The Effect of Donor Race on the Survival of Black Americans Undergoing Liver Transplantation for Chronic Hepatitis C

Abstract

The purpose of this study was to determine the effect of donor race on the outcome of black patients with chronic hepatitis C infection who undergo liver transplantation. The records of deceased donor liver transplantations that occurred in the United States between January 1998 and December 2007 were obtained from the United Network for Organ Sharing (UNOS). 26,212 records contained sufficient data to be included in the analysis. Of these, 11,989 (45.7%) were positive for hepatitis C virus (HCV) and 1,292 (4.9%) were both HCV-positive and black. Black recipients with white donors were found to have significantly worse outcomes compared to all other recipient-donor race combinations (p < 0.001). The crude five-year survival rate for black recipients who had a black donor was 14% higher than the five-year survival rate for black recipients who had a white donor. Multivariate regression analysis determined that a graft from a race-unmatched donor was an independent risk factor for graft failure (HR = 1.41, 95% CI 1.11-1.79) among HCV-positive black recipients, but not among HCV-negative black recipients, after adjusting for donor age, recipient age, cold ischemia time, serum creatinine, serum bilirubin, diabetes mellitus, body mass index (BMI), and donor cytomegalovirus (CMV) status. The observation that race-unmatched grafts are a risk factor in HCV-positive black recipients, but not in HCV-negative black recipients, suggests an alteration of the graft-host relationship by HCV. In conclusion, our results suggest that HCV-positive black recipients who undergo liver transplantation can have increased graft survival if their donor is black, with survival rates approaching that of white liver transplant recipients.

INTRODUCTION

Prior to the Model for End Stage Liver Disease (MELD) era, black patients were underrepresented on the liver transplant waiting list and were more likely to die while awaiting transplantation. The current MELD scoring system appears to have eliminated racial differences in access to transplantation (1). However, for unclear reasons, black recipients continue to experience lower post-transplant graft survival rates when compared to their white peers (2-4). In the United States, the leading indication for liver transplantation for both blacks and whites is the hepatitis C virus (HCV) (5). This infection is twice as common among the black population compared to the white population (3). Black patients are also more likely than white patients to be infected with HCV genotype-1, the genotype that is the most refractory to current therapy (3). Unsurprisingly, a failure to adequately clear HCV increases the risk of cirrhosis and the need for liver transplantation.

Due to the higher prevalence of whites in the liver donor pool, black recipients often receive a graft from a racially unmatched donor. We hypothesized that the poor graft survival rates seen among black liver transplant recipients might be due in part to having a racially unmatched donor. In this study we sought to test this hypothesis by determining the extent to which such mismatch accounts for the higher mortality seen among HCV-infected black patients who undergo liver transplantation.

METHODS

Study Population and Definitions

Records of all adult liver transplants performed in the United States between January 1998 and December 2007 were obtained from UNOS Standard Transplant Analysis and Research files, created on May 20, 2008. These files contain one record per transplant event, along with data from the most recent follow-up visit for each patient. Data are collected by each transplant center and transmitted to UNOS at the time of registration, time of transplant, 6 months after transplant, and annually thereafter. This ten year time period, spanning both the pre-MELD and post-MELD eras, was initially chosen to ensure an adequate number of black patients with black donors. To determine whether our results were applicable to current practice standards, a separate analysis was performed restricted to transplants performed after February, 2002, the post-MELD period. The following exclusion criteria were applied: missing HCV status, a donor or recipient race other than non-Hispanic White or non-Hispanic Black, use of a non-heart beating donor, multi-organ transplant, split liver transplant, and use of a living donor. Race was self-identified by patients when registered in the UNOS database. See Supplementary Figure 1 for exclusion statistics.

The hepatitis C cohort was defined as those patients with documented detection of an antibody to HCV (anti-HCV) and either a diagnosis of “Cirrhosis Type C” or “Hepatitis C: Chronic or Acute”. The hepatitis C-negative cohort consisted of patients with documented negative anti-HCV and a diagnosis other than “Cirrhosis Type C”, “Hepatitis C: Chronic or Acute”, “Cirrhosis: Type B and C”, and “Alcoholic Cirrhosis with Hepatitis C”. The post-MELD cohort was defined as those patients receiving a liver transplant after February 2002. Graft failure was defined as death or the need for retransplantation.

Statistical Analyses

The primary outcome measure was graft survival in white versus black transplant patients who received grafts from race matched versus unmatched donors. Overall survival was computed using the Kaplan-Meier estimator. Comparisons of survival between groups were performed using the log-rank test. Cox Proportional Hazards Regression modeling was used to determine whether donor race was an independent predictor of graft survival following transplantation. To build the model, we first performed univariate analysis for all variables that either (i) had previously been shown to independently predict post-transplant outcome, or (ii) had a biologically plausible reason to impact post-transplant mortality. We then included those variables shown to be significant predictors of outcome in our univariate analysis, in our multivariate analysis.

For the regression analyses, observations that contained a missing value for any predictor were excluded; the percent of records included in each regression analysis is indicated in the regression table. To address the impact of missing data, a second model was constructed that excluded no records and which used single imputation for any missing variables.

The Breslow-Day statistic for homogeneity across strata was used to investigate effect modification. Additionally, a combined model that included for the possibility of 2-way and 3-way interactions among race, HCV status, and donor-recipient mismatch was constructed, to quantitate possible combinatorial effects.

Fisher’s exact tests and the Chi-square statistic (for larger Ns) were used to determine if the proportion of patients with graft-host HLA mismatch at the A, B, and DR loci was statistically different between patients having either a black versus white donor. P values were two-sided and were not adjusted for multiple testing. Analyses were carried out using SAS Version 9.1 (SAS Institute, Cary, NC).

RESULTS

Of the 50,435 adult liver transplants in the UNOS database between January 1998 and December 2007, 26,212 records met inclusion criteria. Of these, 11,989 (45.7%) were HCV-positive, 2,646 (10.1%) were black recipients, and 1,292 (4.9%) were black recipients who were HCV-positive.

The characteristics of these patients are shown in Tables 1A and 1B. The HCV-positive black patient cohort contained more women, was slightly older, had higher baseline creatinine and lower baseline albumin levels, was more likely to be CMV seropositive, and was more likely to have diabetes mellitus compared to the HCV-positive white patient cohort. The median follow-up for all patients was 919 days.

Table 1.

A: Patient Characteristics

	HEPATITISC POSITIVE
	White Recipient			Black Recipient
	White Donor	Black Donor	P Valuea	Black Donor	White Donor	P Valueb	P Valuec
N	9067	1636	-	316	976	-	-
Age (yr)	51.3 (7.30)	51.7 (7.28)	0.02	52.3 (7.33)	52.5 (6.83)	0.71	<.0001
Female sex (%)	22	23	0.47	26	32	0.04	<.0001
Donor age (yr)	40.6 (16.9)	39.3 (16.0)	0.002	40.7 (16.6)	39.4 (16.9)	0.24	0.18
BMI (kg/m2)	28.7 (5.31)	28.9 (5.60)	0.96	28.2 (5.70)	23.3 (5.06)	0.3	0.51
INR	1.77 (1.2)	1.81 (1.8)	0.54	1.96 (1.5)	1.94 (1.1)	0.85	0.0003
Creatinine (mg/dl)	1.27 (0.96)	1.25 (0.95)	0.21	1.38 (1.14)	1.52 (0.98)	0.04	<.0001
Bilirublin (mg/dl)	6.1 (9.1)	5.9 (8.8)	0.54	6.3 (8.5)	6.4 (8.4)	0.81	0.2
Albumin (mg/dl)	2.9 (0.82)	2.9 (0.66)	0.58	2.7 (0.69)	2.7 (0.75)	0.47	<.0001
Cold Ischemia time (hr)	7.82 (3.84)	7.48 (3.89)	0.002	7.49 (3.60)	7.85 (4.01)	0.17	0.95
Warm ischemia time (min)	43 (19.9)	43 (20.0)	0.39	43 (20.0)	44 (21.0)	0.61	0.2
Donor CMV (%)	59	80	<.0001	75	59	<.0001	0.74
Recipient CMV (%)	66	67	0.21	81	83	0.51	<.0001
Diabetes (%)	16	18	0.07	23	25	0.51	<.0001
Black Donor (%)	0	100	-	100	0	-	-

B: Patient Characteristics
HEPATITISC NEGATIVE
	White Recipient	Black Recipient	P Valued
N	12863	1354	-
Age (yr)	52.5 (11.4)	43.1 (13.0)	<.0001
Female sex (%)	39	55	<.0001
Donor age (yr)	41.1 (18.1)	38.9 (17.9)	<.0001
BMI (kg/m2)	28.2 (6.00)	27.2 (6.11)	0.11
INR	1.89 (1.4)	2.25 (2.1)	<.0001
Creatinine (mg/dl)	1.39 (1.08)	1.46 (1.29)	0.05
Bilirublin (mg/dl)	7.8 (10.1)	12.8 (12.8)	<.0001
Albumin (mg/dl)	2.9 (0.69)	2.7 (0.72)	<.0001
Cold Ischemia time (hr)	7.72 (3.62)	7.86 (3.87)	0.23
Warm ischemia time (min)	43 (19.8)	43 (21.4)	0.4
Donor CMV (%)	62	62	0.95
Recipient CMV (%)	62	80	<.0001
Diabetes (%)	21	14	<.0001
Black Donor (%)	14.6	21.9	<.0001

Mean values (S.D.)

^aWhite versus black donors in HCV(+) white transplant recipients

^bBlack versus white donors in HCV(+) black transplant recipients

^cWhite versus black HCV(+) transplant recipients

^dWhite versus black HCV(-) transplant recipients

Figure 1 shows the Kaplan-Meier survival curves for black versus white HCV-positive transplant recipients. As previously observed, HCV-positive black transplant recipients have a lower graft survival rate compared to HCV-positive white recipients (p < 0.001) (4). As the majority of transplant donors in the United States are white, the large majority (76%) of these black transplant recipients had white donors.

Figure 1. Kaplan-Meier survival curves.

HCV-positive white versus HCV-positive black liver transplant recipients.

Figure 2A compares matched versus unmatched recipient-donor race in HCV-positive liver transplant patients. Black recipients with white donors had significantly worse outcomes than all other recipient-donor race combinations (p < 0.001). The five-year graft survival rate for black recipients who had white donors was 45%. By comparison, the five-year survival rate for black recipients who had black donors was 59%, an absolute increase in survival of 14%. HCV-positive black recipients with black donors had survival rates similar to HCV-positive white recipients overall (59% versus 63%).

Figure 2. Kaplan-Meier graft survival curves.

A. HCV-positive liver transplant recipients, by recipient-donor race. B. HCV-negative liver transplant recipients, by recipient-donor race. In panels A and B, the number of patients at risk in each cohort is indicated.

Figure 2B shows the Kaplan-Meier survival curves for matched versus unmatched recipient-donor race in HCV-negative transplant patients. This HCV-negative transplant population did not demonstrate a notable difference in graft survival between black recipients with black donors versus those with white donors; five-year graft survival was 63% versus 66%, respectively. This analysis also found worse overall outcomes for HCV-positive versus HCV-negative transplant patients, consistent with prior observations (6). The corresponding Kaplan-Meier curves for patient survival can be found in Supplemental Figure 2, and depict trends quite similar to the graft survival curves shown in Figure 2.

Multivariate Cox Proportional Hazards Regression modeling (Table 2) determined that having a graft from a white donor was an independent risk factor for graft failure among HCV-positive black recipients (HR = 1.41, p = 0.005), after controlling for factors previously found to be associated with differences in transplant outcome: donor age, recipient age, cold ischemia time, serum creatinine, serum bilirubin, diabetes mellitus, BMI, and donor CMV status (7-9). Of note, in univariate analyses, gender, warm ischemia time, and donor diabetes mellitus status were not found to be significant predictors of outcome and were consequently not included in the multivariate analysis. When the regression analysis was limited to only post-MELD transplants, having a graft from a white donor was again found to be an independent risk factor for graft failure among HCV-positive black recipients (HR = 1.47, p = 0.007).

Table 2.

Multivariate Cox Regression Analysis for Risk of Graft Failure: Hazard Ratios for Donor/Recipient Race Mismatch

	Regression Analysis Cohortsa
	HCV(+) Black Recipients		HCV(-) Black Recipients
	Adjusted HRb (95% CI)	P value	Adjusted HRb (95% CI)	P value
Race Matched Donor	1	-	1	-
Race Unmatched Donor	1.41 (1.11 – 1.79)	0.005	0.89 (0.67 – 1.17)	0.383
	HCV(+) White Recipients		HCV(-) White Recipients
	Adjusted HRb (95% CI)	P value	Adjusted HRb (95% CI)	P value
Race Matched Donor	1	-	1	-
Race Unmatched Donor	1.13 (1.02 – 1.26)	0.019	1.26 (1.13 – 1.39)	<0.001

^aSeparate regression analyses performed on each of the four transplant cohorts to determine the risk attributable to a race unmatched graft.

^bAdjusted for donor age, recipient age, cold ischemia time, serum creatinine, serum bilirubin, diabetes mellitus, BMI, and donor CMV status. Percent usable observations (patients included in the multivariate analysis; others excluded due to missing data) were: HCV(+) black = 86%; HCV(-) black = 79%; HCV(+) white = 85%; HCV(-) white = 82%.

In contrast, a Cox regression analysis in HCV-negative black recipients did not find that having a graft from a white donor was an independent risk factor for graft failure (HR = 0.89, p = NS). Thus, HCV appears to be an effect modifier: only in HCV-infected black recipients is a white donor associated with an increased risk of graft failure. This was formally tested using the Breslow-Day statistic for heterogeneity. As expected, significant heterogeneity in graft survival was seen among black recipients when donor-recipient mismatch was stratified by the presence or absence of HCV (p = 0.003). To further investigate this interaction, we constructed a combined model that included 2-way and 3-way interaction terms among race, HCV status, and mismatch. Each one of these main effects was a significant predictor of mortality, as was the 3-way interaction term (p < 0.002), suggesting that the combination of black race, HCV positive status, and donor/recipient race mismatch is particularity deleterious. Regression coefficients, hazard ratios, and p values are shown in Supplemental Table 1.

When the above regression analyses were then repeated, including all records and using single imputation for all missing variables, the above observations remained significant, suggesting that any impact from missing data was minimal.

To address the potential role of HLA mismatch in these observations, we performed a comparison of the degree of HLA mismatch (A, B, and DR loci) between HCV-positive recipients with black donors versus those with white donors is shown at the top of Figure 3. Among black recipients, having a black versus white donor did not alter the statistical likelihood of having any degree of recipient-donor mismatch at the HLA loci tested. Specifically, the null-hypothesis was not rejected when using Fisher’s exact test to examine the proportion of HCV-positive black recipients with black versus white donors, at HLA match levels of ≥ 2/6 (p = 0.54), ≥ 3/6 (p = 0.18), and ≥ 4/6 (p = 0.55). Thus, the observed difference in outcome between HCV-positive black recipients with black donors versus those with white donors cannot be attributed, per se, to simple HLA mismatch at the A, B, and DR loci. A limitation of this analysis is that the HLA type for both recipient and donor was available in only 46% of HCV-positive transplant cases; the frequency of missing data, however, was similar across the different donor/recipient race combinations.

Figure 3. Degree of HLA Match between Recipient and Donor in HCV-Infected Liver Transplant Patients.

The cumulative percent of patients by HLA match (A, B, and DR loci) is illustrated in the bar graphs (top), from 6/6 match (left) to 0/0 match (right). Actual percent numbers are given in the table below. *P-value for an HLA match level ≥ 3/6, using Fisher’s exact test; **P-value for an HLA match level ≥ 3/6, using Chisquared statistic, which was necessary given the large number of white transplant patients. See text for additional statistics.

We also examined cause of death (COD) in black and white donors, including stroke, head trauma, and central nervous system tumors. Blacks were more likely to have died from cerebrovascular accident or stroke (CVA) than whites (49% vs. 44%, p <0.001). Whites were more likely to have died from head trauma than blacks (43% vs. 38%, p <0.001). In univariate analysis, only donor death from CVA was found to be a risk factor for post-transplant mortality. However, CVA was not found to be a risk factor in multivariate analysis. This is not unexpected, because donors who died from CVA tended to be older than those who died from other causes, and thus the apparent effect of this particular cause of death is approximated by donor age, for which there was no difference between black and white recipients (p = .24, Table 1A), Thus, cause of death does not appear to be an independent factor that explains our observations. This is also consistent with our initial observations: we found that donor-recipient mismatch plays a significant role only in HCV-positive recipients, but not HCV-negative recipients, and yet because both cohorts would be expected to have donors with similar COD, donor COD is an unlikely explanation for these observations.

DISCUSSION

This study shows that black patients who are HCV-positive and undergo liver transplantation have improved graft survival if their donor is black. Cox regression analysis determined that among HCV-positive black recipients, having a black donor is an independent predictor of graft survival. The observed increase in graft survival due to recipient-donor race matching is notably specific to black transplant recipients infected with HCV. It was also determined that simple HLA mismatch at the A, B, and DR loci, per se, does not appear to be responsible for the decreased survival of HCV-positive black recipients who had white donors, as no statistical difference in the degree of HLA mismatch was found between HCV-positive black recipients with black versus white donors. Nevertheless, it is quite possible that database limitations, more subtle differences at these genetic loci, or other immune-related genetic loci, play a role in the outcomes observed here (9) (10).

It has been previously observed that a black donor is associated with a lower graft survival rate (7, 11). This result is not contradictory to our analysis. We also show that having a black donor is an independent predictor of graft failure, but specifically within the white cohort that makes up the large majority of the transplant population. The magnitude of this effect, however, was smaller (HR = 1.13) and not specific to HCV-infected patients; furthermore, differences in the degree of HLA match among white recipients complicated interpretation of this result. Also notable is the study by Nair and Thuluvath (12), which analyzed donor-recipient race mismatch in an exclusively pre-MELD cohort (1988-1996). This smaller study analyzed graft survival at two years, and concluded that black recipients with black donors fared marginally worse than those with white donors. As this study did not stratify by HCV-status, a direct comparison cannot be made. Nevertheless, among our HCV-negative cohort, we also observe a difference in graft survival at two years, similar to the one seen by Nair and Thuluvath, but note that this difference becomes largely immaterial by year five (Figure 2B).

The reason for the improved graft survival conferred by race-matched grafts in HCV-positive black recipients is not clear. Due to the large number of patients (49.8%) with incompletely recorded, multi-factorial, or indeterminate cause of death, we can only speculate on potential factors that may or may not be responsible. We note that the Kaplan-Meier curves for HCV-positive black recipients with black versus white donors diverge by as early as 1 year, and continued to further diverge thereafter (Figure 2A). This suggests that the processes responsible for this difference in survival are ongoing ones, and not likely solely related to the immediate post-transplant period. Similarly, incomplete data precluded an analysis of viral recurrence, rejection rates, or a comparison of immunosuppression regimens. Nevertheless, our results suggest that a genetic component plays a role in HCV-positive black transplant recipient outcomes, as opposed to outcome being determined entirely by demographic factors, given that HCV-positive black transplant recipients are likely to be demographically similar to one another regardless of whether their donor is white or black.

A number of possible processes may be responsible for these outcome differences, including an increased inflammatory response in the presence of HCV or suboptimal control of viral replication. For example, studies suggest that HCV can down-regulate natural killer (NK) cell activity (13); in turn, NK activity has been shown to correlate with the clearance of HCV (14). Thus, genetic differences in NK receptors, which have been shown to vary by race (10), may account for the positive effect of having a black donor that has been observed specifically in the context of black recipients transplanted for chronic hepatitis C. As the liver is also an immuno-privileged organ with a key role in tolerance and T-cell apoptosis (15), other immune pathways may be involved as well. More detailed datasets and further laboratory studies are necessary to address these questions.

Another possibility is a silent infection carried by the graft that affects post-transplant HCV pathogenesis, such as a latent hepatotropic virus. CMV serostatus, however, was not found to account for the deleterious affect of Caucasian grafts in HCV-positive African-Americans. Insufficient data existed to examine the role of EBV or other herpes viruses.

Due to database limitations, donor steatosis was not included in our regression analysis. Donor steatosis has previously been shown to decrease graft survival (8). However, it appears unlikely that this factor is confounding our analysis. The risk factors for steatosis include diabetes mellitus and obesity, both of which are more common among blacks. Thus, if any effect was observed by controlling for donor steatosis, it might actually be to increase the positive effect attributable to placing race-matched grafts in HCV-positive black transplant patients.

Notably, the number of black recipients with black donors who had follow-up data for 5 or more years (n=44) was relatively small (Figure 2A). Despite this small number, however, statistical significance was achieved both in the Kaplan-Meier analysis/log-rank test, as well as in our regression analysis, suggesting that the strength of the observation was substantial. Furthermore, the observation that the curve for HCV-positive black recipients with white donors diverges from that of those with black donors as early as 1 year and in a statistically significant manner, when >200 black transplant recipients remain in the analysis, also provides support for the clinical relevance of these observation (Figure 2A).

Although preliminary, this retrospective study has found a notable survival advantage in HCV-positive black transplant recipients whose donors are black. The mechanism underlying this effect and its HCV-specific nature remains an area for future research. Additionally, case series that are able to detail: (a) graft failure due to rejection versus viral recurrence, as a function of donor and recipient race, while (b) addressing immunosuppressive regimens, are likely to shed further light on this matter. Of note, post-transplant treatment of HCV infected patients is now increasingly common (16). As the number of patients treated increases, the ability to study the role of donor-recipient race mismatch on post-transplant treatment outcome might also shed a new light on the relative role of the liver itself and its response to the immune system, versus the role of the host immune system in response to interferon therapy. Interestingly, a single-center experience with pediatric heart transplants recently demonstrated that black recipients with black donors fared better than those who had white donors (17). An analysis of UNOS-derived data of deceased kidney organ transplantation also found that black recipients with black donors fared better than those who had white donors (18).

In conclusion, this study suggests that race mismatch appears to play a significant role in the low rate of graft survival seen among HCV-positive blacks, an observation which may ultimately provide insight into the genetic and non-genetic factors that govern host control of HCV and liver transplant outcomes.

Abbreviations

UNOS

United Network for Organ Sharing

HCV

Hepatitis C Virus

CMV

Cytomegalovirus

BMI

Body mass index

MELD

Model for End Stage Liver Disease

NK

Natural killer