Early Outcomes of Heart Transplantation Using Donation After Circulatory Death Donors in the United States

Limited donor availability for heart transplantation (HT) in the US continues to drive efforts to expand the donor pool. Most recently, improvements in procurement techniques and encouraging early outcomes have renewed interest in organ donation after circulatory death (DCD).^1,2 The modern experience with DCD HT in adults has been led by St Vincent’s Hospital in Australia and Royal Papworth Hospital in the United Kingdom, which have published a combined 111 DCD heart transplants since 2014.^3–8 Early data from these centers suggest higher rates of primary graft dysfunction with DCD HT but reassuring mortality rates up to 3 to 5 years after HT.^3,7 In the United States, DCD HT has gained increasing popularity since its first adult DCD HT in 2019.⁹ However, there is little published data on the US experience for DCD HT.

In addition, procurement methods for DCD hearts vary across international and US experiences as strategies to mitigate ischemic injury are influenced by national and regional laws and institutional practices. Direct procurement and perfusion (DPP) involves withdrawal of life-sustaining treatment and donor cardiectomy upon confirmation of circulatory death. Due to unmitigated warm ischemia and the inability to assess donor heart suitability before implantation, DPP is routinely augmented with ex vivo machine perfusion (EVP), which has been widely adopted as a prevailing strategy for DCD procurement. An alternative strategy, normothermic regional perfusion (NRP), uses extracorporeal circulation with exclusion of the cerebral circulation to reperfuse and reanimate the donor heart in situ. Evolution in these practices has contributed to increased interest in DCD heart utilization. Continued surveillance is warranted to monitor outcomes of transplantations using DCD donors and assess procurement strategies. The aim of the current study was to evaluate trends in practice and early outcomes of DCD HT in the United States since 2019.

Methods

Data Source

The United Network for Organ Sharing (UNOS) registry contains prospectively collected data on all solid organ transplantations in the United States. Because of the sensitive nature of the data accessed for this study, requests to access the data set from qualified researchers trained in human subject confidentiality protocols may be sent to UNOS. This study was deemed exempt from review by the Institutional Review Board at the Medical University of South Carolina.

Study Population

Patients undergoing HT from January 2019 to September 2021 were considered for inclusion in the analysis. Exclusion criteria were multiorgan transplant and patients under 18 years of age. Patients were excluded if survival data were missing. HT recipients were grouped according to method of donor organ procurement, either conventional donation after brain death (DBD) or DCD.

The primary outcome was 1-year survival. Secondary outcomes included rates of drug-treated acute rejection, dialysis, stroke, and pacemaker implantation before hospital discharge and hospital length of stay.

DCD procurement methods were characterized using data collected by UNOS, including utilization of EVP. DPP and NRP procurements were distinguished by the length of time from circulatory standstill to cross-clamp for cold cardioplegia perfusion, with NRP procurements identified by an interval >30 minutes. This distinction between DPP and NRP procurements is assumed based on available literature about these strategies. Namely, recent United States and international series of DCD HT using NRP universally report reperfusion times >30 minutes.^7,10,11 However, United States and international centers use organs procured with DPP only when functional warm ischemic time (ie, the time after withdrawal of life support after which systolic blood pressure falls to <50 mm Hg until reperfusion) is <30 minutes.^3–8,12 The distribution of times between circulatory arrest to cold cardioplegia in this analysis is shown in Figure S1 and demonstrates 2 unimodal distributions, suggesting 2 distinct groups of patients between which these times do not overlap and supporting our rationale for distinguishing DPP and NRP procurements.

Statistical Analysis

Categorical variables are summarized with counts and percentages. Continuous variables with normal distribution are presented with means and SDs, and those with non-normal distributions are presented with medians and interquartile ranges. Pearson χ² test and Fisher exact test (if the frequency of any variable was <5) were used to compare categorical variables. One-way ANOVA and Kruskal-Wallis tests were used to compare normal and non-normal continuous variables, respectively. Nearest neighbor propensity-score matching without replacement was performed, and an absolute standardized mean difference of <0.10 across all covariates was considered a successful match. All donor, recipient, and donor-recipient matching characteristics were used for propensity matching with some exceptions. Due to the nature of DCD procurements, organ ischemic times are necessarily longer in DCD compared with DBD and encompass a period of EVP or NRP in which the organ is not truly ischemic. Therefore, organ ischemic time was not used for propensity matching. In addition, recipient listing status was not used for propensity matching due to expected differences between DBD and DCD recipients, who were more likely to be listed at lower priority statuses. All variables used for matching and their standardized mean differences between patient groups are shown in Figure S2. Missing variables were managed with complete cases analysis. A Cox proportional hazards model was constructed for the overall cohort including all covariates used for propensity matching to validate the propensity-matched analysis.

Kaplan-Meier analysis was used to estimate 1-year survival, and comparisons were tested using log-rank tests. A Cox proportional hazards model was used to quantify the risk-adjusted hazard of DCD HT for 1-year mortality among propensity-matched HT recipients. Univariable analysis was used to identify risk factors for 1-year mortality among patients undergoing DCD HT. To account for center-level differences between those performing DCD HT and those performing only DBD HT, a subanalysis was performed including only DBD and DCD recipients transplanted at centers performing DCD HT. All multivariable Cox models were constructed using covariates with significant association with 1-year mortality on univariable analysis. A 2-sided P<0.05 was considered statistically significant. Analyses were performed using Stata, version 16.1 (StataCorp LLC, TX).

Results

Patients

A total of 7836 adult patients underwent isolated HT in the observed study period. Of these, 349 patients (4.5%) had missing survival data, and 1 patient had missing data about procurement strategy (DCD versus DBD). Therefore, 7486 patients were included for analysis, among which 229 patients (3.0%) underwent DCD HT. The monthly volume of heart transplants using DCD procurements increased significantly over the study period (Pearson R=0.679, P=0.001; Figure 1). Similarly, the number of centers performing DCD HT increased significantly from 2.5% (n=3/120) of all centers in 2019 to 16.5% (n=20/121) in 2021 (Pearson R=0.978, P<0.001).

• Download figure
• Download PowerPoint

Demographic Characteristics of the Study Population

Baseline characteristics of patients undergoing HT grouped by DCD or DBD heart procurement are summarized in Table 1. Recipients of DCD hearts were more likely to be transplanted at UNOS statuses 3 to 6 compared with DBD recipients, who were more likely to be transplanted at status 1 and 2 (P<0.001). The majority of DCD HTs (90.4%) occurred at centers in the fourth quartile of overall HT volume, compared with 53.8% of DBD HTs (P<0.001). In addition, DCD recipients had longer median times on the waitlist (47 versus 32 days, P=0.02). DCD recipients had a higher median body mass index and were less likely to be intubated or admitted to an intensive care unit at the time of transplantation compared with DBD recipients. In addition, DCD recipients were more likely to receive no mechanical bridging or bridging with a durable ventricular assist device while DBD recipients were more frequently bridged with intraaortic balloon pump, percutaneous endovascular ventricular assist devices, and extracorporeal membrane oxygenation.

DCD donors were more likely to be younger (median age 29 versus 32 years, P<0.001), male (87.3% versus 71.5%, P<0.001), and White (79.4% versus 63.3%, overall P<0.001) compared with DBD donors. In addition, the mechanism of death among DCD donors was more likely to be trauma and less likely to be cerebrovascular or drug overdose compared with DBD donors. The median organ travel distance was significantly longer for DCD organs compared with DBD organs (381 versus 222 miles, P<0.001).

Propensity-Score Matching

Propensity matching yielded 1974 (89.8%) DBD patients and 224 (10.2%) DCD patients whose characteristics are summarized in Table S1. One hundred eighty-seven (2.6%) DBD patients and 4 (1.7%) DCD patients were excluded from matching due to missing covariate data. The propensity-matched cohorts were well-balanced with absolute standardized mean differences <0.10 across all covariates (Figure S1).

Survival Analysis

Kaplan-Meier survival for the study population before and after propensity matching is shown in Figure 2. In the unmatched cohort, 1-year survival was 92.7% among DCD recipients compared with 90.6% for DBD (hazard ratio [HR], 0.77 [95% CI, 0.44–1.37]; P=0.38). In the propensity-matched cohort, 1-year survival was 91.1% among DCD recipients compared with 90.2% for DBD (HR, 0.80 [95% CI, 0.44–1.43]; P=0.44).

• Download figure
• Download PowerPoint

The Cox proportional hazards model for 1-year mortality in the matched cohort is shown in Table S2. Independent predictors of 1-year mortality after HT included higher recipient age, body mass index, creatinine, and bilirubin at transplantation, primary diagnosis of ischemic cardiomyopathy or congenital heart disease, mechanical ventilation, and bridging with durable ventricular assist device. A Cox model, including all patients without propensity matching, was used to validate the propensity-matched analysis and is shown in Table S3. After risk adjustment using all covariates used in the propensity-matched analysis, DCD transplantation was not associated with an elevated risk for 1-year mortality (HR, 0.88 [95% CI, 0.49–1.57]; P=0.658).

Secondary Outcomes

Secondary outcomes in the unmatched and matched cohorts are summarized in Table 2. Before and after propensity matching, acute rejection requiring treatment before discharge occurred significantly more frequently among DCD recipients compared with DBD recipients (14.7% versus 10.1% among matched patients, P=0.03). There were no differences in hospital length of stay or rates of dialysis, stroke, or pacemaker implantation between DBD and DCD recipients before or after propensity matching. Rates of rejection within 1 year of transplantation were not analyzed due to a large proportion of missing data in the DCD group.

Subanalysis Limited to HTs With DCD

Among the 229 DCD HTs occurring in the observed study period, 3 (1.3%) were unable to be confirmed as controlled DCD procurements. Data defining the agonal period and time of circulatory death were not available for 4 patients (1.8%). Thus, a total of 222 DCD HTs with complete data are characterized in Table 3. Among these, 175 were identified as DPP procurements (78.8%) and 47 as NRP (21.2%). The time between confirmed circulatory death and cross-clamp for cold cardioplegia perfusion had 2 distinct unimodal distributions, in which 1 group ranged from 55 seconds to 16 minutes, and the other ranged from 44 minutes to over 3 hours (Figure S1). Therefore, DPP and NRP procurements were distinguished by our predetermined threshold of >30 minutes for NRP. Median times to cold cardioplegia were 5.27 minutes for DPP and 68.13 minutes for NRP. The agonal period, defined by UNOS as the time after which the donor systolic blood pressure drops below 80 mm Hg or systemic oxygen saturation is <80% until circulatory standstill, was a median of 19.7 minutes among all DCD HTs (DPP 17.5 minutes and NRP 21.8 minutes). Median organ travel distance was 415 nautical miles for DPP procurements and 229.5 for NRP, and time from cross-clamp to implantation was a median of 6.24 hours for DPP and 4.17 hours for NRP. EVP utilization was reported in 81.0% of DPP procurements and 33.3% of NRP procurements. Monthly volume for DCD HT since 2019 is shown in Figure 3. DPP was the first and most frequently used strategy over the study period. Monthly frequencies of NRP transplantations have increased in volume since January 2020 (Pearson R=0.681, P<0.001) and accounted for 26.6% of DCD HT in 2021. One-month, 6-month, and 1-year survival were 98.8%, 92.9%, and 91.7% for DPP and 100%, 96.6%, and 96.6% for NRP (Figure 4). Procurement strategy was not significantly associated with 1-year mortality after DCD HT (HR, 0.41 for NRP relative to DPP [95% CI, 0.05–3.16]; P=0.39).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Univariable analysis for 1-year mortality risk after DCD HT is shown in Table 4. Increasing recipient age and waitlist time were associated with an increased risk for 1-year mortality (HR, 1.07 per year [95% CI, 1.00–1.13]; P=0.04, and HR, 1.03 per 30 days [95% CI, 1.00–1.06]; P=0.027, respectively).

Subanalysis Limited to DCD HT-Performing Centers

A subanalysis was performed including only HT recipients at DCD-performing centers. This included 2277 patients, of whom 2048 (89.9%) underwent DBD HT and 229 (10.1%) underwent DCD HT. Demographic characteristics are summarized in Table S4. Differences in donor and recipient characteristics between DBD and DCD HT were similar in this subanalysis of DCD HT centers as in the primary cohort, including lower rate of recipient intensive care unit admission and mechanical ventilation, lower priority listing status, longer organ travel distances, younger donor age, and higher donor male prevalence among DCD HTs compared with DBD.

Kaplan-Meier survival among DBD and DCD HT recipients at DCD HT centers is shown in Figure S3. Unadjusted 1-year survival was 92.7% for DCD recipients compared with 91.3% for DBD (HR, 0.84 [95% CI, 0.56–1.25]; P=0.55). After risk adjustment, DCD HT was not significantly associated with an increased risk for 1-year mortality after HT (HR, 0.81 [95% CI, 0.44–1.47]; P=0.49; Table S5). There were no significant differences in secondary outcomes after DBD and DCD HT (Table S6).

Discussion

In the past decade, the use of extended criteria and hepatitis C-positive donors in the United States has contributed to a >60% increase in the number of recovered donor hearts and has shown acceptable outcomes after transplantation.^13–15 HT with DCD presents a new strategy to further increase donor availability, and surveillance of outcomes is imperative as adoption of DCD HT expands. This analysis investigated trends and early outcomes of DCD HT in the United States.

With 229 patients, this report represents the largest registry analysis of DCD HT in the US published to date. We demonstrate that although DCD HT occurred in a highly selected group of donors and recipients, DCD was not associated with an elevated hazard for 1-year posttransplant mortality in a propensity-matched cohort. Additionally, the rate of DCD HT has increased steadily since 2019 and accounted for >5% of US heart transplants in 2021. However, DCD HT is still performed at a minority of centers. These findings suggest that DCD HT may be safely adopted more broadly in the United States with the potential to deliver up to 700 additional heart donors per year, according to a previous analysis of the UNOS registry.¹

This analysis did not detect a difference in early survival between DCD and DBD recipients up to 1 year, which is corroborated by findings from international centers and other US series. In the Royal Papworth Hospital series, Messer et al⁷ reported comparable 1-year survival between 79 DCD recipients and a propensity-matched DBD group. In Australia, Dhital et al⁴ reported a higher rate of immediate graft dysfunction but similar 5-year survival among 32 DCD recipients compared with unmatched contemporary DBD recipients. Most recently, the first US series of 127 DCD HT was published from the UNOS registry.¹⁶ This analysis by Madan et al¹⁶ demonstrated no difference in 30-day and 6-month survival between propensity-matched DCD and DBD recipients. However, none of these series have detected a difference in early rejection. The current analysis demonstrated a nearly 50% increase in drug-treated rejection before hospital discharge among DCD recipients compared with DBD recipients both before and after propensity matching. While clinical research and animal studies have identified that primary graft dysfunction may occur after DCD HT with prolonged warm ischemic times, the biologic basis for acute rejection after DCD HT requires further investigation.^3,17–19 Additionally, this finding should be interpreted in the context of possible differences in the posttransplant management of DCD recipients, who may be more aggressively treated for rejection compared with DBD recipients. Further surveillance for acute and chronic rejection after DCD HT is warranted.

Importantly, procurement strategies differ across DCD HT practices, and increasing utilization of NRP has introduced heterogeneity to analyses of DCD HT outcomes.^20,21 DPP with EVP is the sole procurement strategy in the Australian series and the predominant strategy in the 5-year Papworth series (72.2% DPP with EVP versus 27.8% NRP).^4,7 In the only published comparative analysis of DPP versus NRP, the Papworth series demonstrated no difference in survival between the 2 strategies, although the NRP group had fewer days of ventilatory support, a reduced frequency of dialysis, and a shorter hospital length of stay.⁷ The analysis presented here is the first to describe practice trends and outcomes with DPP and NRP strategies in the United States. We show that the US experience with DPP preceded NRP by 1 year, and NRP utilization started in January 2020, coinciding with the first published series of DCD HT using NRP in the United States.¹¹ In their pilot study of 8 DCD HTs using NRP, Smith et al¹¹ reported that all donor hearts were successfully reanimated in situ within 1 minute of extracorporeal circulation. Despite 1 heart-lung recipient requiring ECMO for primary graft dysfunction, survival was 100% at a median follow-up of 304 days. Another US series of DCD HT using NRP by Hoffman et al¹⁰ included 15 patients and reported an organ recovery rate of 71.4% from DCD donors using NRP and a 100% 30-day survival for recipients of hearts procured using this strategy. Heart recovery and utilization rates from DCD donors using DPP and NRP strategies are outside the scope of this analysis, and the low mortality rates among DPP and NRP recipients in this analysis limits the statistical power to compare survival between these 2 strategies. However, we demonstrate acceptable early survival with both. Importantly, registry analyses comparing these strategies require the reliable collection of variables that characterize procurement methods. Notably, the rate of EVP utilization with DCD HT in the United States is likely to be higher than this analysis reports, particularly for DPP procurements. DPP with cold static storage only (ie, without EVP) has not been reported in the international or US experience, except for a series of pediatric DCD HT.²² This suggests that the reported EVP utilization rate of 81% among DPP transplants in our analysis may represent underreported data. Importantly, the results of the multicenter prospective randomized DCD Heart Trial will soon be published comparing early survival among recipients of DCD HT using DPP and EVP with the TransMedics Organ Care System to recipients of standard criteria DBD donors.¹² This will provide long-awaited data on DPP with EVP in 180 patients in the United States.

Several authors cite favorable characteristics of NRP procurements over DPP with EVP.^8,10,11,20 With NRP, in situ reanimation allows earlier reperfusion to abort warm ischemia, which centers speculate may improve rates of DCD heart utilization and rates of primary graft dysfunction after implantation.^8,11 While robust comparative analyses have yet to be published, the experience with NRP at Papworth has been favorable with an implantation rate of 100% compared with 83% of DPP/EVP hearts.⁸ NRP also allows donor functional assessment upon weaning of extracorporeal circulation, which may provide a truer representation of contractile function under loaded conditions compared with EVP.^23,24 However, the broader adoption of NRP has been met with several ethical concerns addressing its potential conflict with the definition of irreversible circulatory death.^25–27 While outside the scope of this discussion, ethical boundaries require clear definition as NRP protocols are established at new centers. In addition, further investigation is needed to elucidate any relative benefits between DPP/EVP and NRP strategies.

This analysis demonstrates that recipients and donors for DCD HTs are highly selected for lower risk characteristics, including avoidance of nondischargeable mechanical bridging strategies and younger, male donors. However, there was no difference in age between DCD and DBD recipients. Notably, this analysis demonstrated that recipient age was associated with an increased hazard for 1-year mortality after DCD HT (HR, 1.07 per year), and this should remain a consideration in candidate selection. There was also a significant association between centers’ overall HT volume and DCD HT volume, and the majority of DCD HTs in this analysis occurred at centers in the highest quartile of overall volume. This may bias the outcomes of this analysis by overrepresenting high-performing or highly resourced centers. However, this analysis found similar posttransplant outcomes between the primary cohort and the subanalysis cohort which excluded DBD recipients transplanted at non-DCD–performing centers. Nevertheless, the small number of low- and mid-volume HT centers represented in this analysis limits the generalizability of our findings.

We also report that DCD recipients were more likely to be transplanted from UNOS listing status 3 to 6 compared with DBD recipients, who were transplanted more frequently as status 1 and 2. In addition, median waitlist time was >50% longer among patients receiving DCD hearts compared with DBD. In their analysis of the UNOS registry, Madan et al¹⁶ analyzed waitlist time after transplant candidates modified their donor criteria to accept DCD hearts. Compared with DBD recipients, DCD recipients waited less than half the time for HT after being designated as accepting DCD hearts.¹⁶ These findings suggest that DCD HT is an effective strategy for reducing waitlist time and increasing the likelihood of transplantation. However, in this analysis, longer waitlist times were associated with an increased hazard for 1-year mortality after DCD HT (HR, 1.03 per 30 days), which should be considered in recipient selection. Further dedicated analyses on waitlist outcomes may better characterize the impact of DCD HT in expanding organ availability and improving waitlist survival, transplantation rates, and post-HT survival.

This study has several additional limitations. First, DCD HT remains a novel strategy in the United States, and follow-up time is, therefore, limited. A minority of patients in the unmatched and matched DCD cohorts reached the primary end point of 1-year posttransplantation, and further surveillance for 1-year mortality outcomes is warranted. In addition, selection bias inherent to this retrospective analysis may not be fully accounted for with propensity matching. For example, centers with reduced regional competition may be able to select DCD donors who have lower risk characteristics compared with DBD donors, and these inherent differences in donor characteristics may be incompletely captured in this analysis. Importantly, the DCD cohort in this analysis likely represents many of the patients enrolled in the DCD Heart Trial, which may introduce further selection bias (eg, inclusion of recipients transplanted at high-volume, high-performing centers, exclusion of DCD candidates with prior organ transplantation or diagnosis of chronic renal insufficiency). This may also influence the frequency of DPP/EVP relative to NRP strategies presented in this analysis.

Additionally, registry data are limited by variability and errors in data collection, particularly for variables recently introduced in the data set for characterization of DCD HT such as hemodynamic data characterizing the agonal period and the time of circulatory standstill. Furthermore, any variables not collected in the UNOS registry are not accounted for in this analysis, including several relevant variables describing DCD procurement which may vary significantly between institutions. Variables capturing primary graft dysfunction, such as requirement for ECMO, are not well-defined in the registry or are frequently missing. In addition, this study lacks the granularity of data to fully characterize DCD heart procurements, including hemodynamic parameters and biomarker levels during functional assessment and precisely defined times for warm ischemia, machine perfusion, in vivo reperfusion, and cold storage. Finally, the onset of the COVID-19 pandemic occurred within the study period, and its influence on center volume, recipient and donor selection, and recipient outcomes is not accounted for in this analysis.

This study investigated the trends and early outcomes of DCD HT in the United States since adoption of this strategy in 2019. Observation over 3 years demonstrates a growing volume of DCD HT with increasing participation of centers across the country. There is no detectable difference in survival between DCD and DBD recipients at 1 year, both before and after propensity matching, although increased rates of early drug-treated rejection among DCD recipients warrants close surveillance. Outcomes of the early experience for DCD HT in the United States is consistent with existing international data and encourages broader adoption of this practice.

Article Information

Supplemental Material

Tables S1–S6

Figures S1–S3

Footnotes