2000s began to plateau. Pediatric cases continued to rise slowly, related to new reporting centers, but not to the extent of the expansion into adults. Adult ECMO trials showing better efficacy and survival, including the CESAR trial,¹⁷ H1N1 virus outbreak¹⁸ and EOLIA trial,¹⁹ These results may have contributed to the large expansion of cases and new centers. Although unknown, more experienced centers may support a wider variety of more difficult cases. At the same time, respiratory support, the primary application of the technology, was being supplemented by a growing number of cardiac applications, including postcardiotomy support, bridge to transplantation or ventricular assist device implantation and extracorporeal cardiopulmonary resuscitation.^{16, 25–31} In 2020 ECMO support was being implemented in another new population of severe respiratory failure related to the COVID-19 pandemic.^32–34

The ECMO II era (after 2008) is defined by the rapid expansion into adults along with the implementation of second-generation membrane lungs and centrifugal pumps. New centrifugal pumps, some being magnetically levitated while others having more efficient designs with pivot bearings, became available and were paired with lower resistance membrane lungs made from polymethylpentene fibers.³⁵ Polymethylpentene fibers offered technical advantages in terms of more efficient priming, reduced hemodynamic resistance and better control and preservation of coagulation proteins compared to their predecessors.^36–38 Although there was a brief interlude in which microporous polypropylene hollow fiber membrane lungs were evaluated, their longevity was limited by their fiber wettability that had a propensity to leak plasma, resulting in frequent device replacement.^39,40 Some ECLS circuits had also been adapted to integrate both the membrane lung and pump into one unit.^41,42

Device longevity has increased, and performance has improved over time as cases have become more complex. In such cases of bridge to recovery, bridge to device or bridge to transplantation, the circuit and devices might require more frequent replacement related to time, wear or declining performance. One notable application of multiple device use and variations of support modalities was on a single patient supported for 605 days.⁴³

There are limitations in describing such a large set of data. The data can be described, analyzed and interpreted in a myriad of ways. Some of the complication reporting is subjective and reporting could vary across centers. For instance, a small residual air bubble may or may not be reported as circuit air based on its size or position in the circuit. In addition, it would be expected that clots would form within the circuit or some circuit components after days or weeks of use. There are different anticoagulants, anticoagulation strategies and measurement techniques that also vary amongst the reporting centers. In addition, considering that there are differences in circuit designs, length of use and blood flow rates and patterns, a certain amount of “acceptable or visible” thrombus in the circuit may be tolerated more or less by each individual center. These centers may or may not report it as a circuit complication, especially if the circuit supports the patient’s gas exchange and metabolic requirements. There could also be varying institutional protocols to electively change a circuit or component, regardless of its function, based on time or an abnormal coagulation panel after some predetermined timepoint. In contrast, other centers may choose not to replace a device until it becomes dysfunctional or it fails altogether. Obviously the longer the patient is supported, the higher the risk of having any circuit-related complication. For instance, patients supported for several weeks or months have likely had device or circuit changeouts.

Additionally, the focus of the study was to describe device use and to not associate patient outcome, survival, mode of support (e.g., venoarterial, venovenous), and type of support (pulmonary, cardiac, ECPR) with any specific device or device type. Outcome data are routinely reported biannually by ELSO, as well as the number of cases and reporting centers. These data can be found on the ELSO web site (https://www.elso.org/Registry/Statistics/InternationalSummary).

ECMO devices and related disposables have always been used off-label, with very few exceptions. These devices were never labeled or cleared for long-term use but rather for shorter-term procedures (<6 hours) associated with cardiac surgery. Thus, ECMO patients have been supported successfully on devices well past their labelled indications for use. New devices or those previously used off label may be resubmitted to regulatory bodies for a longer labelled period of use. Data from the ELSO Registry may be a valuable tool in relabeling some devices.

Conclusion

The ELSO data described in this study (2002– 2017), clearly show that certain membrane lungs and blood pumps can be used safely off-label for extended periods. Complications related to hemolysis, circuit thrombosis, circuit air and pump and membrane failure do exist, but many have actually declined over time. Device changeout was relatively low. Data, as described in this report may be a useful benchmarking tool in conjunction submission of device performance to regulatory agencies. Future devices and new biomaterials will likely continue to improve and lower the risks of using ECLS for even longer periods, and future work may include measuring outcome standardized by hours of support to address length and time.

Figure 5:

Trends in the incidence of circuit-associated complications. Incidence calculated as a percentage of all ECLS runs among all ages for a given year. Raceway rupture incidence was calculated as a percent of runs using a roller pump.