Cannulation Configuration and its Effects on Bridging to Lung Transplant: Analysis of the Extracorporeal Life Support Organization Registry

Abstract

Background:

Donor stagnation and modification of lung allocation scores has resulted in a higher acuity of patient presentation prior to lung transplantation. Extracorporeal membrane oxygenation (ECMO) has been utilized as a bridge to lung transplant (BTT) although the effect of cannulation strategy on outcomes has not been well investigated. We sought to analyze contemporary data on ECMO BTT utilizing a large, international registry of patients.

Methods:

Utilizing the Extracorporeal Life Support Organization registry, all adult patients from 2010–2022 undergoing ECMO as a bridge to lung transplantation (ECMO BTT) were identified. Patients were stratified by support type: venovenous or venoarterial.

Results:

A total of 1066 patients were identified. ECMO BTT increased over the study period (p<0.001) as did survival to hospital discharge (p<0.001) with overall survival of 87.7%. Venovenous patients experienced less complications on ECMO including dialysis (16.7% vs. 25.3%, p=0.006), stroke (1.4% vs. 5.1%, p=0.004), limb ischemia (0.2% vs. 3.4%, p<0.001) and required ECMO less frequently in the postoperative period (41.0%% vs. 53.4%, p=0.002) and for less time (4 days [2–7] vs. 5 days [3–9], p=0.01). In-hospital mortality was significantly lower for venovenous patients compared to venoarterial (11.0% vs. 18.5%, p=0.005). Increasing center volume of ECMO BTT was protective of in-hospital mortality (p<0.001) with benefit observed after around 45 total BTT intent cannulations.

Conclusions:

ECMO BTT has resulted in improved post-transplant survival to discharge. Due to a higher rate of complications and worsened mortality, thoughtful implementation of venoarterial ECMO in BTT should be undertaken when assessing patient candidacy.

INTRODUCTION

Lung transplantation has been established as a life-saving treatment for patients with end-stage respiratory disease. However, the discrepancy between a high demand for organs and a scarcity of donors has led to long waiting periods and increased mortality among patients in need of transplant¹. While initially associated with unfavorable outcomes in pre-transplant patients, extracorporeal membrane oxygenation (ECMO) has re-emerged as a crucial bridge to lung transplantation (BTT), enabling patients to survive and stabilize while awaiting a suitable donor organ^2,3. Over the past decade, several retrospective studies have demonstrated ECMO BTT may be an effective management strategy for patients with end-stage lung disease^4–6.

While current literature has suggested ECMO BTT is a viable strategy for patients in end-stage pulmonary disease, there are still many questions about its current use. Notably, there is limited understanding of the factors that affect mortality and adverse outcomes of patients placed on ECMO pre-operatively especially regarding cannulation strategy.

The objective of this study was to analyze outcomes of a venovenous (VV) strategy for ECMO BTT as compared to a venoarterial (VA) strategy using data from the Extracorporeal Life Support Organization (ELSO) international registry. We evaluated the population characteristics, center volume, and ECMO-related variables and complications as predictors of mortality for patients who underwent cannulation with BTT intent and those surviving to ECMO BTT. We hypothesized that use of VA ECMO would be associated with worsened hospital survival secondary to higher complications compared to VV.

METHODS

Patients

Utilizing the ELSO registry, all adult patients 18 years or older requiring ECMO support from 2010–2022 were identified. Only patients cannulated for an indication of pulmonary support with BTT intention were included. Patients undergoing multi-organ transplant those with missing cannulation configuration, missing mortality data, or unknown reason for ECMO decannulation were excluded. To better capture the BTT population, only patients requiring ECMO support more than 24 hours prior to lung transplantation were included. Those patients decannulated prior to transplantation were considered to have recovered and were not included in the BTT group despite some undergoing lung transplant later while off ECMO support. Postoperative ECMO support was defined as requiring ECMO greater than 24 hours after lung transplantation due to variability with exact timing reports of decannulation timeframes after transplant. Lung disease classifications were based upon Lung Allocation Score groups: A – obstructive pulmonary disease, B – pulmonary vascular disease, C – cystic fibrosis, and D – restrictive lung disease. Patients that underwent venopulmonary (VP) cannulation or venovenous arterial (VVA) cannulation were included in the VV and VA groups, respectively, due to small absolute numbers of these subgroups. Patients were dichotomized based on their cannulation strategy; either VV or VA and then compared statistically. Lastly, there were 2 patient cohorts analyzed: 1) all patients cannulated with BTT intent (BTT Intent) and 2) patients that entered the operating room while on ECMO support (BTT). Our primary outcome of interest was survival to hospital discharge. Secondary outcomes included survival to transplant or recovery, and ECMO-related complications between the two cannulation configurations.

Statistical Analysis

Continuous variables were analyzed via Wilcoxon rank sum test with expression as either mean and standard deviation or median with interquartile ranges (IQR). Categorical variables were expressed as absolute numbers with percentages and analyzed via Chi-square testing or Fisher’s Exact Test for expected outcomes of less than 5 events. Linear regression was utilized to determine any timewise trends in utilization of ECMO BTT as well as changes in probability of survival over time. Logistic regression was utilized to determine potential factors associated with survival to hospital discharge following ECMO BTT. Variables included in the model were based on statistical significance on univariate analysis and known clinical associations. A stepwise approach for model creation was done in which covariates with p value >0.3 were excluded. A plot of the marginal effect of increasing case volume on predicted mortality was created utilizing logistic regression to control for other risk factors. Median imputation was utilized for missing data with all missingness less than 5%. All statistical analyses were carried out using SAS Version 9.4 (SAS Institutive, Cary, NC), R Version 4.3.0 (Posit, PBC, Boston, MA), and GraphPad Prism (GraphPad Software, Boston, MA) with a p-value less than 0.05 determining significance. This study was approved through the University of Virginia’s IRB (#20533)

RESULTS

Population Characteristics

From January 2010 until December 2022 there were 2,214 total patients identified from the ELSO registry as being cannulated for ECMO BTT intent. Of those patients, 1066 (48.2%) went on to undergo lung transplantation while on ECMO (BTT), 36 patients underwent transplant following decannulation from ECMO (1.6%), 428 (19.3%%) recovered and were decannulated without transplant, and 684 (30.9%) patients died on ECMO. BTT patients had a median ECMO duration of 11 days [5–26] until transplant, those decannulated following recovery had a median ECMO duration of 11 days [5–24], those decannulated prior to transplant had a median duration of 17 days [6–56] and those dying prior to transplant or recovery/decannulation had a median ECMO duration of 15 days [8–29]. Overall survival to discharge for all patients that were cannulated with BTT intent was 61.0%. For patients recovering to decannulation, survival to discharge was 89.5% (383/428) while BTT patients undergoing transplant on ECMO support had overall survival to hospital discharge of 87.7% (935/1066).

Figure 1. demonstrates the trend in outcomes for all patients cannulated with BTT intent; there were no significant trends by linear regression for the groups. There was a significant increase in the number of BTT intent ECMO cannulations over the study period from 2010 to 2022 with a peak of 403 total cannulations in 2021 (R²=0.84, p<0.001) (Figure 2) although no change in survival to discharge among all patients cannulated with BTT intent (R²=0.09, p=0.31, Figure 3).

Figure 1.

Linear regression demonstrating outcomes for all patients cannulated with BTT intent during 2010–2022. No change in proportion of patient outcomes.

Figure 2.

Trend displaying number of cannulations per year with BTT intent and number of patients going on to undergo transplant with significant increases in both groups over time.

Figure 3.

Survival to discharge over time for patients cannulated with BTT intent and patients undergoing transplant with significant improvement in survival for BTT patients over time.

Of all patients cannulated with BTT intent, 1809 patients (81.7%) were initially cannulated with a VV configuration with 14 (0.8%) of VV patients having a VP configuration, and 405 patients (18.3%) had a VA configuration with 38 (9.4%) of VA patients having a VVA configuration. Baseline characteristics and outcomes of BTT intent patients are shown in supplemental table 1. There was a lower proportion of females in the VV group (38.6% vs. 51.1%, p<0.001). VV patients were less likely to experience cardiac arrest prior to cannulation (3.0% vs. 10.6%, p<0.001), more likely to have an existing tracheostomy (30.4% vs. 17.0%, p<0.001) and less likely to be off the ventilator within 24 hours of cannulation (14.9% vs. 20.3%, p=0.007). VV ECMO patients experienced significantly less complications including circuit thrombosis (9.7% vs. 14.1%, p=0.01), oxygenator failure (9.4% vs. 13.5%, p=0.02), peripheral cannula bleeding (2.6% vs. 6.2%, p<0.001), stroke (2.7% vs. 9.1%, p<0.001), and limb ischemia (0.4% vs. 4.7%, p<0.001). Despite a lower number of complications, in-hospital mortality for all BTT intent patients was similar for VV patients compared to VA patients (38.6% vs. 40.7%, p=0.43).

There were 1066 total patients that entered the operating room on ECMO and underwent transplant. This group is referred to as the “BTT” group. BTT significantly increased over time with a peak of 229 cases in 2021 (R²=0.79, p<0.001, Figure 2), with improvement in survival to hospital discharge after transplant (R²=0.55, p=0.004, Figure 3); overall survival to hospital discharge was 87.7% among those that underwent transplant with a BTT strategy. Those decannulated prior to transplant had a survival to discharge of 90.3%.

Patients cannulated via VV strategy initially had a slower time to transplant compared to those with a VA strategy on Kaplan-Meier analysis (Log-Rank p<0.001) (Figure 4). Baseline characteristics and outcomes of BTT patients are shown in supplemental table 2. BTT patients that were initially cannulated in VV configuration were a smaller proportion female (36.6% vs. 48.9%, p<0.001). VV patients were less likely to have group B lung disease (5.1% vs. 31.5%, p<0.001) and less likely to be off the ventilator within 24 hours after cannulation (17.8% vs. 26.4%, p=0.008). Again, VV patients experienced less complications on ECMO including surgical site bleeding (12.5% vs. 20.2%, p=0.006), need for dialysis (16.7% vs. 25.3%, p=0.006), stroke (1.4% vs. 5.1%, p=0.004), or limb ischemia (0.2% vs. 3.4%, p<0.001). VV patients also required ECMO less frequently in the postoperative period (41.0%% vs. 53.4%, p=0.002) and for less time (4 days [2–7] vs. 5 days [3–9], p=0.01). In-hospital mortality was significantly lower for VV patients compared to VA (11.0% vs. 18.5%, p=0.005).

Figure 4.

Kaplan-Meier curve displaying time from ECMO cannulation to transplant based on cannulation strategy. Significantly faster time with VA cannulation compared to VV (Log-Rank p<0.001).

Risk-Adjusted Associations

A multivariable logistic regression was performed for the 1066 BTT patients. Center volume was included as a continuous variable and was based on total number of ECMO cannulations with BTT intent during the study period. Factors protective from mortality included increasing center volume (OR=0.98 [0.98–0.99], p<0.001), no ventilator requirement by 24 hours post cannulation (OR=0.14 [0.05–0.35], p<0.001), and VV configuration (OR=0.59 [0.35–0.99], p=0.048). Stroke (OR=10.58 [3.74–29.96], p<0.001), dialysis [OR=2.65 [1.69–4.15], p<0.001) and need for postoperative ECMO (OR=2.47 [1.60–3.81], p<0.001) were the most significant predictors of mortality (Table 1, Figure 5). A plot of the marginal effect of increasing center volume on predicted mortality was created using logistic regression and demonstrated a significant decrease in in-hospital mortality with increasing center volume when controlling for additional factors from the multivariable logistic regression model (Figure 6). The plot suggests that the mortality benefit of case volume occurs after around 45 total ECMO cannulations with BTT intent per center and continues as center volume rises.

Table 1.

Multivariable logistic regression of predictors of in-hospital mortality among patients successfully bridged to transplant while accounting for center volume.

	Odds Ratio	95% Confidence Interval	P Value
Prone within 24 hours of cannulation	0.56	[0.19–1.59]	0.27
Neuromuscular blockade within 24 hours of cannulation	0.70	[0.40–1.24]	0.22
Hyperbilirubinemia	3.49	[1.63–7.49]	0.001
Year of cannulation	0.93	[0.87–1.00]	0.04
Center volume (per case)	0.98	[0.98–0.99]	<0.001
Need for postoperative ECMO	2.47	[1.60–3.81]	<0.001
Stroke	10.58	[3.74–29.96]	<0.001
Off ventilator within 24 hours after cannulation	0.14	[0.05–0.35]	<0.001
Dialysis	2.65	[1.69–4.15]	<0.001
VV configuration (vs. VA)	0.59	[0.35–0.99]	0.048
Female	0.79	[0.51–1.22]	0.28

Figure 5.

Forest plot of predictors of mortality in BTT patients following transplant. Blue color indicates protective factors while red indicates factors increasing risk.

Figure 6.

Plot of marginal effects displaying adjusted mortality prediction compared to increasing volume of ECMO BTT cases. Estimates suggest greater than 45 total cases during the study period confers mortality benefit.

COMMENT

Through analysis of the ELSO registry, we demonstrated a significant increase in utilization of ECMO BTT over the past decade. Overall survival to discharge for all patients cannulated with BTT intent has remained somewhat stagnant. However, there is evidence of improving survival to discharge for patients undergoing lung transplant from ECMO bridging. Despite VA ECMO resulting in faster time to transplant, there was higher risk of mortality prior to discharge after transplant, likely due to a higher risk of complications related to ECMO and increased risk of postoperative ECMO.

Initial studies of ECMO use in acute respiratory distress syndrome (ARDS) did not demonstrate improvement in survival, halting its initial adoption and application toward bridging patients to lung transplant^7,8. However, as technology has evolved, contemporary results with ECMO for respiratory pathologies have improved with ECMO serving as a rescue therapy for patients failing conventional approaches⁹. This is especially important in the setting of limited donor lung supply despite use of donation after circulatory death donors to decrease waitlist times^1,10.

Outcomes of BTT ECMO have varied across the literature. Tipograf et al. demonstrated successful bridging to transplant with ECMO in 59% of patients with 91% survival in those transplanted ¹¹. Other studies comparing preoperative ECMO to no ECMO have demonstrated similar short and long-term survival without risk to graft survival^3,12–14. The overall survival to hospital discharge for all patients cannulated with BTT intent was 61.0% and remained somewhat stable over the decade. This survival percentage is similar to overall outcomes based on the ELSO registry report through 2022 which describes a 58% survival to discharge or transfer for pulmonary ECMO. A recent report by Loor et al. suggested prognostic indicators for ECMO BTT patients in hopes of improving patient selection and subsequent survival¹⁵. These factors include things such as younger age, ventilation status, evidence of hepatic congestion and pulmonary arterial pressures among others. Awake status during ECMO, defined as no sedation or light sedation, non-intubated and spontaneously breathing, was considered the most important prognostic indicator as this may allow for physiotherapy. We found that patients not requiring ventilator support by 24 hours after cannulation had better survival to discharge after lung transplant. Fuehner et al. compared awake ECMO bridge to lung transplant to mechanical ventilation and found significantly better survival to discharge in the ECMO group. Additional studies have suggested this may be one of the most important prognostic survival indicators^13,16.

Higher center volume of cannulations has previously been shown to correlate with survival to discharge in all ECMO patients as well as for BTT patients^17–19. Our analysis indicated that this benefit was evident after around 45 total cannulations with BTT intent with a continued increase in benefit as center volume increased. A study by Habertheuer et al. developed a tool called the Recipient Stratification Risk Analysis in Bridging Patients to Lung Transplant on ECMO (STABLE) score to better identify the mortality risk of patients bridging on ECMO to transplant²⁰. One of their variables included center volume of lung transplant per year as an important predictive factor. We are unable to scrutinize the total number of transplants, including non-BTT cases, performed at each participating center due to the deidentified nature of the database. We instead utilized the volume of BTT intent cannulations which is more comprehensive and better captured with our database. However, higher volume BTT centers may also be higher volume transplant centers which may be the reason for their improved outcomes in the postoperative period.

The decision when bridging a patient to lung transplant of whether to employ VA or VV ECMO is not always intuitive. In circumstances where patients are hemodynamically unstable due to poor oxygenation, employment of VV ECMO may restore hemodynamic stability and cardiac function. Majority of patients with end stage lung disease may also possess some degree of right ventricular dysfunction because of pulmonary hypertension and may require cardiac support in the form of VA ECMO. We found that utilization of VA ECMO was most common in group B patients where increased pulmonary vascular resistance is their main pathology. This was also true of female patients in our cohort, which tended to have a higher rate of group B disease.

Complications on ECMO were the main drivers of mortality in BTT patients and VA ECMO patients experienced a significantly higher number of complications such as renal failure, stroke, circuit related complications, limb complications, and higher risk for postoperative ECMO. This was in comparison to VV ECMO which had lower rates of almost every single circuit related complication. Many of these complications have previously been demonstrated to be markers of poor outcomes while on ECMO^11,21. VA ECMO has inherent risks to its utilization. Given utilization of arterial flow, there is higher risk of disrupting important vasculature and the threshold for circuit, oxygenator, and cannula exchange may be lower to avoid dangerous arterial emboli. Additionally, cannula placement into higher pressure vessels may increase risk of surgical complications and bleeding. Lastly, VA ECMO requires a higher degree of anticoagulation which may additionally increase the risk of bleeding.

Postoperative need for ECMO was also significantly associated with in-hospital mortality. This may be a marker of primary graft dysfunction (PGD) following lung transplant although the current database does not explicitly define PGD. A study of the United Organ Sharing Network (UNOS) by Mulvihill et al. identified preoperative ECMO use as associated with need for ECMO following lung transplant ²². However, other studies have demonstrated that continuation of ECMO into the postoperative phase for patients high at risk for PGD is sometimes warranted and may result in improved hospital survival prior to development of PGD or clinical deterioration leading to secondary ECMO cannulation ^23,24. We found not only that postoperative ECMO was independently associated with in-hospital mortality but it also occurred at a much higher rate in patients on VA ECMO and was required for a longer duration in VA ECMO patients. The reason for this is not known but should be further investigated and may perhaps be related to lung pathology as majority of patients requiring VA ECMO had pulmonary vascular disease. Due to database limitations, we lack information on ischemic times for grafts which also plays a major factor in the development of PGD and need for postoperative ECMO support.

There are limitations to the present study. As a retrospective study, there remains a risk of unmeasured confounding. Enrollment of ELSO centers have changed significantly over the study period, so this dataset may not have captured potential ECMO candidates and BTT patients. This could also explain the stark increase in ECMO BTT over time as perhaps several centers were not part of the registry in earlier years. We also do not know from the participating centers, how much their volume of lung transplant increased over the time period which could help distinguish whether we are using ECMO BTT more often or the increase of usage of this strategy is correlated with just overall number of increased transplants. Complication timing while on ECMO in relation to lung transplant is unknown for patients that required ECMO in the postoperative period. We would suspect that for patients undergoing transplant, complications such as stroke and renal failure likely would have occurred in the postoperative period as these complications may preclude transplant eligibility. Due to limitations of the database, we also do not possess data on long-term outcomes and how these may be changing over time. Lastly, baseline risk stratification for these patients is difficult due to lack of lung allocation scores for transplant and the wide variety of presenting pathologies.

ECMO BTT remains a viable strategy for patients with a significant increase in the number of cannulations and improvements in survival to hospital discharge over the last decade for patients surviving to transplant. VA ECMO use for BTT is associated with higher risk of ECMO-related complications and increased risk of ECMO following transplant, resulting in worsened mortality. The decision to pursue a VA strategy should be considered carefully but may be necessary in situations of hemodynamic collapse requiring cardiac support in addition to gas exchange.

Glossary of Abbreviations

BTT

Bridge to lung transplantation

ECLS

Extracorporeal life support

ECMO

Extracorporeal membrane oxygenation

ELSO

Extracorporeal Life Support Organization

PGD

Primary graft dysfunction

VA

Veno-arterial

VV

Veno-venous

VVA

Venovenous arterial