Evolving utilization of donation after circulatory death livers in liver transplantation: The day of DCD has come

Abstract

Compared to donation after brain death (DBD), livers procured for transplantation from donation after circulatory death (DCD) donors experience more ischemia-reperfusion injury and higher rates of ischemic cholangiopathy due to the period of warm ischemic time (WIT) following withdrawal of life support. As a result, utilization of DCD livers for liver transplant (LT) has generally been limited to short WITs and younger aged donor grafts, causing many recovered DCD organs to be discarded without consideration for transplant. This study assesses how DCD liver utilization and outcomes have changed over time, using OPTN data from adult, first-time, deceased donor, whole-organ LTs between January 1995 and December 2019. Results show that increased clinical experience with DCD LT has translated into increased use of livers from DCD donors, shorter ischemic times, shorter lengths of hospitalization after transplant, and lower rates of retransplantation. The data also reveal that over the past decade, the rate of increase in DCD LTs conducted in the United States has outpaced that of DBD. Together, these trends signal an opportunity for the field of liver transplantation to mitigate the organ shortage by capitalizing on DCD liver allografts that are currently not being utilized.

1 ∣. INTRODUCTION

Donation after circulatory death (DCD) refers to the procurement of organs after diagnosed and confirmed irreversible cessation of circulation. It is generally employed in candidate donors that do not meet strict brain death criteria but who have irrecoverable neurologic injury, for which the patient’s medical team and next of kin agree that further interventions are futile. With growing reliance on DCD donors in the in 1990s, the Maastricht classification categorized DCD donors into 4 types: Category I donors were dead on arrival at the hospital, category II donors died after unsuccessful resuscitation, category III donors were awaiting cardiac death, and category IV donors experienced a cardiac arrest while brain dead. Category II was later subdivided into IIa and IIb as out of hospital and in hospital deaths, respectively.¹ Categories I, II, and IV are uncontrolled procurements (meaning time of cardiac arrest is unplanned), while III is a controlled procurement and can be planned for at the time of withdrawal of life support.²

Reservations around transplanting DCD livers derive from reported higher rates of graft failure compared to donation after brain death (DBD) livers due to ischemia-reperfusion (IR)-induced early allograft dysfunction (EAD) and primary non-function (PNF),^3-5 as well as delayed biliary complications and ischemic cholangiopathy (IC).^6-9 DCD liver transplants (LTs) are also associated with increased need for retransplantation and prolonged waitlist times for those needing retransplant due to lack of waitlist priority, until recently.¹⁰ Collectively, these factors resulted in less than 10% of LTs arising from DCD donors in 2019, despite the fact that DCD comprised 22% percent of total organ donors.¹¹

The current practice of frequently discarding livers from DCD donors provides an opportunity to expand the donor pool if approaches can be defined to mitigate the IR injury-related EAD and high rates of IC. Scalea et al. showed that young DCD donors (<50 years old) with short cold ischemic times (CIT) (<6 hours) had equivalent graft survival (GS) compared to DBD LTs.¹⁰ In addition, a single-center experience with 38 transplants suggested that selected extended criteria DCD liver donors (donor age >50 years, body mass index >35 kg/m², warm ischemic times (WITs) >30 min, or macrosteatosis >30%) with an optimization protocol showed equivalent survival outcomes to matched deceased donor LT.¹² Other single-center experiences at Mayo Clinic Jacksonville and the Ochsner Clinic have also reported equivalent 1- and 3-year patient and graft survival rates between DCD and DBD allografts.^13,14

The emergence of machine perfusion technology in LT has the potential to expand the utilization of DCD livers. Ex vivo liver perfusion at normothermic (35-38°C), subnormothermic (20-25°C), and hypothermic temperatures (2-10°C) can assess the performance of marginal liver allografts and possibly resuscitate DCD livers to meet transplant suitability.^15-20 This study examines the trends in the practice of DCD liver utilization and evolving outcomes between January 1995 and December 2019. We also explore the potential for expanding the organ pool through greater use of currently discarded DCD livers, perhaps the largest untapped pool of potentially transplantable livers. Finally, we examine the early experience with machine perfusion of DCD livers to assess the prospect for greater DCD utilization.

2 ∣. METHODS

2.1 ∣. Data sources

This study analyzed data from the Organ Procurement and Transplantation Network (OPTN) STAR file released on April 2020 based on data collected through March 2020. The content in this paper is the responsibility of the authors alone and does not necessarily reflect the views or policies of the Department of Health and Human Services, nor mentions of trade names, commercial products, or organizations imply endorsement by the U.S. Government

2.2 ∣. Study population

For donor and recipients’ distribution analyses, we identified all deceased donors who had at least one organ donated and all LTs from deceased donors from January 1995 to December 2019. Those with missing data regarding DCD versus DBD information (n = 174), liver disposition (n = 2025), donor age (n = 1), and without WIT information (n = 3731) were excluded. For outcome comparison, we identified all adult, first-time, liver-only whole-organ transplants from deceased donors from January 1995 to December 2019 without missingness in DCD or donor age.

2.3 ∣. Outcome classification

The outcomes of interest were hospital length of stay, rate of retransplantation, CIT, WIT, patient survival (PS), and all-cause graft survival (GS) following LT. GS was defined as time until either retransplantation or patient death with a functioning graft. We calculated WIT using the withdrawal of support time and the cross-clamp time, as these were consistently recorded in the OPTN database for DCD donors regardless of transplant center.

2.4 ∣. Statistical analysis

Demographics and clinical characteristics were compared using chi-square tests, Student’s t-tests, and Wilcoxon-Mann-Whitney as appropriate. The Bonferroni correction method was used for multiple comparison. Survival rates were presented in Kaplan-Meier curves and analyzed by log-rank tests. Time to the outcome was defined as the time from the date of transplant to the date of outcome (death or graft failure) and censored for loss to follow-up or end of the study period.

We performed comparisons of PS and GS after propensity score matching to assess the effect of DCD on post-transplant survival. The concept of using propensity score matching in analyzing the survival data was well-described by Dr. Peter C. Austin.²¹ A propensity score matching method was used to eliminate the baseline confounders between comparing groups. According to Austin’s method, factors related to outcome but possibly not related to treatment should be included in the propensity prediction model. We used the nearest neighbor matching with 1:1 ratio, without replacement, and with a caliper of width equal to 0.2 of the standard deviation (SD) of the logit of the propensity score. A balance diagnosis was performed by comparing the characteristics between matched groups using a method as described by Zhang et al.²² An SD greater than 0.1 was considered as a sign of imbalance, and the propensity score prediction model was refitted to ensure the balance between matched groups (Figure S1).

All analyses were performed using RStudio software, version 1.1.456 (R. RStudio, Inc., Boston, MA). A p-value <.05 identified statistical significance, and all confidence intervals used a 95% threshold.

3 ∣. RESULTS

Our study assessed 4 aspects of DCD LT over time: characteristics of DCD donors and LTs, characteristics of the recipients of DCD livers, DCD versus DBD LT outcomes, and machine-perfused LT outcomes. Overall, we analyzed 169 470 DBD donors, 107 685 DBD LTs, 20 779 DCD donors, and 5506 DCD LTs between January 1995 to December 2019.

3.1 ∣. Characteristics of DCD donors and liver transplants

The number of DBD donor livers increased from 5272 in 1995 to 9043 in 2019. During the same period, the total number of DCD donor livers increased from 64 to 2711, with the largest escalation of 27% occurring between 2018 (2132 DCD donors) and 2019 (2711 donors) (Figure 1A). Compared to DBD LT, the growth in number of DCD LTs has been more robust, especially in recent years. From 2009 to 2019, the number of DCD LTs performed per year increased by 147%, compared to only 32.3% for DBD LTs. In addition, between January 2018 to December 2019, the number of DCD LTs rose by 33.0%, compared to only 4.66% for DBD LTs over the same period. Finally, between 1995 to 2019, DCD comprised less than 10% of all LTs performed in the United States but the relative rise in number performed per year continued to outpace that of DBD LT (Figure 1B); the absolute number increase was much greater for DBD. The number of DCD donors that led to LT increased across all age ranges: <35 years old, between 35-55 years old, and >55 years old. Of note, the rate of increase for DCD LT donors <35 years and between 35-55 years was higher than that of DCD donors > 55 years old (Figure 1C). However, the number of deceased donor livers procured for transplant but not transplanted (the discard rate) increased by 24% from 2018 (707 liver allografts) to 2019 (875 liver allografts) (Figure 1D).

FIGURE 1.

Trends of DCD liver donors and LTs over time. (A) All deceased donors over time. (B) All deceased donor LTs over time. (C) Donor age in DCD LTs over time. (D) Deceased donor livers procured but not transplanted (DCD discard) over time

Among DCD donor allografts that were transplanted, the mean WIT from 2009 to 2019 was 23 minutes (Figure 2A). Of note, WIT was included in the STAR file only after March 1, 2008. The proportion of transplanted DCD donor allografts with WITs shorter than 20 min and between 20-30 min also rose steadily between 2008 and 2019 (Figure 2B). Mean CIT decreased in both DBD and DCD allografts from 9 hours in 1995 to less than 6 hours in 2019 (Figure 2C). The overall trend was an increasing percentage of DCD liver transplants that had fewer than 6 hours of CIT (Figure 2D).

FIGURE 2.

Trends in warm and cold ischemic times in LT. (A) Mean warm ischemic time in DCD LTs over time. (WIT was included in the STAR file only after March 1, 2008). (B) Warm ischemic time distribution (minutes) in DCD LTs. (C) Mean cold ischemic time in deceased donor LTs over time. (D) Cold ischemic time distribution (hours) in DCD LTs

3.2 ∣. Characteristics of DCD LT recipients

The mean age of DCD LT recipients increased over time, with the sharpest rise in the number of transplants occurring in the age group of individuals over 55 years old. Meanwhile, the absolute number of DCD LT recipients younger than 34 years old remained low from 2002 to 2019 (Figure 3A). The average model for end-stage liver disease (MELD) scores for DBD and DCD at time of LT (lab MELD) diverged after MELD-based allocation was implemented in 2002. From 2002-2003, both mean DCD and DBD lab MELD scores for LT recipients at time of transplant were below 20. However, from 2010 to 2019, the mean DCD liver recipient MELD score remained under 20, while mean DBD liver recipient MELD score rose to 23.4 in 2019 (Figure 3B). By 2019, the number of DCD LT recipients with MELD scores higher than 30 was only 30% (Figure 3C). Finally, the most common reasons for liver transplant among DCD LT recipients over the past decade were liver disease diagnosis (acute hepatic necrosis, cirrhosis, hepatitis), tumors (primary liver malignancies, liver metastasis), and biliary disease (primary biliary cirrhosis, primary sclerosing cholangitis). Among these, the largest increases in the indications for DCD LT were tumor diagnosis followed by liver disease (Figure 3D).

FIGURE 3.

Trends in DCD LT recipients. (A) Recipient age in DCD LT over time. (B) Mean lab MELD score of deceased donor liver recipients over time. (C) Mean lab MELD range of DCD LT recipients. (D) Most common diagnoses of DCD LT recipients. *Biliary disease: primary biliary cirrhosis, secondary biliary cirrhosis, primary sclerosing cholangitis, biliary atresia, biliary hypoplasia. Liver disease: acute hepatic necrosis, cirrhosis, alcoholic cirrhosis, hepatitis. Metabolic: familial cholestasis, metabolic disorders. Tumor: primary liver malignancy, secondary liver malignancy. Other: Crohn's disease, cystic fibrosis, total parentalnutrition/hyperalimentation, graft vs host disease, trauma

3.3 ∣. DCD versus DBD LT outcomes

Table 1 shows the comparisons between 107 685 DBD LTs and 5506 DCD LTs between January 1995 and December 2019. These were all adult (>18 years old), first-time, liver-only, whole-organ transplants from deceased donors. Our analysis revealed that compared to DBD donors, DCD donors were younger (mean 34 years vs. 42 years, p-value <.001) and more likely male (67.3% vs. 59.2%, p-value <.001), but had similar BMI (25.9 kg/m² vs. 26 kg/m²). DCD donors were also more likely to be white (80.8% vs. 67.9%, p-value <.001) and had lower rates of smoking, hypertension, diabetes, HCV, and myocardial infarction compared to DBD donors.

TABLE 1.

DCD vs DBD donor and recipient characteristics

Characteristic	DBD (n = 107 685)	DCD (n = 5506)	p-value
Donors
Age (median, [IQR]) in years	42.0 [26.0, 55.0]	34.0 [24.0, 47.0]	<.001
Gender (n, % male)	63 765 (59.2)	3706 (67.3)	<.001
BMI (median, IQR) in kg/m2	26.0 [22.8, 30.1]	25.9 [22.6, 30.1]	.04
Ethnicity (n, %)
White	73 160 (67.9)	4447 (80.8)	<.001
African American	17 361 (16.1)	487 (8.8)
Hispanic	13 107 (12.2)	434 (7.9)
Other	4057 (3.8)	138 (2.5)
Cause of death (n, %)
Anoxia	23 718 (22.0)	2372 (43.1)	<.001
Cerebrovascular / Stroke	43 036 (40.0)	974 (17.7)
Head Trauma	38 264 (35.5)	1918 (34.8)
Other	2667 (2.5)	242 (4.4)
History of smoking (n, %)
No smoking history	74 093 (68.8)	4332 (78.7)	<.001
Smoking but not continued	6008 (5.6)	177 (3.2)
Continued smoking	25 892 (24.0)	947 (17.2)
Unknown	1692 (1.6)	50 (0.9)
History of hypertension (n, %)
No	70 763 (65.7)	4288 (77.9)	<.001
Yes	36 007 (33.4)	1188 (21.6)
Unknown	915 (0.8)	30 (0.5)
History of diabetes (n, %)
No	95 886 (89.0)	5113 (92.9)	<.001
Yes	11 123 (10.3)	375 (6.8)
Unknown	676 (0.6)	18 (0.3)
History of myocardial infarction (n, %)
No	94 034 (87.3)	5062 (91.9)	<.001
Yes	3473 (3.2)	142 (2.6)
Unknown	10 178 (9.5)	302 (5.5)
Hepatitis C Status (n, %)
Negative	102 576 (95.3)	5340 (97.0)	<.001
Positive	4854 (4.5)	154 (2.8)
Unknown	255 (0.2)	12 (0.2)
Recipients
Age (median, [IQR]) in years	55.0 [48.0, 61.0]	57.0 [51.0, 63.0]	<.001
Gender (n, % male)	71 257 (66.2)	3823 (69.4)	<.001
BMI (median, IQR) in kg/m2	28.0 [24.6, 32.2]	28.2 [24.9, 32.2]	.03
Ethnicity (n, %)
White	78 742 (73.1)	4150 (75.4)	.002
African American	9044 (8.4)	407 (7.4)
Hispanic	13 946 (13.0)	671 (12.2)
Other	5953 (5.5)	278 (5.0)
Blood Type (n, %)
A	40 607 (37.7)	2182 (39.6)	<.001
B	14 628 (13.6)	559 (10.2)
AB	5583 (5.2)	162 (2.9)
O	46 867 (43.5)	2603 (47.3)
History of diabetes (n, %)
No	81 167 (75.4)	4006 (72.8)	<.001
Yes	24 385 (22.6)	1449 (26.3)
Unknown	2133 (2.0)	51 (0.9)
Hepatitis C Status (n, %)
Negative	59 078 (54.9)	3175 (57.7)	<.001
Positive	39 451 (36.6)	2021 (36.7)
Unknown	9156 (8.5)	310 (5.6)
Cold Ischemic Time (median, [IQR]) in hours	6.5 [5.0, 8.5]	5.8 [4.7, 7.2]	<.001

^*Significant difference, p-value <.05.

Recipients of DCD donors were on average older (mean 57 years vs. 55 years, p-value <.001), more likely male (69.4% vs. 66.2%, p-value <.001), and had similar BMIs (28 kg/m² vs. 28.2 kg/m²) compared to recipients of DBD livers. Recipients of DCD donor livers were also more likely white, had lower rates of diabetes, hepatitis C, and CITs (mean 5.8 hours vs. 6.5 hours, p-value <.001), and were 2.85% more likely to have a diagnosis of hepatocellular carcinoma at time of LT (3.72% in DBD versus 6.57% in DCD, p-value <.001).

Outcomes comparing DCD to DBD LTs demonstrated that mean length of stay after LT was over 5 days longer for DCD LT recipients compared to DBD recipients in 2003, but after 2009, DCD LT recipients had a shorter average length of stay after surgery (Figure 4A). Of note, mean lab MELD scores of DCD LT recipients were 1.3 points lower than those of DBD LT recipients at this time. However, DCD LTs had a higher rate of retransplant, although these rates did converge from 2005 to 2019 (Figure 4B). In 2009, the difference in retransplant rates between DCD and DBD LTs was 7.5%; by 2019, this gap had decreased to just 1.2%. Just as promising, the percent of DCD LT recipients who had biliary complications (such as IC) causing graft failure severely decreased from 9.1% in 2002 to 0.92% in 2019. By 2019, the difference in biliary complications causing graft failure in DCD versus DBD allografts was just 0.77% (Figure 4C).

FIGURE 4.

Convergence in DCD vs. DBD (A) hospital length of stay after LT, (B) percent of deceased donor LT recipients requiring retransplant (*no recorded DCD LTs in 1998), and (C) percent of deceased donor liver transplant recipients who had biliary complications (BC) leading to graft failure (GF). (Biliary complications data limited in the STAR file prior to 2002)

Kaplan-Meier survival analysis of DCD versus DBD livers revealed that DBD recipients had statistically higher PS (p-value = .027) (Figure 5A). The 1-, 3-, 5-, and 10-year PS was 89.8%, 82.2%, 76.1%, and 60.9% after DBD LT, compared to 89.5%, 81.3%, 74.9%, and 58.2% after DCD LT. Overall GS revealed analogous results, with superior DBD GS over time (p-value <.001) (Figure 5B):The 1-,3-, 5-, and 10-year GS was 86.7%, 78.5%, 72.4%, and 58.1% for DBD allografts and 83.7%, 74.2%, 67.1%, and 52.0% for DCD allografts. When these results were stratified into LTs conducted between 1995-2008 and 2009-2019, DCD LT 1-, 3-, 5-, and 10-year PS increased by 5.3%, 6.5%, 6.0%, and 7.3%, respectively, in the 2009-2019 group. GS also showed significant improvements between the two groups, with 1-, 3-, 5-, and 10-year GS improving by 10.8%, 12.7%, 12.9%, and 11.7%, respectively, in the 2009-2019 group. Propensity score matched data comparing DCD and DBD patient and graft survival had comparable outcomes, with DBD having higher PS (p-value < 0.001) and death-censored graft survival (DCGS) (p-value <.001) (Figure S2). After adjusting for baseline confounders (age, sex, race, BMI, cause of death, history of smoking, diabetes, hypertension, previous myocardial infarction, CIT, and WIT), the matched survival analysis showed DCD LTs had a PS hazard ratio of 1.19 (95% CI 1.10, 1.28) compared to DBD LTs. DCD allografts had a GS hazard ratio of 1.06 (95% CI 1.01, 1.12) compared to DBD allografts (Figure 5C).

FIGURE 5.

DCD vs. DBD LT outcomes. (A) Kaplan-Meier curve of DCD vs. DBD LT survival, p-value = 0.027. (B) Kaplan-Meier curve of DCD vs. DBD LT graft survival, p-value <0.001. (C) forest plot of the patient survival (PS) and graft survival (GS) in the propensity score matched cohorts

Finally, comparing improvements in DCD LT in the era of MELD implementation (2002 to the present) revealed significant Pearson correlation coefficients between LT metrics and DCD LT outcomes (PS, GS). Strong negative associations were found between CIT and 1-year PS (R= −0.7, p-value = .0012) and 1-year GS (R= −0.78, p-value <.001). A strong negative associations was also seen between retransplantation rate and 1-year PS (R= −0.84, p-value <.001) (Figure S3).

3.4 ∣. Machine-perfused liver transplant outcomes

There was an increase in both DCD and DBD machine-perfused transplanted livers from 2016 to 2019 (from 16 total MP livers in 2016 to 99 in 2019). Kaplan-Meier curves depicting PS with machine-perfused vs. non-machine-perfused livers showed no difference in overall survival (p-value = .53) (Figure 6A). However, machine-perfused livers that were transplanted did exhibit lower GS compared to transplanted livers without machine perfusion (p-value <.001) (Figure 6B). Further stratification showed that machine-perfused transplanted livers (at normothermic, subnormothermic, and hypothermic temperatures) had longer mean CITs in both DCD allografts (7.15 hours versus 5.46 hours) and DBD allografts (7.26 hours versus 5.94 hours). Machine-perfused transplanted livers did have comparable recipient mean lab MELD scores (18.1 versus 19.1 for DCD and 21.0 versus 23.3 for DBD) and mean WITs (22.5 minutes for DCD versus 22.8 minutes for DBD) compared to transplanted livers without machine perfusion (Figure 6C).

FIGURE 6.

Machine perfusion effect on DCD and DBD LT outcomes. (A) Kaplan-Meier curve of machine perfused vs non-machine perfused LT patient survival, p-value = 0.53 and (B) graft survival, p-value <0.001. (C)Comparisons of mean lab MELD, CIT, and WIT between machine perfused and non-machine perfused liver allografts

4 ∣. DISCUSSION

This analysis characterizes the rapidly changing landscape in DCD LT as clinical experience with DCD livers has increased, especially over the past two years. The adoption of expanded-use protocols combined with the growing worldwide liver organ shortage has likely stimulated the increased the number of DCD donors¹²; this is particularly evident in the most recent years in our analysis, 2018 and 2019 (Figure 1A). Increased clinical experience with DCD LT in the United States has led to reduction in mean CIT, mean post-LT hospital length of stay, and lower rates of retransplant. These changes are consistent with the factors that lead to improved DCD LT outcomes seen in the United Kingdom DCD Risk Score.²³ Improved clinical outcomes with DCD LTs are likely due to better patient and organ selection, but there remains tremendous potential for the additional utilization of DCD livers in the United States. In 2019 alone, 306 DCD livers were recovered for LT but not transplanted (discarded), 428 were recovered for other purposes (mostly for research), and 1211 DCD livers were not recovered at all. Thus, in total, 1945 potential DCD organs were not transplanted into patients on the LT waitlist in 2019.

Our results indicate that while there has been an increase in the number of DCD donors across all age groups, this has been matched with a higher organ discard rate. In 2019, there were 7453 DBD LTs with 569 discarded allografts (7.1% DBD discard rate) versus 714 DCD LTs with 306 discarded allografts (30% DCD discard rate) (Figure 1D). However, the relative rise in number of LTs performed from 2018 to 2019 continues to be higher for DCD LTs (33.0% increase) versus DBD (4.66% increase) (Figure 1B). 58.5% of these discards (179 organs) meet standard DCD acceptance criteria proposed by ASTS guidelines of WIT <30 min, donor age <55 years, and CIT of <8 hours.²⁴ There is also a much larger pool of DCD organs, from donors over 55 years of age that are that are not considered for transplant, as well as livers from younger donors that are discarded. While some of these DCD livers are deemed unusable for reasons such as poor flush, prolonged WIT, poor biopsy, or procurement issues, finding ways to transplant the high-quality DCD allografts safely and with acceptable outcomes could have tremendous impact on the liver organ shortage.²⁵

One recent trend that may help in this regard is use of physiologic events such as SpO₂ falling below 70% or blood pressure <50mmHg following withdrawal of life-sustaining therapy in controlled DCD procurement as the start of functional donor warm ischemia time.²⁶ This method might allow procurement teams to better gauge organ ischemia compared to the traditional measures of reliance on simple time from withdrawal of support to initiating organ flush with preservation solution.

The focus on WIT is critical as a considerable body of literature has linked WIT, defined as the time from cessation of cardiopulmonary support to in situ cold preservation, as a metric by which to estimate risk and to decide whether the organ is suitable for transplant versus discard.²⁷ In 2012, an analysis by the Mayo Clinic group showed that every minute added to WIT from asystole to cross-clamp increased the chances of IC by 16.1%.²⁸ The concern of IC and biliary complications in DCD donors related to prolonged WIT²⁹ has led to ASTS recommendations limiting WIT to 30-45 min.²⁴ Our analysis indicates that in 2019, less than 10% of transplanted DCD donor livers had WITs >30 min (Figure 2B). The median DCD donor age was 45 years for those with WITs >30 min and 41 years for those with WITs <30 min. Furthermore, the mean WIT of DCD donor liver allografts in 2019 was 23 minutes (Figure 2A).

Second, CIT has a disproportionately greater negative impact on DCD LT recipients compared to their DBD counterparts. Paterno et al. noted that longer CIT was associated with longer post-transplant hospital stay, higher rate of PNF, and hyperbilirubinemia.³⁰ An analysis by Sher et al. also revealed that minimizing DCD CIT to less than 6 hours increased 3-year liver allograft survival rates by 4%, perhaps explaining the trend we observed favoring shorter CIT in recent years.³¹ Our results indicate that mean CITs for DCD liver allografts have been less than 6 hours since 2011 (Figure 2C). In 2019, over 66% of DCD LTs had CITs less than 6 hours and less than 1% had CITs over 12 hours (Figure 2D). Thus, perhaps one of the most effective improvements to DCD liver transplantation has been the reduction of CITs. This is further supported by the strong negative association seen between CIT and 1-year DCD PS and GS (Figure S3).

The observation that DCD organs are preferentially transplanted to patients with lower MELD scores at transplant compared to DBD recipients has been regarded as another route to optimize DCD utilization and outcome.³² Studies find that transplanting patients with lab MELD scores <20 prolonged GS, suggesting that DCD organs compromised by an ischemic insult function better in less critically ill recipients.³³ Evidence also recommends that even for patients with severe liver disease, PS is enhanced by accepting a DCD liver compared to remaining on the waitlist for a DBD liver in the future.³⁴ Since MELD tracking was implemented in 2002, mean recipient lab MELD scores at the time of LT have remained between 19 and 20 while those for DBD recipients have increased overall with time (Figure 3B). In 2019, 40% of DCD LT recipients had lab MELD scores less than 20, and only 30% had higher than 30 (Figure 3C). This trend can be verified as the biliary disease became a smaller percentage of indications for DCD LT, while tumor diagnosis became a more common reason to use DCD livers (Figure 3D).

The reasons behind preferential placement of DCD livers in low MELD patients are not entirely clear. One potential rationale is that patients critically ill with liver failure may not tolerate the hemodynamic instability caused by severe IR injury that may accompany DCD liver reperfusion. In addition, DCD allografts transplanted to critically ill recipients exhibit high rates of PNF.³⁵ Thus, just as some critically ill recipients do not tolerate marginal DCD organs such as DCD, some DCD livers do not function well in compromised hosts.

Historic reasoning suggests that compared to a DBD LT recipient, a DCD recipient should be able to tolerate a higher chance of graft non-function but also should stand to benefit from receiving the organ.^13,36 Conversely, placing DCD liver allografts in low-risk recipients or young patients (<34 years old) potentially compromises life years gained.³⁵ This rationale is reflected in clinical practice, as our results show that DCD recipients are older on average by 2 years but also healthier (lower MELD at time of transplant, lower rates of diabetes, less likely to be hepatitis C positive, and shorter length of hospital stay after LT).

Lower patient and graft survival of DCD LTs compared to DBD LTs is also cited as contributing to lower DCD utilization rates. However, as clinical experience with DCD LT and donor and recipient selection has been refined, the rates of retransplantation between DCD and DBD organs have converged, with both being less than 5% at one year in 2019 (Figure 4B). This improvement in retransplantation rate was shown to be strongly associated with improved 1-year DCD PS and GS (Figure S3). A possible explanation for this trend is improved selectivity of DCD organs based greater clinical experience with DCD and the ASTS recommendations in 2009 of WIT <30-45 min and CIT <8 hours, leading to lower rates of biliary complications causing graft failure seen in Figure 4C.²⁴ Also, while our patient and graft survival analysis showed statistically significant differences between DBD and DCD LTs (Figure 5A, 5B), the divergence occurred at 15 years post-transplant. Collectively, these data suggest that clinical judgement in effective utilization of DCD livers for transplants has improved. In 2019, the mean CIT for DCD livers was 5.6 hours, the mean WIT was 23 minutes, and the mean DCD donor age was 45 years, all well within the ASTS DCD LT recommendations.²⁴

Finally, ex vivo machine perfusion is currently transforming the use of marginal DCD liver allografts.¹⁶ The experience with transplanting machine-perfused livers for the past 4 years in the United States shows equivalent rates of PS, but lower rates of GS compared to transplanting non-machine-perfused livers (Figure 6A, 6B). However, lower GS is confounded by the fact that marginal organs with longer CITs are preferentially selected for machine perfusion, while high-quality liver allografts are transplanted immediately (Figure 6C). Additionally, most of the non-perfused livers were not in clinical trials applying strict entry criteria and eliminating the sickest recipients. Thus, a multivariable analysis is needed to adjust for differences in the recipient population to draw further conclusions comparing machine-perfused and non-perfused groups.

Nevertheless, normothermic machine perfusion (NMP) provides an enticing opportunity to recover marginal livers ex vivo from warm ischemic damage incurred during procurement by providing oxygen and nutrition at physiologic temperatures.²⁵ Furthermore, NMP allows assessment of viability of a marginal DCD allograft without risk to the recipient. Currently, there are several clinical trials assessing the safety and effectiveness of NMP in resuscitating and evaluating marginal human liver allografts for transplantation. These studies show a reduction in ischemia-reperfusion injury and recipient hemodynamic instability, which could significant expand the recipient pool for DCD livers.^17,37-39 As liver perfusion machines are incorporated into clinical practice, another avenue of investigation is the problem of IC in DCD allografts, a major cause of graft failure and need for retransplant. Fibrinolytic agents have been utilized during procurement and have been found to reduce thrombosis in the peribiliary plexus^40,41; however, this comes at the price of an increased risk of significant bleeding in a procedure already prone to coagulopathy.⁴² Ex vivo machine perfusion provides the opportunity to treat the DCD liver allograft with fibrinolytic agents prior to transplantation with minimal risk of bleeding in the recipient. Trials testing this approach are currently underway.

The opportunity to expand the LT organ pool with DCD organs comes with the challenge of tackling a wide range of practices and attitudes toward DCD LT.³¹ A clear manifestation of this variation is that the majority of the DCD LT experience is clustered at a few LT centers. Hobeika et al. showed that only 3 centers in the United States performed over 100 DCD LTs during a 5-year study period between January 2013 to December 2017, and this comprised 17.5% of all DCD LTs in the United States. The authors conjecture that understanding high-utilization center practices and their relationships with Organ Procurement Organizations could help minimize the variation in DCD LT volume and practice.⁴³ It is also possible that with broader sharing of DCD livers, the greater regional competition for organs will increase DCD use in those areas.

This work has several limitations. First, liver transplantation has evolved over the 25 years we analyzed, with advances in immunosuppression, surgical technique, and post-operative care. Thus, overall trends need to be interpreted with potential confounding due to clinical innovation. Second, the OPTN database used for this study has inherent limitations due to missing data, specifically for complications such as IC and biliary complications. For this reason, the comparison between DCD and DBD IC rates over time was not possible. Finally, machine-perfused livers that subsequently were transplanted were recorded in the OPTN database starting in 2016 and this comprised a relatively small data set. Comparing Kaplan-Meier survival curves between these two groups may leave conclusions underpowered. However, as ex vivo normothermic machine perfusion continues to be incorporated into clinical liver transplant practice, more data assessing the risks and benefits of transplanting machine-perfused organs will be available.

In conclusion, our study reveals that increased clinical experience with DCD over time has translated into increased utilization, shorter CITs, older yet healthier DCD LT recipients, improved patient selection, lower length of hospitalization after transplant, and lower rates of retransplantation from 1995 to 2019. Over the past decade, the growth in number of DCD LTs conducted in the United States has outpaced that of DBD on a percentage basis, particularly in recent years. These trends, together with emerging application of machine perfusion technology, raise hope that we will be soon able to capitalize on the large number of DCD organs currently being discarded and allow their consideration for transplantation.

DATA AVAILABILITY STATEMENT

The authors declare that the data supporting the findings of this study are available and, if needed, will be provided upon request.