Extracorporeal membrane oxygenation liberation before heart transplantation in the modern allocation era: Predictors and one-year outcomes from the National OPTN/UNOS Registry

Abstract

Background

The clinical significance of ECMO liberation before heart transplantation remains poorly defined.

Objectives

To identify predictors of pre-transplant ECMO liberation and evaluate its association with 1-year post-transplant survival.

Methods

We conducted a retrospective cohort study using the national OPTN/UNOS registry, including adult isolated heart transplant recipients supported with ECMO at listing between October 18, 2018 and March 31, 2026. Predictors of pre-transplant ECMO liberation were evaluated using multivariable logistic regression, and the association between ECMO liberation and one-year post-transplant survival was assessed using Kaplan-Meier methods and multivariable Cox regression.

Results

Of the 1153 eligible recipients included in the final cohort, 208 (18.0%) were liberated from ECMO before heart transplantation. In multivariable analysis, mechanical ventilation at listing was the only independent predictor of a lower likelihood of ECMO liberation (aOR 0.51, 95% CI 0.34–0.78, p=0.002). Regarding post-transplant outcomes, recipients liberated from ECMO prior to transplantation demonstrated lower unadjusted post-transplant mortality; however, after adjustment for recipient demographic and clinical characteristics, ECMO liberation was not independently associated with 1-year post-transplant survival (aHR 0.81, 95% CI 0.49–1.35, p=0.42).

Conclusion

In this contemporary national cohort, ECMO liberation before transplantation occurred in about one in six heart transplant recipients. Mechanical ventilation was associated with reduced likelihood of ECMO liberation, whereas liberation itself was not independently associated with 1-year post-transplant survival after adjustment for recipient demographic and clinical characteristics.

Graphical abstract

Introduction

Extracorporeal membrane oxygenation (ECMO) is a vital form of temporary mechanical circulatory support for patients with advanced heart failure and refractory cardiogenic shock awaiting heart transplantation1, 2. The 2018 revision of the United States adult heart allocation system significantly increased priority for candidates supported with ECMO, resulting in its widespread utilization as a bridge-to-transplant strategy and shorter waitlist times for critically ill candidates. Consequently, ECMO-supported patients now represent a growing and clinically important subgroup of heart transplant recipients1, 3, 4, 5, 6, 7, 8.

Despite these advantages, ECMO support remains associated with substantial morbidity and mortality7, 9. Prolonged exposure has been linked to bleeding, thromboembolic events, infection, end-organ dysfunction, neurologic complications, and progressive deconditioning. These complications raise concerns that continued pre-transplant ECMO dependence may adversely influence post-transplant outcomes10, 11, 12, 13, 14. Accordingly, considerable clinical effort is directed toward achieving ECMO liberation whenever feasible, whether through heart transplantation, myocardial recovery, transition to alternative mechanical support devices, or stabilization with medical therapy alone15, 16.

However, the clinical significance of pre-transplant ECMO liberation remains poorly understood. Also, relatively little is known about the characteristics of recipients who achieve ECMO liberation before transplantation, or their subsequent post-transplant outcomes¹⁷. These knowledge gaps are highly relevant in the modern allocation era, where ECMO utilization has expanded considerably and management strategies continue to evolve. Therefore, using a contemporary national cohort of adult heart transplant recipients from the United Network for Organ Sharing (UNOS) registry, we sought to identify predictors of pre-transplant ECMO liberation and its association with 1-year post-transplant survival.

Methods

Study design

We performed a retrospective cohort study using the UNOS thoracic transplant registry, a national database that captures recipient, donor, waitlist, transplant, and follow-up information for heart and lung transplant candidates across transplant centers in the United States. Thoracic transplant records registered between May 10, 1984 and March 31, 2026 were initially screened. To ensure relevance to contemporary clinical practice and the current heart allocation system, we restricted our analyses to heart transplant recipients listed during the modern adult heart allocation era, beginning October 18, 2018, when the revised 6-tier adult heart allocation system was implemented.

Study population

Recipients were eligible if they were adults (≥18 years) receiving ECMO support at the time of heart transplant listing and subsequently underwent isolated heart transplantation between October 18, 2018-March 31, 2026. Pediatric (<18 years) and multi-organ transplant recipients were excluded. Recipients transplanted on the day of listing were excluded because they had no opportunity to experience ECMO liberation before transplantation. Recipients with missing ECMO support status at transplantation were also excluded because ECMO liberation status could not be determined.

Study size

No formal a priori sample size calculation was performed. The final sample size was determined by the number of adult isolated heart transplant recipients supported with ECMO at listing who met all prespecified eligibility criteria during the contemporary allocation era. All eligible recipients were included to maximize statistical power and minimize selection bias.

ECMO status at transplantation

ECMO status at transplantation was categorized according to whether recipients remained on ECMO support or had been liberated from ECMO before transplantation. All recipients were receiving ECMO support at the time of listing. Recipients who remained on ECMO support at transplantation were classified as having persistent ECMO support, whereas those who were no longer receiving ECMO at transplantation were classified as liberated. Accordingly, ECMO liberation was operationally defined as ECMO support at listing but not at transplantation.

Covariates

Covariates were selected a priori based on biological plausibility, prior transplant literature, and their potential associations with both ECMO liberation and post-transplant survival. Baseline variables included recipient age, sex, race or ethnicity, body mass index (BMI), primary cardiac diagnosis, ventilator support, inotrope use, and dialysis status at listing.

Age, sex, and BMI were included because of their established associations with transplant outcomes and overall clinical status. Ventilator support, inotrope use, and dialysis were included as markers of illness severity and end-organ dysfunction among critically ill transplant candidates. Dialysis was selected over serum creatinine because it more directly reflects clinically significant renal failure while minimizing sample loss from missing laboratory data. Race or ethnicity was included to account for possible disparities in transplant access, management, and outcomes. Primary cardiac diagnosis was included because the likelihood of ECMO liberation and post-transplant prognosis may differ across underlying disease etiologies.

Variable grouping

Several variables were categorized to improve clinical interpretability and reduce sparse-data instability. Race or ethnicity was grouped into White, Black, Hispanic, and Asian or Other. Due to small subgroup sizes, Native American, Pacific Islander, multiracial, and recipients with unknown race or ethnicity were combined with Asian recipients into the Asian/Other category to preserve model stability and reduce the risk of overfitting. Primary cardiac diagnoses were grouped into ischemic cardiomyopathy, nonischemic dilated cardiomyopathy, retransplantation or graft failure, and other diagnoses. Retransplantation/graft failure was maintained as a separate category because of its distinct clinical characteristics and prognostic implications. Dialysis status was categorized as no dialysis versus any dialysis because the clinically relevant distinction was the presence of severe renal dysfunction rather than dialysis modality.

Functional status was explored during model development but excluded from the final analyses because of concerns regarding data validity. Specifically, pediatric-specific functional status codes were identified among adult recipients, raising the possibility of significant misclassifications. The specific UNOS codes used to derive these variables are provided in Supplemental Table S1. Recipient ABO blood group was also evaluated as a potential predictor of ECMO liberation. Because it was not associated with ECMO liberation in exploratory analyses, it was not retained in the final multivariable model, consistent with our objective of maintaining a parsimonious adjustment model.

Additionally, time from listing to transplantation was evaluated during model development but was excluded from the adjustment models because waitlist duration occurs after listing and may lie on the causal pathway between ECMO status at transplantation and post-transplant outcomes. Hence, adjustment for such intermediate variables may obscure the associations of interest.

Statistical analysis

Descriptive statistics

Baseline recipient characteristics were summarized according to ECMO liberation status. Continuous variables are presented as means ± standard deviations (SD) or medians with interquartile ranges (IQR), as appropriate, and were compared using Student's t-test or the Wilcoxon rank-sum test, respectively. Categorical variables are presented as frequencies and percentages and were compared using the chi-square or Fisher's exact tests, as appropriate.

Predictors of ECMO liberation

Predictors of ECMO liberation before transplantation were evaluated using univariable and multivariable logistic regression. Crude odds ratios (cORs), adjusted odds ratios (aORs), and corresponding 95% confidence intervals (CIs) are reported.

Post-transplantation survival

The primary outcome of interest was 1-year all-cause post-transplant survival. Survival time was defined from the date of transplantation until death, last available follow-up, or administrative right censoring at 365 days after transplantation, whichever occurred first. Recipients experiencing day-zero mortality were handled by shifting the origin back by a half-day to ensure complete cohort inclusion and valid risk contribution.

Unadjusted survival curves according to ECMO liberation status were constructed using Kaplan-Meier (K-M) methods and compared using the log-rank test. Multivariable Cox proportional hazards (PH) regression with robust variance estimation was used to estimate adjusted hazard ratios (aHRs) and corresponding 95% CIs. The PH assumption was formally evaluated using Schoenfeld residuals. Variables violating the PH assumption were modeled using time-dependent coefficient interactions with the natural logarithm of analysis time to ensure more precise hazard estimation.

As a sensitivity analysis to address potential confounding by indication, we utilized propensity score matching using age, sex, race/ethnicity, BMI, primary cardiac diagnosis, ventilator support, inotrope use, and dialysis status at listing. Propensity scores were estimated using logistic regression, and recipients were matched 1:1 using nearest-neighbor matching without replacement and a caliper width of 0.2. We assessed covariate balance using standardized mean differences (SMDs) and Rubin's balance diagnostics, with Rubin's B statistic <25 and Rubin’s R between 0.5 and 2.0 considered indicative of acceptable balance.

All analyses were performed using StataNow version 19.5 (StataCorp LLC, College Station, Texas). Two-sided p-values <0.05 were considered statistically significant.

Ethics

This study used de-identified data from the Organ Procurement and Transplantation Network (OPTN) database obtained through UNOS under a Data Use Agreement. An Institutional Review Board waiver of informed consent was granted.

Results

Application of the study eligibility criteria yielded a final analytic cohort of 1153 adult heart transplant recipients who were supported with ECMO at the time of listing and subsequently underwent transplantation. Of these, 208 (18.0%) were liberated from ECMO before transplantation, whereas 945 (82.0%) remained on ECMO support at the time of transplantation. Among the ECMO-liberated recipients, 44 (21.2%) were supported with ECMO for less than 24 h. For the remaining 164 recipients who were on ECMO support for ≥1 day, the median ECMO duration was 6 days (IQR 3–10), compared with 4 days (IQR 2–6) among recipients who remained on ECMO until transplantation (p<0.001).

Among those liberated, 93 (44.7%) transitioned to medical therapy alone, 43 (20.7%) were bridged to Impella support, 18 (8.7%) to durable ventricular assist device (VAD), 21 (10.1%) to non-Impella percutaneous VADs, 5 (2.4%) to a total artificial heart, 24 (11.5%) to isolated intra-aortic balloon pump (IABP) support, and 4 (1.9%) to dual Impella and IABP support prior to transplantation (Figure 1). The median time from listing to transplantation was 4 days (IQR 3–8) for the overall cohort, 4 days (IQR 2–6) for recipients who remained on ECMO at transplantation and 13.5 days (IQR 7–24) for ECMO-liberated recipients.

Figure 1.

Study cohort selection and classification according to extracorporeal membrane oxygenation (ECMO) status at transplantation. Based on OPTN data as of March 31.2026. Abbreviation: UNOS=United Network for Organ Sharing.

The cohort had a mean age of 48.4 ± 14.4 years, was predominantly male (73.2%, n=844), and had a mean BMI of 28.1 ± 5.4 kg/m². The racial and ethnic distribution comprised 59.2% White (n=682), 20.4% Black (n=235), 10.2% Hispanic (n=117), and 10.3% Asian or Other recipients (n=119). Nonischemic dilated cardiomyopathy was the most common underlying cardiac diagnosis (51.5%, n=594), followed by ischemic cardiomyopathy (28.7%, n=331), retransplantation/graft failure (5.2%, n=60), and other diagnoses (14.6%, n=168).

Markers of advanced illness were common at listing; 291 recipients (25.2%) were on mechanical ventilation, 605 (52.5%) on inotropic support and 91 (7.9%) on dialysis.

Recipients who were subsequently liberated from ECMO before transplantation were less likely to be on mechanical ventilation at listing than those who remained on ECMO support at transplantation (15.4% vs. 27.4%, p<0.001). No significant differences were observed with respect to age, sex, race/ethnicity, BMI, primary cardiac diagnosis, inotrope use, or dialysis status (Table 1).

Table 1.

Characteristics at listing among adult heart transplant recipients supported with extracorporeal membrane oxygenation (ECMO), Organ Procurement and Transplantation Network (OPTN) data (October 18, 2018- March 31, 2026)

Characteristic	Total Cohort (N=1153)	Remained on ECMO (n=945)	Liberated from ECMO(n=208)	p-value
Age, years (mean±SD)	48.4±14.4	48.6±14.2	47.6±15.0	0.39
Body mass index, kg/m²(mean±SD)*	28.1±5.4	28.1±5.3	27.8±5.8	0.51
Male sex	844 (73.2)	694 (73.4)	150 (72.1)	0.15
Race/ethnicity				0.16
White	682 (59.2)	572 (60.5)	110 (52.9)
Black	235 (20.4)	183 (19.4)	52 (25.0)
Hispanic	117 (10.2)	92 (9.7)	25 (12.0)
Asian/Other	119 (10.3)	98 (10.4)	21 (10.1)
Primary cardiac diagnosis				0.02
Ischemic cardiomyopathy	331 (28.7)	273 (28.9)	58 (27.9)
Nonischemic dilated cardiomyopathy	594 (51.5)	471 (49.8)	123 (59.1)
Retransplantation /graft Failure	60 (5.2)	55 (5.8)	5 (2.4)
Other diagnoses	168 (14.6)	146 (15.5)	22 (10.6)
Mechanical ventilation	291 (25.2)	259 (27.4)	32 (15.4)	<0.001
Inotropic support	605 (52.5)	495 (52.4)	110 (52.9)	0.90
Any dialysis	91 (7.9)	79 (8.4)	12 (5.8)	0.21

Data are presented as mean ± standard deviation (SD) for continuous variables and n (column %) for categorical variables. Continuous variables were compared using independent-samples t-tests, and categorical variables were compared using Pearson's chi-square test or Fisher’s exact test as appropriate.

*Body mass index data were missing for 2 recipients; analyses were performed using available cases (Total N=1151, remained on ECMO n=943, liberated from ECMO n=208).

Asian/Other includes Asian (n=71), Native Hawaiian or other Pacific Islander (7), American Indian or Alaska Native (7), multiracial (8), and unknown (26).

Based on Organ Procurement and Transplantation Network (OPTN) data (October 18, 2018- May 31, 2026).

Predictors of ECMO liberation

In univariable analyses, Black recipients had higher odds of ECMO liberation than White recipients (cOR 1.48, 95% CI 1.02–2.14; p=0.04), and recipients on mechanical ventilation at listing were less likely to be liberated from ECMO than those without ventilatory support (cOR 0.48, 95% CI 0.32–0.72; p<0.001). However, after multivariable adjustment, only mechanical ventilation remained independently associated with ECMO liberation (aOR 0.51, 95% CI 0.34–0.78; p=0.002).

Age, sex, BMI, primary cardiac diagnosis, inotrope use, and dialysis status were not significantly associated with ECMO liberation in either univariable or multivariable analyses (Table 2).

Table 2.

Univariable and multivariable predictors of extracorporeal membrane oxygenation (ECMO) liberation prior to adult heart transplantation

Characteristic of recipient at listing		Crude odds ratio (95% CI)	Adjusted odds ratio (95% CI)	Adjusted p-value
Age	(per 1-year increase)	1.00 (0.99–1.01)	1.00 (0.99–1.01)	0.73
Body mass index	(per 1-kg/m² increase)	0.99 (0.96–1.02)	0.99 (0.97–1.02)	0.68
Sex	Female	reference
	Male	0.94 (0.67–1.31)	0.85 (0.60–1.21)	0.37
Race/ethnicity	White	reference
	Black	1.48 (1.02–2.14)	1.33 (0.91–1.96)	0.14
	Hispanic	1.41 (0.87–2.30)	1.25 (0.76–2.06)	0.38
	Asian/Other	1.11 (0.67–1.87)	1.02 (0.61–1.72)	0.93
Primary cardiac diagnosis	Ischemic cardiomyopathy	reference
	Nonischemic dilated cardiomyopathy	1.23 (0.87–1.74)	1.04 (0.71–1.52)	0.86
	Retransplantation/graft failure	0.43 (0.16–1.11)	0.45 (0.17–1.20)	0.11
	Other diagnoses	0.71 (0.42–1.21)	0.65 (0.38–1.14)	0.13
Mechanical ventilation	No	reference
	Yes	0.48 (0.32–72)	0.51 (0.34–0.78)	0.002
Inotropic support	No	reference
	Yes	1.02 (0.76–1.38)	1.10 (0.80–1.50)	0.56
Any dialysis	No	reference
	Yes	0.67 (0.36–1.25)	0.80 (0.42–1.51)	0.49

Multivariable analyses were performed using complete-case data (N=1150), whereas univariable analyses utilized all available observations (N=1153). Body mass index and dialysis status were missing for two recipients and one recipient, respectively. Adjusted odds ratios were derived from the multivariable logistic regression model including all baseline characteristics shown in the table. The composite Asian/Other category includes Asian, Native Hawaiian or Other Pacific Islander, American Indian or Alaska Native, multiracial, and recipients with unknown race/ethnicity. CI=confidence interval.

Based on Organ Procurement and Transplantation Network (OPTN) data (October 18, 2018- May 31, 2026).

One-year post-transplant survival

Of the 1153 contemporary heart transplant recipients supported with ECMO at listing, 1131 had complete survival data and were included in the survival analysis. This cohort comprised 926 recipients who remained on ECMO at transplantation and 205 who were liberated from ECMO support before surgery. During the first post-transplant year, 137 all-cause deaths occurred over 315,655.5 person-days of follow-up (864.2 person-years), yielding an overall mortality rate of 15.9 deaths per 100 person-years. The estimated 1-year post-transplant survival was 86.6% (95% CI 84.4–88.6%).

The persistent ECMO group contributed 699.4 person-years of follow-up and experienced 119 deaths (12.9%), corresponding to a mortality rate of 17.0 deaths per 100 person-years. In comparison, the liberated ECMO group contributed 164.8 person-years of follow-up and experienced 18 deaths (8.8%), translating to a mortality rate of 10.9 deaths per 100 person-years. Despite lower crude mortality among liberated recipients, K-M analysis demonstrated no statistically significant difference in 1-year post-transplant survival between recipients who remained on ECMO at transplantation and those liberated before transplantation (log-rank p=0.09, Figure 2). Consistent with these findings, univariable Cox regression showed no significant association between pre-transplant ECMO liberation and post-transplant all-cause mortality (crude HR 0.66, 95% CI 0.40–1.08; p=0.10).

Figure 2.

Crude 1-year post-transplant survival. Panel A: Kaplan-Meier estimates of post-transplant survival with 95% confidence intervals among adult heart transplant recipients supported with extracorporeal membrane oxygenation (ECMO) at listing. Panel B: Cumulative mortality according to ECMO status at transplantation. ECMO+ denotes recipients who remained on ECMO at transplantation, whereas ECMO- denotes recipients liberated from ECMO before transplantation. Survival distributions were compared using the log-rank test (p=0.48). Based on OPTN data as of March 31.2026.

In the multivariable Cox PH model, which included 1128 recipients with complete covariate data, ECMO liberation remained not associated with 1-year post-transplant mortality (aHR 0.81, 95% CI 0.49–1.35, p=0.42, Figure 3). Adjusted associations for all covariates are presented in Figure 4.

Figure 5.

Time-varying associations with post-transplant mortality. Predicted hazard ratios and 95% confidence intervals derived from exploratory time-varying coefficient Cox models. Panel A demonstrates the changing association between mechanical ventilation at listing and mortality over time. Panel B demonstrates the changing association between Asian/Other race and mortality relative to White recipients. Hazard ratios were estimated using interactions between covariates and the natural logarithm of follow-up time. Based on OPTN data as of March 31, 2026 (N=1128).

Figure 3.

Adjusted 1-year post-transplant survival. Panel A shows adjusted survival curves and Panel B shows adjusted cumulative mortality curves for recipients who remained on ECMO at transplantation (ECMO+) and those liberated from ECMO before transplantation (ECMO-). Curves were derived from the multivariable Cox proportional hazards model adjusting for recipient age, sex, race/ethnicity, body mass index, primary cardiac diagnosis, mechanical ventilation, inotropic support, and dialysis status at listing. Based on OPTN data as of March 31.2026.

Figure 4.

Adjusted predictors of all-cause 1-year mortality following heart transplantation among recipients supported with extracorporeal membrane oxygenation (ECMO) at listing. Hazard ratios and 95% confidence intervals were estimated using a multivariable Cox proportional hazards model adjusted for all variables shown. The solid vertical reference line indicates a hazard ratio of 1.0. Reference groups were persistent ECMO support at transplantation, female sex, White race/ethnicity, ischemic cardiomyopathy, absence of mechanical ventilation, absence of inotropic support, and absence of dialysis. Based on OPTN data as of March 31, 2026 (N=1128). Abbreviations: aHR=adjusted hazard ratio; CI=confidence interval; NICM=nonischemic dilated cardiomyopathy.

Proportional hazards assumption

Testing of the PH assumption using Schoenfeld residuals showed no evidence of global nonproportionality (global test p=0.10). The primary exposure, ECMO liberation status, also satisfied the PH assumption (p=0.43), confirming the validity of the regression estimates.

Although the global test did not indicate model-wide nonproportionality, individual Schoenfeld residual tests suggested time-varying effects for mechanical ventilation at listing (p=0.03) and Asian/Other race (p=0.004). Therefore, an exploratory time-varying coefficient Cox model was fitted. In this model, the adverse association between mechanical ventilation and mortality was greatest immediately after transplantation (aHR 2.94, 95% CI 1.18–7.35 at day 1) and progressively attenuated over time, becoming nonsignificant by day 28 (aHR 1.23, 95% CI 0.83–1.82; Figure 4A). Similarly, the excess mortality risk associated with Asian/Other race compared with White recipients was most pronounced immediately after transplantation (aHR 7.44, 95% CI 2.28–24.22 at day 1), but subsequently attenuated, becoming nonsignificant by day 28 (aHR 1.64, 95% CI 0.90–2.98; Figure 4B).

However, these findings should be interpreted cautiously. The overall PH assumption remained satisfied and several estimates were characterized by wide confidence intervals, particularly during the earliest phases of follow-up.

Sensitivity analysis: Propensity score-matched cohort

A propensity score-matched cohort of 410 recipients was created, with 205 recipients in each ECMO status group. Matching improved overall covariate balance, reducing the mean standardized bias from 10.6% to 5.5% and Rubin's B statistic from 40.8 to 21.5 (Supplementary Figure S1). During the first post-transplant year, 40 deaths occurred in the matched cohort: 18 deaths among recipients liberated from ECMO and 22 among those who remained on ECMO support at transplantation. K-M analysis demonstrated no significant difference in 1-year post-transplant survival between the two groups (log-rank p=0.49). In a Cox PH model, ECMO liberation was not associated with 1-year post-transplant mortality (HR 0.80, 95% CI 0.43–1.50; p=0.49), consistent with the primary analysis. There was not violation in the PH assumption (global Schoenfeld residual test p=0.52).

Discussion

In this contemporary national cohort of adult heart transplant recipients supported with ECMO at listing, ECMO liberation before transplantation occurred in approximately one in six recipients. Mechanical ventilation at listing was independently associated with a significantly lower likelihood of ECMO liberation, whereas age, sex, BMI, primary cardiac diagnosis, inotrope use, and dialysis status showed no independent associations. Importantly, pre-transplant ECMO liberation was not associated with altered 1-year post-transplant survival in either univariable, multivariable or propensity-matched analyses. Instead, post-transplant mortality was primarily driven by intrinsic recipient characteristics, including older age, female sex, higher BMI, retransplantation, and dialysis dependence. These findings suggest that while ECMO liberation is often regarded as a favorable pre-transplant milestone, ECMO status at transplantation is not an independent determinant of post-transplant survival after accounting for recipient characteristics.

Predictors of ECMO liberation

The finding that only 18% of the recipients were liberated from ECMO before their transplantation highlights the profound severity of illness among candidates listed while on ECMO. ECMO is generally reserved for patients with severe cardiogenic shock and advanced circulatory failure, many of whom require continuous mechanical circulatory support until transplantation1, 18. This low liberation rate likely stems from two factors, the sheer severity of illness in this cohort and the impact of the 2018 heart allocation policy, which substantially shortened waitlist times for status 1 candidates3, 19. Because patients on ECMO at listing are given highest priority, they get transplanted earlier (with median waitlist times decreasing from approximately 11 days to 5 days) such that the window for myocardial recovery and planned pre-transplant decannulation may be significantly compressed3, 5.

Mechanical ventilation emerged as the sole independent predictor of persistent ECMO support through to transplantation. This finding is biologically plausible because the requirement for invasive ventilatory support frequently reflects advanced cardiopulmonary compromise, greater hemodynamic instability, multisystem organ dysfunction, and diminished physiologic reserve. Such candidates are less likely to experience the clinical recovery necessary to permit discontinuation of ECMO support prior to transplantation. In contrast, inotropic support was similarly common in both groups (approximately 52%) and was not independently associated with ECMO liberation. This likely reflects the routine use of inotropes among patients receiving ECMO, where their presence alone may be less discriminatory of overall illness severity than the requirement for invasive mechanical ventilation. Prior studies have similarly identified mechanical ventilation as a marker of increased risk among ECMO-supported transplant candidates and recipients17, 20. Collectively, these observations suggest that respiratory failure identifies a particularly vulnerable subgroup of ECMO-supported candidates and may represent an important target for intensified multidisciplinary optimization while awaiting transplantation.

Post-transplant survival

One of the important findings of this study is the absence of a significant association between pre-transplant ECMO liberation and 1-year post-transplant survival in both unadjusted and adjusted models, and propensity-matched analyses. Several explanations may account for this finding. First, discontinuation of ECMO does not necessarily indicate complete physiologic recovery. Recipients who are liberated may continue to harbor substantial residual disease burden, end-organ dysfunction, frailty, or hemodynamic instability despite decannulation. Prior studies have shown that liberation from VA-ECMO for cardiogenic shock remains associated with substantial subsequent mortality, indicating that discontinuation of support does not fully reverse the adverse physiologic consequences of the initial shock insult21, 22.

Second, ECMO liberation occurred through heterogeneous clinical pathways within this cohort, including transition to VAD, IABP, or medical therapy alone. These pathways likely represent distinct clinical phenotypes with varying degrees of recovery and residual risk. Third, transplantation itself may mitigate many of the adverse physiologic consequences associated with advanced heart failure and cardiogenic shock, thereby reducing differences between recipients who remain on ECMO and those who are liberated before transplantation. Previous studies demonstrated substantial improvements in post-transplant survival among ECMO-supported recipients in the modern allocation era5, 23.

Notably, several recipient characteristics at listing emerged as important determinants of post-transplant survival. Older age was independently associated with an increased risk of death, consistent with prior transplant literature demonstrating the adverse impact of age-related frailty, greater comorbidity burden, and reduced physiologic reserve on post-transplant outcomes²⁴. Similarly, dialysis dependence was associated with substantially worse survival, likely reflecting advanced renal dysfunction, systemic illness, and increased susceptibility to postoperative complications25, 26. Retransplantation also emerged as a strong independent predictor of mortality. Consistent with prior reports, recipients undergoing transplantation for graft failure frequently represent a uniquely complex population characterized by prior sensitization, repeat sternotomy, greater surgical complexity, and accumulated comorbidity burden, all of which may contribute to inferior outcomes following transplantation27, 28. Female sex and higher BMI were also associated with increased mortality risk, further emphasizing the importance of recipient-specific characteristics in determining post-transplant outcomes. Multiple analyses have demonstrated that obesity, particularly BMI ≥35 kg/m², is associated with increased post-transplant mortality, higher rates of primary graft dysfunction, and treated rejection29, 30, 31, 32.

Together, these findings suggest that post-transplant survival among ECMO-supported recipients is driven predominantly by underlying recipient characteristics rather than ECMO status alone at the time of transplantation. ECMO liberation may therefore not be viewed as a surrogate marker of improved post-transplant prognosis. At the same time, liberated recipients did not experience worse outcomes, supporting the safety of transplantation in appropriately selected candidates regardless of ECMO status at transplantation. Future risk stratification efforts should focus on identifying high-risk recipient phenotypes and determining whether targeted pre-transplant optimization strategies can improve post-transplant outcomes.

Study strengths and limitations

This study has several strengths. It utilized a modern national cohort derived from the UNOS registry, enhancing generalizability across transplant centers in the United States. Restriction to the contemporary heart allocation era ensures direct relevance to current clinical practice and avoids many of the limitations associated with historical cohorts. Furthermore, the study evaluated both predictors of ECMO liberation and subsequent post-transplant outcomes, providing a comprehensive assessment of this increasingly important patient population. In addition, a propensity score-matched sensitivity analysis was performed to address potential confounding by indication and demonstrated findings consistent with the primary analysis, thereby strengthening the robustness of the observed associations. Finally, PH assumptions were formally evaluated, and exploratory time-varying analyses were performed when potential violations were identified.

Several limitations need also to be acknowledged. The observational design precludes causal inference and remains susceptible to residual confounding (clinician judgment, transplant center practices, and factors influencing decisions regarding ECMO discontinuation) despite multivariable adjustment and propensity score matching. Although the support strategy at transplantation among liberated recipients was characterized, the registry lacks detailed information regarding hemodynamic recovery, center-specific management practices, complications occurring during ECMO support, and clinical decision-making surrounding support transitions. Consequently, the mechanisms underlying successful ECMO liberation could not be fully evaluated. In addition, only recipients who ultimately underwent transplantation were included; therefore, these findings should not be extrapolated to all ECMO-supported candidates, particularly those who died or were removed from the waitlist before transplantation. Furthermore, our analysis of post-transplant survival did not account for donor characteristics, which are important determinants of post-transplant outcomes and may have contributed to residual confounding. The relatively small number of ECMO liberated recipients and post-transplant deaths may have reduced study power to detect modest differences in survival.

Conclusions

Among adult heart transplant recipients supported with ECMO at listing, ECMO liberation happened in approximately one in six recipients before transplantation. Mechanical ventilation at listing was independently associated with a lower likelihood of liberation. Although ECMO-liberated recipients demonstrated numerically lower post-transplant mortality, liberation was not independently associated with 1-year survival after adjustment for recipient demographic characteristics and illness severity. Instead, post-transplant outcomes were driven primarily by recipient factors, including age, sex, BMI, retransplantation, and dialysis dependence. These findings do not support the use of ECMO liberation status as a prognostic marker of post-transplant outcomes among recipients who ultimately undergo heart transplantation. Future work should focus on better understanding the clinical factors and support strategies associated with ECMO status transitions before transplantation and their relationship with post-transplant outcomes.

Funding statement

None.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplemental material

Supplementary material