This study used data from the Organ Procurement and Transplantation Network/United Network for Organ Sharing (OPTN/UNOS) Standard Transplant and Research (STAR) files for LT. The study period was set from 1 January 2014 to 31 December 2019, with 1 year of post-transplant observation for each patient. Study procedures were approved by the Henry Ford Health System Institutional Review Board; the requirement for written informed consent was waived due to the deidentified nature of the data. Patients who were 18 years or older at the time of transplant were eligible for this study. Patients who received a partial/split graft or combined organ transplant with thoracic organs, kidney, intestine, and/or pancreas or patients for whom donor age was unknown were excluded. Also, if patients who had one or more missing data which was evaluated in this study, those were excluded (Figure 1). To assess the impact of donor age on post-LT outcomes and to determine whether specific recipient characteristics were associated with worse post-LT outcomes with liver grafts from older donors, the cohort was divided into three groups according to the donor age. Age categories were determined using the liver donor risk index [3]: older donor (≥70 years); middle-aged donor (40–69 years); and younger donor (<40 years).

Binary variables included: recipient gender; recipient diabetes; primary liver disease etiologies including HCV infection; alcohol related disease, non-alcoholic steatohepatitis and other diseases, diagnosis of HCC; history of abdominal surgery; previous liver transplant; registration as status 1; dialysis requirement at transplant; mechanical ventilation at transplant; portal thrombosis at transplant; donation after circulatory death (DCD); donor diabetes; donor history of heavy alcohol use; donor history of hypertension; and donor history of myocardial infarction. In the risk factor analysis, model for end-stage liver disease (MELD) score or MELD-sodium score was not included. Instead, 4 parameters of MELD-sodium score (serum total bilirubin, creatinine, sodium, and INR) were separately included. MELD-sodium score was calculated using the following formula; MELD-sodium = MELD +1.32 x (137-serum sodium)—[0.033 x MELD x (137 - serum sodium)] [11]. Continuous variables were classified into multilevel categorical variables. Recipient data at time of transplant included: age (<50, 50–64, and ≥65 years); BMI (<18.5, 18.5–24.9, 25.0–29.9, and ≥30.0 kg/m²) [12]; serum bilirubin (<2.0, [2.0–4.4, 4.5–11.9, and ≥12.0 mg/dl [total bilirubin of 2.0 mg/dl: based Child-Pugh score [13], 4.5 and 12 mg/dl were 33 and 66%tile in the cohort]); serum creatinine (<1.5, 1.5–1.7, 1.8–2.5, and ≥2.5 mg/dl [creatinine of 1.5 mg/dl is used for a diagnosis hepatorenal syndrome criteria in patients with cirrhosis] [14], 1.8 and 2.5 mg/dl were 33 and 66%tile in the cohort]), serum sodium (<125, 125–134, 135–145, and ≥146 mEq/L) [15]; and international normalized ratio (INR; <1.5, 1.5–1.7, 1.8–2.4, and ≥2.5 [INR ≥1.5; a factor of acute liver failure] [16], 1.8 and 2.5 were 33 and 66%tile in the cohort]). Organ related variables included donor age at transplantation (<40, 40–69, and ≥70 years old) and cold ischemia time (<6.0, 6.0–7.9, and ≥8 h [6 h was median value in the cohort, 8 hour-cut off point was decided according to liver donor risk index] [3]). Additional multilevel categorical variables included: recipient race (White, Black/African American, Hispanic [of any race], and other); Karnofsky Performance Status score (10–30, 40–60, and 70–100%); donor cause of death (trauma, anoxia, cerebrovascular accident [CVA], and other); and organ share type (local, regional, or national). All covariates were collected prior to or at the time of LT.

Risk of one-year graft loss after receipt of an organ from the ≥70 donor group was compared to the 40–69 and <40 donor groups. Graft loss was defined as death or re-transplantation. Analyses were adjusted for recipient demographic (age, race, gender) and clinical characteristics (BMI, diabetes, primary liver disease etiologies including HCV infection; alcohol related disease, non-alcoholic steatohepatitis and other diseases, presence of HCC, history of abdominal surgery, portal thrombus, previous liver transplant, status 1 [yes/no], laboratory values [bilirubin, creatinine, INR, sodium], Karnofsky score, and need for mechanical ventilation or dialysis) at the time of transplantation. Analyses were also adjusted for donor (age category, race, gender, BMI, diabetes, history of heavy alcohol use, history of hypertension, and history of myocardial infarction) and organ characteristics (cause of death, donation after cardiac death [DCD; yes/no], cold ischemia time, and organ share type).

Recipient risk factors for one-year graft loss were determined with multivariable Cox regression. The total risk score for each patient was calculated by multiplying the adjusted hazard ratios (aHRs) of recipient risk factors according to a previously used methodology [17]. If one risk factor, score is equal to the HR of that particular factor. If no risk factor, score is zero. Our risk stratification system classified recipients into low- and high-risk groups; the cut-off risk score value was calculated from Youden index and determined by the receiver operating characteristic (ROC) curve for one-year graft survival. We then compared one-year graft loss in the ≥70 donor group to these risk score categories. We also compared the low- and high-risk ≥70 donor groups to the 40–69 and <40 groups (both with and without DCD). This risk stratification system was then applied to the validation cohort using patient cohort who received LT in 2020 from the STAR files (Figure 2).

Patient demographic and clinical characteristics, as well as donor and organ characteristics, were described by donor age groups, using median and interquartile range (IQR) for continuous variables and number and percentages for categorical variables. We used the Mann-Whitney-U test for continuous variables and chi-square test for categorical variables to study differences in patient characteristics among the three donor age groups. Post-transplant graft survival was evaluated using Kaplan-Meier curve analysis and compared by log-rank tests. A multivariable Cox regression model assessed hazards of post-transplant graft loss. For the risk factor analysis in each donor group (older donor, middle-aged donor, and younger donor), multivariable Cox regression models were created using factors which had p value less than 0.157 in univariable analyses [18]. p-values <0.05 were considered statistically significant for all analyses. All statistical analyses were completed using SPSS version 27 (IBM, Chicago IL, United States) and R version 3.5.1 (R Foundation for Statistical Computing, Vienna, Austria).

Of the 31,290 patients eligible for this study, 1,225 received livers from donors in the ≥70 group, 15,445 received livers from donors aged 40–69, and 14,620 received livers from donors <40 years old (Figure 1). Table 1 showed details characteristics of patients from the three donor age groups. Recipients of organs from older donors were themselves significantly older (median age 62 vs. 58 [donors aged 40–69] and 57 [donors <40 years], p < 0.001 for both).

Median recipient MELD (MELD-sodium) score was significantly lower in older donor group (18 vs. 22 [donors aged 40–69] and 24 [donors <40 years], p < 0.001 for both). More of recipients from older donors had HCC compared to recipients with younger donors (17.6% vs. 14.0% [40–69] and 11.6% [<40], p = 0.001 and p < 0.001, respectively) but fewer had HCV (18.9% vs. 21.8% and 22.0%, p = 0.01 for both). Recipients of organs from donors ≥70 were less likely to have Karnofsky scores of 10–30% (14.7% vs. 27.2% [40–69] and 34.0% [<40], p < 0.001 for both), to have previously received a liver transplant (1.1% vs. 3.1% [40–69] and 5.6% [<40], p < 0.001 for both), or to be registered as status 1 (1.0% vs. 1.7% [40–69] and 3.0% [<40], p = 0.06 and p < 0.001 respectively), were also more likely to have recipient diabetes (32.4% vs 29.7% [40–69] and 26.7% [<40], p = 0.051 and p < 0.001, respectively). Organs from the older donor group were more likely to have <6 h cold ischemia time than from other age groups (60.3% vs. 56.0% [40–69] and 53.6% [<40], p < 0.001 for both), to be allocated from a national organ share (11.8% vs. 5.3% [40–69] and 3.6% [<40], p < 0.001 for both), to have donor diabetes (30.0% vs 20.0% [40–69] and 4.1% [<40], p < 0.001 for both), history of hypertension (75.5% vs 57.0% [40–69] and 12.4% [<40], p < 0.001 for both), history of myocardial infarction (11.8% vs 6.8% [40–69] and 1.3% [<40], p < 0.001 for both), and to have liver biopsy (70.7% vs 52.0% [40–69] and 26.5% [<40], p < 0.001 for both), but those were less likely to have history of heavy alcohol use (8.7% vs 20.7% [40–69] and 13.2% [<40], p < 0.001 for both). There were no cases of donation after cardiac death (DCD) among recipients of organs from the ≥70 donor group (Table 1).

The adjusted risk of one-year graft loss was significantly higher among recipients of organs from donors aged ≥70 years than from donors aged 40–69 years (aHR 1.30, 95%CI 1.09–1.56, p = 0.004) and aged <40 years (aHR 1.39, 95%CI 1.15–1.69, p < 0.001; Table 2).

Demographic comparisons between the derivation and validation cohorts are shown in the Table 3. In the derivation dataset, the following recipient characteristics were associated with significantly increased risk of graft loss: previous liver transplant (aHR 4.37, 95%CI 1.98–9.65, p < 0.001); need for mechanical ventilation (aHR 4.28, 95%CI 1.95–9.43, p < 0.001); portal thrombus (aHR 1.87, 95%CI 1.26–2.77, p = 0.001); serum sodium <125mEq/L (aHR 2.88, 95%CI 1.34–6.20, p = 0.007); Karnofsky score between 10 and 30% (aHR 2.03, 95%CI 1.13–3.65, p = 0.01), between 40%–60% (aHR 1.65, 95%CI 1.08–2.51, p = 0.02; Table 4). HCV status did not increase risk of graft loss in the older donor group. Using these results, a risk stratification system was created using same (derivation) dataset by multiplying the aHRs of the significant risk factors (Table 5). Based on ROC curve analysis (Supplementary Figure S1), a risk score cut-off value of 2.03 was used to divide patients into low-risk (<2.03; n = 931) and high-risk groups (≥2.03; n = 294).

Although patient age was not a significant risk factor for graft loss among recipients in the older donor group, age ≥65 years was a significant risk factor for graft loss among recipients who received organs from the middle-aged or younger donors (aHR 1.19, 95%CI 1.01–1.40 and aHR 1.71, 95%CI 1.44–2.04; p = 0.04 and p < 0.001, respectively). Recipient age of 50–64%years was also a significant risk factor for graft loss in the younger donor group (aHR 1.27, 95%CI 1.09–1.47, p = 0.001; Supplementary Tables S1, S2).

One-year graft survival rate was significantly higher in low-risk recipients than in high-risk recipients (91.7% [low-risk] vs. 78.0% [high-risk], p < 0.001) (Supplementary Figure S2A). Three- and 5-year graft survival rate were also significantly higher in low-risk recipients than in high-risk recipients (3-year; 82.5% [low-risk] vs. 70.5% [high-risk], p < 0.001, 5-year; 76.0% [low-risk] vs. 64.1% [high-risk], p < 0.001) (Supplementary Figures S2B,C). One-year graft survival rate in low-risk recipients of the older donor group was similar to the younger or middle-aged donor group (Figure 2A). After excluding DCD cases from the middle-aged and younger donor groups (for consistency with the older donor group, in which there was no DCD donors), similar trends were observed (Figure 2B).

The adjusted risk of one-year graft loss in the low-risk older donor group was similar to that of the younger donor group (aHR 1.08, 95% CI 0.85–1.38, p = 0.53; Figure 2C ). In contrast, the adjusted risk of one-year graft loss in the high-risk recipients was significantly higher than in the younger donor group (aHR 2.00, 95% CI 1.53–2.61, p < 0.001). While the adjusted risk of one-year graft loss in the high-risk recipients was also significantly higher compared to the middle-aged donor group (aHR 1.84, 95% CI 1.42–2.38, p < 0.001), those in the low-risk older donor group was similar to that of the middle-aged donor group (aHR 0.99, 95% CI 0.78–1.26, p = 0.95; Figure 2D).

Among the validation cohort, one-year graft survival rate in low-risk recipients of the older donor group was similar to those in the younger or middle-aged donor group (Figure 3A). After excluding DCD cases from the younger and middle-aged donor groups, the one-year graft survival rate in low-risk recipients of the older donor group was similar to those in the younger or middle-aged donor group (Figure 3B).

The adjusted risk of one-year graft loss was similar between low-risk older donor recipients and younger donor recipients (aHR 0.95, 95% CI 0.56–1.62, p = 0.86; Figure 3C), but was significantly higher for high-risk recipients (aHR 2.00, 95% CI 1.02–3.92, p = 0.04). While the adjusted risk of one-year graft loss in the high-risk recipients was also significantly higher compared to the middle-aged donor group (aHR 2.14, 95% CI 1.11–4.12, p = 0.02), those in the low-risk older donor group was similar to that of the middle-aged donor group (aHR 1.00, 95% CI 0.60–1.66, p = 0.97; Figure 3D).