Gender Differences in Liver Donor Quality Are Predictive of Graft Loss


  • Grant support: this project was supported by the National Institute of Diabetes and Digestive and Kidney Diseases (T32 DK060414, JCL) and the University of California San Francisco Liver Center (JCL, NAT).

Corresponding author: Norah A. Terrault,


Some studies have found that donor–recipient gender mismatch predicts posttransplant outcomes but whether this is independent of donor quality is unknown. To evaluate the association between gender mismatch and graft loss, 11 508 females (F) and 16 714 males (M) who underwent liver transplant from March 1, 2002 to December 31, 2007 were studied. Of 11 donor characteristics, clinically relevant differences between F and M donors were median age (47 vs. 39 years), height (165 vs. 178 cm) and proportion dying of stroke (59 vs. 35%) (p < 0.001 for all). The donor risk index was significantly lower for F than M donors (1.3 vs. 1.6, p < 0.001). Recipients of gender-mismatched grafts had an 11% higher risk of graft loss (p < 0.001). Compared to M→M donor–recipient-matched transplants in univariable analysis, F→M mismatch was associated with a 17% increased risk of graft loss (95% CI = 1.11–1.24, p < 0.001), whereas M→F mismatch was not (HR = 1.02; 95% CI = 0.96–1.09; p = 0.46). However, adjustment for significant recipient and donor factors eliminated the association between F→M mismatch and graft loss (HR = 0.95; 95% CI = 0.89–1.02; p = 0.18). In conclusion, donor quality differs significantly between female and male donors—female donors are older, shorter and die more frequently of stroke—and gender differences in donor quality, rather than gender mismatch are predictive of graft loss.


body mass index


donor risk index


hepatitis B virus


hepatitis C virus


interquartile range


model for end-stage liver disease


non-alcoholic fatty liver disease


Gender plays an important role in the outcomes of liver transplant recipients (1–3). Some studies suggest that donor gender influences outcomes after liver transplantation, although this association is controversial. A large international cohort of over 16 000 liver transplant recipients found that donor gender, per se, had no significant effect on graft loss (4), whereas three separate United States-based studies reported that donor gender—in the form of donor–recipient gender mismatch—was associated with graft failure (5–7). Specifically, in one study evaluating nearly 14 000 gender-mismatched liver transplants, male recipients of female liver grafts had a 20% increased risk of graft loss compared to gender-matched male recipients (6). Another study confirmed these findings in nonhepatitis C virus (HCV) liver transplant recipients and also found that female HCV recipients of female liver grafts experienced the lowest graft survival rates among all the donor–recipient gender pairings (5). However, all of these studies included patients from the earlier transplant experience dating back to 1987, and the analyses were not adjusted for other transplant-related and donor factors that are now widely recognized to be associated with graft loss.

In recent years, the importance of donor quality in posttransplant outcomes has been emphasized. Feng et al. developed the donor risk index (DRI), a composite score consisting of seven donor characteristics (age, African American race, height, split liver, donation after cardiac death and death from cerebrovascular accident or other causes), to assist transplant clinicians with predicting the risk of graft loss associated with a specific liver donor (8). Notably, donor gender did not emerge as a factor in the DRI, raising the question of whether donor gender or donor–recipient gender mismatch truly plays a role in posttransplant outcomes. We hypothesized that gender differences in donor quality rather than donor–recipient gender mismatch were relevant in predicting graft outcomes and tested this hypothesis in a national cohort of liver transplant recipients.


Study population

This retrospective cohort study included all patients ≥18 years of age who underwent primary, single-organ liver transplantation in the United States from March 1, 2002 through December 31, 2007. Data regarding recipient and donor characteristics were obtained from the UNOS Standard Transplant Analysis and Research (STAR) database as of March 1, 2010.

Patients of all races were included in this study but the race covariate in our analyses compared African American to non-African American. Etiologies of liver disease were grouped as follows: alcohol, HCV, hepatitis B virus (HBV), nonalcoholic fatty liver disease (NAFLD, including cryptogenic and nonalcoholic steatohepatitis), autoimmune (including autoimmune hepatitis, primary biliary cirrhosis and primary sclerosing cholangitis), acute (listed as status 1, acute hepatic necrosis or fulminant hepatic failure) and other (including alpha-1-antitrypsin deficiency, Budd-Chiari, hemochromatosis and others). Patients who were listed with HCV in addition to other diagnoses were included under a listing diagnosis of HCV. Patients who had a listing diagnosis of hepatocellular carcinoma were included in the cohort under their primary etiology of liver disease, although concurrent diagnosis of hepatocellular carcinoma was evaluated in the multivariable models as a covariable. The MELD score was calculated using the standard formula (9):

MELD = 3.8 * ln(bilirubin [mg/dL]) + 11.2 * ln(INR) + 9.6 * ln (creatinine [mg/dL]). For the calculation of MELD at transplant, a lower limit of 1 was set for all variables, and an upper limit of 40 was set for the MELD score. Body mass index (BMI) was calculated using the formula: weight (kilograms)/height (meters)2.

Donor variables

Gender mismatch was defined as either a male recipient receiving a graft from a female donor (F→M) or a female recipient receiving a graft from a male donor (M→F). All hazard ratios involving gender mismatch are reported with M → M matched transplants as the reference group. The DRI was calculated using the formula established by Feng et al. (8) Missing values for cold and warm ischemia times were imputed with the median times per region. Cutoffs for selected variables that were deemed to be implausible for an adult recipient were as follows: recipient height <120 cm or > 240 cm, recipient weight <30 kg or >180 kg, donor age <10 years or > 95 years, donor height <100 cm or >180 cm and donor weight <20 kg or >180 kg, cold ischemia time < 1 h or > 24 h and warm ischemia time <10 min or > 120 min. Observations including these implausible values were set as missing. Sensitivity analyses comparing the multivariable models using case-wise deletion for missing values confirmed that imputation did not substantially change the interpretation of the final results.

Outcomes and censoring

The primary predictor in our study was donor–recipient gender mismatch. The primary outcome was graft failure. Analyses focused on the key differences between female and male donors and their effect on the association between donor–recipient gender mismatch and graft outcome. Patient follow-up began on the day of transplantation and ended either at the time of graft failure or date of last data update in the UNOS STAR database.

Statistical analysis

Recipient characteristics immediately prior to transplantation and their donors were compared using Wilcoxon and chi-square tests as appropriate. A cutoff p-value <0.05 was used for all tests of significance; all tests were two-sided. Associations between donor variables and graft failure were evaluated using univariable and backward selection multivariable Cox regression methods. Continuous variables that were nonlinearly associated with graft failure in univariable Cox analysis were modeled in the multivariable analyses using linear splines. Multiple clinically relevant two-way interactions were evaluated in the multivariable Cox model and included in the final model if significant at a p-value <0.05. The proportional hazards assumption of the final adjusted model was tested visually by plotting the scaled Schoenfeld residuals of donor–recipient gender mismatch, the main predictor of interest. Accounting for nonproportionality by stratification of the model on significant categorical variables and deciles of continuous variables confirmed the validity of the association between the main predictors of interest and the primary outcome. To account for cohort effects, all analyses were adjusted for year of transplant.

Analyses were performed using Stata®11.0 statistical software (College Station, TX, USA). The institutional review board at the University of California San Francisco approved this study.


A total of 28 222 adult patients underwent primary, single-organ liver transplantation in the United States during the study period—9226 (33%) were female and 18 996 (67%) were male. Baseline characteristics of the recipients are shown in Table 1. There was no significant difference in median age between female and male recipients, but females were smaller as measured by height, weight and BMI. A greater proportion of females were of African American race, had NAFLD or autoimmune disease as their etiology of liver failure and underwent living donor liver transplantation. Fewer women were transplanted for HCV-related liver disease, alcohol-related liver disease and hepatocellular carcinoma. The MELD score at transplant was significantly higher in females than in males, and this was largely driven by a higher median total bilirubin level in females compared to males, in spite of a lower median serum creatinine level. Cold ischemia time was not significantly different between the two groups.

Table 1.  Characteristics of liver transplant recipients from March 1, 2002 through December 31, 2007
Recipient characteristics1Recipient female (n = 9226)Recipient male (n = 18 996)p-Value
  1. 1IQR = interquartile range.

  2. 2Body mass index = (Weight in kilograms)/(Height in meters)2.

Age in years: median (IQR)54 (46–60)53 (48–59)0.28
Race: N (%)
 Caucasian6466 (70%)14 360 (75%)<0.001
 African American992 (11%)1371 (7%) 
 Hispanic1231 (13%)2229 (12%) 
 Asian425 (5%)866 (5%) 
 Other112 (1%)170 (1%) 
Etiology of liver disease: N (%)
 Hepatitis C virus2406 (26%)6997 (37%)<0.001
 Hepatitis B virus154 (2%)625 (3%) 
 Alcohol722 (8%)2729 (14%) 
 Nonalcoholic fatty liver disease1222 (13%)1506 (8%) 
 Autoimmune1919 (21%)1406 (7%) 
 Acute271 (3%)99 (1%) 
 Other2532 (27%)5634 (30%) 
Size: median (IQR)
 Height in centimeters163 (158–168)178 (173–183)<0.001
 Weight in kilograms72 (62–85)87 (77–100)<0.001
 Body mass index227 (24–32)28 (25–32)<0.001
Living donor liver transplant: N (%)615 (7%)813 (4%)<0.001
Hepatocellular carcinoma: N (%)1356 (15%)4525 (24%)<0.001
Laboratory values at transplant: median (IQR)
 Total bilirubin3.7 (1.8–9.1)3.2 (1.7–6.5)<0.001
 International normalized ratio1.6 (1.3–2.1)1.5 (1.3–2.0)<0.001
 Creatinine1 (1–1.5)1.1 (1–1.5)<0.001
 MELD score at transplant19.0 (13.3–28.4)17.7 (12.9–24.8)<0.001

Among donors, 11 508 (41%) were female and 16 714 (59%) were male. Compared to male donors, female donors were significantly older (47 vs. 39 years; p < 0.001), smaller according to height (165 vs. 178 cm; p < 0.001) and weight (69 vs. 81 kg; p < 0.001), and more likely to die of a cerebrovascular accident (59 vs. 35%; p < 0.001). The DRI was significantly higher for female compared with male donors (1.6 vs. 1.3; p < 0.001). Differences in the proportions of female grafts from donation after cardiac death, nonlocal origin (regional or national), split livers or from donors with diabetes or hypertension were statistically significant, but of unlikely clinical significance (Table 2). There were no significant differences in the proportion of female donors by region or by year of transplant (data not shown).

Table 2.  Characteristics of liver donors from March 1, 2002 through December 31, 2007
Donor characteristics1Donor female (n = 11 508)Donor male (n = 16 714)p-Value
  1. 1IQR = interquartile range.

  2. 2Body mass index = (Weight in kilograms)/(Height in meters)2.

  3. 3Donor risk index = exp[(0.154 if 40 ≤ donor age < 50) + (0.274 if 50 ≤ donor age < 60) + (0.424 if 60 ≤ donor age < 70) + (0.501 if 70 ≤ donor age) + (0.079 if cause of death = anoxia) + (0.145 if cause of death = cerebrovascular accident) + (0.184 if cause of death = other) + (0.176 if donor race = African American) + (0.126 if donor race = other) + (0.411 if donation after cardiac death) + (0.422 if partial/split) + (0.066 ((170–donor height)/10)) + (0.105 if regional share) + (0.244 if national share) + (0.010 × cold ischemia time)].

Age in years: median (IQR)47 (33–57)39 (23–52)<0.001
Race: N (%)
 Caucasian8375 (73%)11 503 (69%)<0.001
 African American1598 (14%)2346 (14%) 
 Hispanic1133 (10%)2342(14%) 
 Asian272 (2%)308 (2%) 
 Other130 (1%)215 (1%) 
Size: median (IQR)
 Height in centimeters165 (160–168)178 (173–183)<0.001
 Weight in kilograms69 (60–82)81 (71–93)<0.001
 Body mass index226 (22–30)26 (23–29) 0.004
Cause of death: N (%)
 Anoxia1649 (15%)1900 (12%)<0.001
 Cerebrovascular accident6322 (59%)5567 (35%) 
 Head trauma2527 (23%)8125 (51%) 
 Other322 (3%)380 (2%) 
Donation after cardiac death371 (3%)729 (5%)<0.001
Share region: N (%)
 Local8287 (72%)12 196 (73%) 0.25
 Regional2442 (21%)3445 (21%) 
 National779 (7%)1072 (6%) 
Split liver: N (%)803 (7%)959 (6%)<0.001
Diabetes: N (%)1082 (9%)1390 (8%) 0.002
Hypertension: N (%)669 (6%)736 (4%)<0.001
Cold ischemia time in hours: median (IQR)7 (5.5–8.5)7 (5.5–8.7) 0.34
Warm ischemia time in minutes: median (IQR)39 (36–43)40 (37–44)<0.001
Donor risk index,3 median (IQR)1.6 (1.3–2.0)1.3 (1.1–1.7)<0.001

Male recipients were less likely than female recipients to receive a gender-mismatched graft (37% vs. 50%; p < 0.001). One-, 3- and 5-year survival rates for recipients of a gender-mismatched graft (regardless of recipient gender) were 83%, 72% and 65%, respectively, versus 86%, 75% and 68% for a gender-matched graft (p < 0.001). Compared to recipients of a gender-matched graft, recipients of a gender-mismatched graft were at 11% higher risk of graft loss (hazard ratio [HR], 1.11; 95% confidence interval [CI], 1.06–1.16; p < 0.001).

The Kaplan–Meier survival curves for each donor–recipient gender pairing are shown in Figure 1. Graft survival rates at 1, 3 and 5 years were 86%, 75%, 67% for M→M matched recipients, 85%, 75%, 68% for F→F matched recipients, 83%, 71%, 64% for F→M mismatched recipients and 84%, 74% and 68% for M→F mismatched recipients (p < 0.001). In univariable analysis, F→M mismatched recipients, compared to M→M matched recipients, were at increased risk of graft loss (HR, 1.17; 95% CI, 1.11–1.23; p < 0.001), whereas M→F recipients were not (HR, 1.02; 95% CI, 0.96–1.09; p = 0.46; Table 3). After adjustment for clinically relevant recipient, donor and transplant-related factors, there was no longer an association between F→M mismatch and graft loss (Table 3). The dominant donor characteristics affecting the association between gender mismatch and graft loss were donor age and donor height, although donation after cardiac death, stroke as the cause of death, split liver and donor diabetes were also predictive of graft loss (Table 3).

Figure 1.

Unadjusted Kaplan–Meier survival curves by donor–recipient gender pairing.

Table 3.  Cox analyses of the association between donor–recipient gender mismatch and graft failure
PredictorUnivariable HR (95% CI) p-ValueMultivariable1 HR (95% CI) p-Value
  1. 1Each model is also adjusted for: recipient age, recipient African American race, hepatocellular carcinoma, recipient weight, etiology of liver disease, MELD at transplant, region of transplant, cold ischemia time, and the interaction between split liver and donor–recipient gender mismatch.

M→M match (reference)
F→F match1.00 (0.94–1.07) 0.991.02 (0.95–1.10) 0.540.96 (0.89–1.03) 0.220.85 (0.78–0.92) <0.0010.86 (0.79–0.93) <0.001
F→M mismatch1.17 (1.11–1.23) <0.0011.16 (1.09–1.22) <0.0011.08 (1.02–1.14) 0.0070.96 (0.90–1.03) 0.240.95 (0.89–1.02) 0.18
M→F mismatch1.02 (0.96–1.09) 0.461.01 (0.95–1.09) 0.681.04 (0.97–1.11) 0.281.02 (0.95–1.10) 0.501.02 (0.95–1.10) 0.51
Donor age (per 10 years)1.13 (1.12–1.15) <0.0011.13 (1.12–1.15) <0.0011.13 (1.11–1.14) <0.001
Donor height (per 10 cm)0.92 (0.90–0.94) <0.0010.92 (0.89–0.94) <0.001
Donation after cardiac death1.70 (1.54–1.89) <0.001
Cause of death-Stroke1.07 (1.01–1.13) 0.02
Split liver1.71 (1.23–2.37) 0.001
Donor diabetes1.15 (1.07–1.24) <0.001

In a post hoc analysis to further evaluate the association between donor–recipient gender pairing and graft loss, we observed an important effect modification by recipient HCV status (Table 4). In univariable analysis, compared to M→M matched recipients, female HCV recipients experienced an increased hazard of graft failure regardless of donor gender. This was not seen among non-HCV recipients (Table 4). Adjustment for other factors associated with graft loss eliminated the association between F→F matching and graft loss among HCV recipients, but not among M→F mismatched recipients (Table 4). In contrast, among non-HCV recipients, there was no increased risk of graft loss among the M→F group (Table 4).

Table 4.  Comparison of univariable and multivariable analyses to evaluate the association between donor–recipient gender mismatch and graft loss, stratified by recipient HCV-status
Univariable HR (95% CI) p-ValueMultivariable1 HR (95% CI) p-ValueUnivariable HR (95% CI) p-ValueMultivariable2 HR (95% CI) p-Value
  1. 1Adjusted for recipient age; African American race; hepatocellular carcinoma; recipient weight; MELD score at transplant; cold ischemia time; donor age; donation after cardiac death; donor height; donor African American race; split liver; donor diabetes; region of transplant; and year of transplant.

  2. 2Adjusted for recipient age; African American race; MELD score at transplant; region of transplant; year of transplant; donor age; donation after cardiac death; donor cause of death-stroke; donor height; donor origin (local, regional, national); donor diabetes and cold ischemia time.

M→M match (reference)
F→F match0.91 (0.84–0.99) 0.020.77 (0.69–0.85) <0.0011.29 (1.16–1.44) <0.0011.06 (0.93–1.21) 0.39
F→M mismatch1.17 (1.09–1.25) <0.0010.96 (0.88–1.05) 0.351.19 (1.09–1.30) <0.0010.92 (0.83–1.03) 0.14
M→F mismatch0.95 (0.88–1.03) 0.200.93 (0.85–1.02) 0.121.26 (1.13–1.40) <0.0011.23 (1.10–1.38) <0.001


Our study is the first to show that donor quality differs significantly between female and male donors. As a result of older age, shorter height and increased proportion of death from cerebrovascular accident, the DRI was 23% higher for female compared to male donors. This highlights substantial differences in the quality of the national donor pool secondary to gender. Our analyses demonstrate that these donor quality variables account for the F→M gender mismatch effect in liver transplantation that had been previously reported (5,6). The dominant explanatory donor factors were age, height, donation after cardiac death and split liver, all of which have previously been shown to be associated with graft failure (8). Additionally, we found that donor diabetes was associated with a 15% increased risk of graft failure, and although this factor did not emerge as predictive of graft loss in the DRI, our findings likely reflect the rising prevalence of diabetes in the deceased donor pool (10).

Given the demographic distribution of women and men in the United States, the gender-based differences that we have identified among deceased liver donors are not surprising. The life expectancy for women is 5 years longer than for men (11), likely resulting in the older median donor age of female liver donors. Overall, women are more likely to suffer from a fatal stroke compared to men (12), and a recent study using data from the National Health and Nutrition Examination Surveys (NHANES) found that women between the ages of 35–54 years of age were more than twice as likely as men to report a stroke. There was no gender difference in the prevalence of stroke in the 55–64 years of age group (13). In addition, men between the ages of 15–45 years are 1.4 times more likely to experience a traumatic accident compared to women in the same age group, whereas they were less likely than women to experience trauma above the age of 65 years, further contributing to the significant differences in age and cause of death among male and female donors (14). As the average organ utilization is three organs per deceased donor (15), these gender differences among liver donors are clearly applicable to transplantation of other organs. Whether these differences explain the gender mismatch effect seen in kidney, lung and heart transplantation warrants additional investigation (16–18).

Our findings differ from those of prior studies that found that F→M gender mismatch was associated with an increased risk of graft failure. There are several potential explanations for this difference. Importantly, prior analyses did not account for the multiple donor characteristics highlighted by the DRI that are associated with graft loss (5–7). Our multivariable analyses evaluated not only all of these donor factors but also included important predictive recipient (e.g. age, African American race, MELD, etiology of liver disease, hepatocellular carcinoma) and transplant-related factors (e.g. cold and warm ischemia times, region of transplant). Additionally, gender mismatch has been linked with chronic rejection (19), and in our more contemporary cohort of liver transplant recipients, with improved immunosuppression regimens, the contribution of chronic rejection to graft losses may be less.

Interestingly, our multivariable analyses revealed that F→F matched transplants were associated with a decreased risk of graft failure compared to M→M matched transplants. The biology of this association is not entirely clear. However, in an exploratory analysis, this effect was seen only among female non-HCV recipients. Among female HCV recipients compared to the entire cohort, a similar direction and magnitude of risk of graft failure was seen, regardless of donor–recipient gender pairing. Specifically, for F→F matched recipients, the hazard ratio increased from 0.86 to 1.06 and for M→F mismatched recipients, the hazard ratio increased from 1.03 to 1.23, suggesting that female recipient HCV status is an important effect modifier in the association between donor–recipient gender pairing and graft failure. This is consistent with prior studies that have demonstrated that female HCV recipients experience worse outcomes compared to female non-HCV recipients (20) and compared to male HCV recipients (1). Although not within the scope of our UNOS-registry-based study, further cohort studies that can adjust for posttransplant factors including treatment of acute rejection, immunosuppressive therapy and antiviral treatment for HCV are needed to better understand the effects of F→M mismatch on graft survival among HCV recipients.

Our study has important implications. Given the challenging task of allocating scarce liver grafts to the appropriate candidates, this study highlights the key donor aspects that impact graft failure and demonstrates that donor gender per se is not a detrimental donor factor. Instead, our findings underscore the importance of specific donor characteristics, including age, height, cause of death and donation after cardiac death—rather than donor–recipient gender mismatch—as predictors of graft loss after liver transplantation. In light of the current liver donor graft deficit, development of posttransplant strategies that mitigate the negative effect of these unfavorable donor factors is needed.


We would like to thank Peter Bacchetti, PhD, Professor of the Division of Biostatistics, for his statistical advice for which he did not receive any compensation.


The authors of this manuscript have no conflicts of interest to disclose as described by the American Journal of Transplantation.