Objectives: An age-dependent interaction has been described for the effect of donor-recipient sex mismatch on outcomes after kidney transplant in the United States. However, this has not been verified or tested in a different cohort from another country.
Materials and Methods: Data of 25 140 deceased donor kidney transplant recipients (2000-2016) were retrospectively analyzed at a population-cohort level using the United Kingdom transplant registry. Within subgroups of donor sex, associations between recipient sex and death-censored graft survival were assessed for the cohort as a whole and within recipient age subgroups.
Results: No differences in graft survival were detected between female versus male recipients of male donor kidneys (adjusted hazard ratio: 1.05; P = .227). However, a significant interaction between the age and sex of recipients was identified (P = .007). Female recipients aged 25 to 44 years had significantly shorter graft survival than male recipients (adjusted hazard ratio: 1.27; P = .003), but this effect was reversed in recipients who were 45 years or older (adjusted hazard ratio: 0.94; P = .258), where there was a nonsignificant tendency for longer graft survival in females. No such effect was observed in the subgroup of female donor transplants, with no significant difference between female versus male recipients overall (adjusted hazard ratio: 1.02; P = .638) and no significant interaction with age (P = .470).
Conclusions: Graft survival is influenced by the combination of recipient age and sex in recipients of male donor kidneys only. This work demonstrates findings broadly similar to published reports but presents differences in observed effect sizes among certain subgroups. Our research suggests further work is warranted to explore personalized approaches to age- and sex-adapted immunosuppression.
Key words : Older donors, Renal transplant
Studies comparing graft outcomes between male and female kidney transplant recipients have shown conflicting results. Analyses of pediatric or young adult recipients have consistently shown poorer outcomes for female versus male recipients,1-5 whereas reciprocal analyses in older adults have shown equivalent or superior survival outcomes for females versus male recipients.6-8 This discrepancy raises the possibility that sex differences in kidney graft survival may be influenced by recipient age. To investigate this hypothesis further, Lepeytre and colleagues9 evaluated 159 417 patients registered in the Scientific Registry of Transplant Recipients who received their first deceased donor kidney transplant between 1995 and 2013. They found graft survival to be significantly shorter in female versus male recipients of male donor kidneys. However, the size of this effect was found to decrease with recipient age, from an adjusted hazard ratio (aHR) of 1.51 (95% CI, 1.19-1.90) for recipients from 0 to 14 years old to 1.05 (95% CI, 1.01-1.09) in those 45 years and older. A different relationship was observed for female donor organs, in which female recipients had a significantly shorter graft survival than male recipients in age group from 15 to 24 years (aHR: 1.28; 95% CI, 1.06-1.53) but a significantly longer graft survival when age 45 years and older (aHR: 0.95; 95% CI, 0.91-0.99). The authors hypothesized this effect could relate to several factors, including sex-determined minor histocompatibility antigens, influence of sex hormones, medication adherence, and differences in body size.
As the first study to assess differences in graft outcomes between female and male recipients across the entire age spectrum, the data from Lepeytre and colleagues9 are important because the data raise a suggestion of age- and sex-specific immunosuppression regimens and/or therapeutic targets after kidney transplant. However, it is important to validate the findings from Lepeytre and colleagues and to verify using a separate population cohort. For example, long-term outcomes beyond the first year after kidney transplant in the United States are inferior to other established kidney transplant programs, such as in the United Kingdom, Australia, and New Zealand, by approximately 25%.10 Therefore, it is unclear whether the findings of Lepeytre and colleagues are generalizable to countries outside the United States. Here, we aimed to replicate the analysis of Lepeytre and colleagues, using a cohort of patients from the United Kingdom, to investigate whether the association between recipient sex and death-censored graft failure is influenced by recipient age and donor sex.
Materials and Methods
We undertook a retrospective, observational population-cohort study of all patients who received a single-organ, primary, deceased donor kidney transplant in the United Kingdom between January 1, 2000 and December 31, 2016 (with follow-up data to August 13, 2018). Data were obtained from the United Kingdom (UK) Transplant Registry, held by NHS Blood and Transplant, to which every kidney transplant center within the United Kingdom is mandated to submit demographic and clinical data for each transplant performed. This study was reported in accordance with the STROBE checklist for reporting of observational studies (checklist available upon request).
Data extracted from the UK Transplant Registry included donor-, recipient- and transplant-related variables. Donor-related factors included age, sex, ethnicity (defined as White, Black, South Asian, mixed/other), body mass index (BMI), terminal creatinine, cause of death (intracranial hemorrhage, hypoxic brain damage, trauma, other), donor type (donation after brain death or donation after cardiac death [DCD]), medical comorbidities (diabetes, hypertension, smoking), hepatitis C, donor risk index, and cytomegalovirus (CMV) serostatus. Recipient-related factors included age, sex, ethnicity, BMI, diabetes, hepatitis C, CMV serostatus, wait list time, and dialysis status. Transplant-related factors included transplant year, cold ischemic time (CIT), HLA mismatch (level 1 [HLA mismatch 0], level 2 [HLA mismatch 0 DR and 0/1 B], level 3 [HLA mismatch 0 DR and 2B, or 1 DR and 0/1 B], and level 4 [1 DR and 2B, or 2 DR]), antibody-incompatible transplant, and calculated reaction frequency (CRF).
The primary exposure of interest was recipient sex. To assess how the effect of recipient sex differed by recipient age, the interaction between these variables was also assessed, with age divided into categories as per Lepeytre and colleagues (ie, 0-14, 15-24, 25-44, and 45+ years).
The primary outcome of interest was death-censored graft survival, in which graft loss was treated as the event, and patients were censored either at death with a functioning graft or at the end of follow-up. The secondary outcome was overall graft survival, which used death or graft loss as a composite outcome, censoring patients at the end of follow-up.
For the analysis, the cohort was divided by donor sex, with analyses performed separately for male and female donors. Initially, a range of demographic factors was compared between male and female recipients. Continuous variables are reported as medians with interquartile ranges, with comparisons between groups performed with the Mann-Whitney test. Ordinal variables were also assessed with Mann-Whitney tests, whereas Fisher exact or chi-square tests were used for nominal variables with 2 or more than 2 categories, respectively. Comparisons across the categories of recipient age were then performed using Jonckheere-Terpstra tests for ordinal or continuous factors, with Mann-Whitney or Kruskal-Wallis tests used to compare between nominal factors, treating the age category as the dependent variable.
Death-censored graft survival was then compared between male and female recipients. Initially, univariable Cox regression models were produced to measure the overall effects of recipient sex. These were then extended to include recipient sex and a recipient age × sex interaction term, to test whether the effects of recipient sex differed with recipient age. In these models, recipient age was treated as a time-dependent covariate to account for the fact that recipients would move between age categories over the period of follow-up. The model was then evaluated within each of the categories of recipient age to measure the effect of recipient sex within each of these subgroups.
Adjusted models were then produced, in order to assess the effects of recipient age and sex, after accounting for the effects of potentially confounding factors. These models took the same form as those described previously but used a backward stepwise approach to select other independent predictors of death-censored graft survival. To improve model fit, continuous variables were categorized before the analysis was started, based on the quartiles of the distribution or using the standard categories for BMI. Cox regression models used listwise deletion of cases with missing data. As a result, factors with substantial quantities of missing data, namely recipient BMI and recipient HCV serostatus, were excluded from the initial analyses, to maximize the included sample size. Factors selected for inclusion by the stepwise procedure were then entered into a new model to prevent exclusions due to missing data for factors not in the final model. Sensitivity analyses were then performed, in which the variables that were initially excluded (recipient BMI and HCV serostatus) were entered into the final models, to ensure that their exclusion had not unduly influenced the results. For these variables, missing values were replaced with the mean, in order to maximize the included sample size. The unadjusted and adjusted analyses were then repeated for the secondary outcome of overall graft survival.
All analyses were performed using IBM SPSS version 22, with P < .05 deemed to be indicative of statistical significance throughout.
Formal ethical approval was not required for this study as no patient identifiable data were used. This analysis was undertaken on routinely collected data that had been submitted anonymously to the UK Transplant Registry by national transplant centers. The clinical and research activities being reported are consistent with the Principles of the Declaration of Istanbul as outlined in the Declaration of Istanbul on Organ Trafficking and Transplant Tourism.
Data were available for a total of 25 158 deceased donor transplants. Of these, 18 cases did not have either donor or recipient sex recorded and were excluded, leaving 25 140 transplants for analysis. Male donor kidneys were used in 13 414 transplants (53.4%), of which 8403 (62.6%) had male recipients. Of female donor kidneys, 7287 of 11 726 (62.1%) were transplanted into male recipients.
Demographics by recipient sex
Donor factors were generally similar in male and female recipients, with some small but statistically significant differences detected (Table 1). The largest observed difference was in the donor type, with male recipients being significantly more likely than female recipients to receive a DCD organ, regardless of whether the donor was male (32.5% vs 28.0%; P < .001) or female (24.4% vs 21.1%; P < .001). Analyses of recipient- and transplant-related factors (Table 2) found male recipients to be significantly older, with higher BMIs and higher rates of diabetes and hemodialysis at transplant, whereas female recipients had higher rates of CMV positivity and longer wait times. Female recipients also had longer CITs and higher CRFs at transplant, but male recipients had significantly higher mismatch levels at transplant.
Factors associated with recipient age at transplant
Recipient age was unavailable for a single transplant, with the remainder being categorized as age 0 to 14 (n = 771), 15 to 24 (n = 1272), 25 to 44 (n = 7602), and 45+ (n = 15 494) years at time of transplant. A full breakdown of the numbers of transplants within each combination of the age and sex categories is reported in Table 3. Older recipients tended to receive kidneys from higher risk donors, with the median donor risk index for recipients aged 0 to 14 versus 45+ years increasing from 0.82 to 1.09 (P < .001) in male donors and 0.97 to 1.12 (P < .001) in female donors. Correspondingly, the average donor age and BMI and rates of donor diabetes, hypertension, CMV, and DCD donation were also found to increase significantly with recipient age (Table 4 and Table 5). Of the recipient factors considered (Table 6 and Table 7), older recipients had significantly higher BMIs and rates of diabetes and CMV. Older recipients were also significantly more likely to be on hemodialysis at the time of transplant and had significantly longer wait times and higher mismatch levels. However, older recipients also had significant shorter CITs and lower CRFs.
Death-censored graft survival
Graft outcome data were available for 25 098 recipients. Over a median of 5.0 years of follow-up (interquartile range, 2.0-9.1), there were 5046 graft failures, giving Kaplan-Meier estimated death-censored graft survival rates of 92.8%, 88.9%, 84.5%, and 73.9% at 1, 3, 5, and 10 years, respectively.
Comparisons by recipient sex are reported in Table 8. On unadjusted analysis of male donor transplants, death-censored graft survival was not found to differ significantly between female and male recipients (HR: 1.00; 95% CI, 0.92-1.08; P = .977). However, a significant interaction between recipient age and sex was detected (P = .009), suggesting that the effect of recipient sex was age-dependent. Subgroup analysis within the categories of recipient age found a tendency for female recipients to have shorter death-censored graft survival than males in age groups of 0 to 14, 15 to 25, and 25 to 44 years, although only the latter was statistically significant (HR: 1.16; 95% CI, 1.01-1.34; P = .037). However, for recipients aged 45+ years, this effect was reversed, with females found to have significantly longer graft survival than males (HR: 0.90, 95% CI, 0.81-0.99; P = .036). This effect is visualized in Figure 1. For the subgroup of female donor transplants, no overall effect of recipient sex was observed (HR: 1.07; 95% CI, 0.98-1.16), and there was no significant interaction shown between recipient age and sex (P = .119) on unadjusted analysis.
Adjusted analyses were then performed to account for the effect of potentially confounding factors. All factors from Table 1 and Table 2 were considered for inclusion in these analyses, with the exception of recipient BMI and HCV serostatus, due to excessive quantities of missing data (42.7% and 25.2% of cases, respectively). A complete list of the included factors, as well as the categorizations used, is shown in Table 9, and the full models are shown in Table 10 and Table 11. The adjusted analyses returned results similar to those of the unadjusted analyses previously described, with no significant overall effect of recipient sex, regardless of the sex of the donor; in addition, a significant interaction was observed between recipient age and sex in transplants from male (P = .007) but not female (P = .470) organ donors (Table 8).
Sensitivity analyses were then performed, in which recipient BMI and HCV serostatus were entered into the final models. Missing data for these factors were replaced with the mean, namely, a BMI of 26 kg/m2 and a negative HCV serostatus. After we accounted for these 2 additional factors, the adjusted effects of recipient sex remained consistent with those previously described (Table 12).
Overall graft survival
A total of 3882 patients died during the follow-up period, of whom 2826 had a functioning graft at the time of death. For the composite outcome of overall graft survival, Kaplan-Meier estimated rates were 90.5%, 84.4%, 77.8%, and 60.0% at 1, 3, 5, and 10 years, respectively. Analyses of this outcome, with respect to recipient age and sex, gave results that were consistent with those previously reported for the outcome of death-censored graft survival; these results are summarized in Table 13, Table 14, and Table 15.
Comparison with results from Lepeytre and colleagues9
For the primary outcome of death-censored graft survival, the interaction between recipient age and sex from the adjusted analysis is visualized alongside the results reported by Lepeytre and colleagues9 in Figure 2. For transplants of male donor kidneys, both studies showed a similar trend, with younger female recipients having shorter death-censored graft survival than males, with the size of this difference declining with recipient age. However, although Lepeytre and colleagues observed significantly shorter death-censored graft survival in female recipients to persist in the 45+ year age group (aHR: 1.05; 95% CI, 1.01-1.09), our study found the effect to be reversed in this group, although this did not achieve statistical significance (aHR: 0.94; 95% CI, 0.84-1.05; P = .258).
Within the subgroup of female donor kidneys, although our study found no significant interaction between recipient age and sex on adjusted analysis (P = .470), the general trend across the recipient age categories was similar to that reported by Lepeytre and colleagues. For both studies, an inverted U-shaped relationship was observed, with death-censored graft survival being similar or longer for female versus male recipients aged 0 to 14 years or 45+ years but shorter in female versus male recipients aged 15 to 44 years. However, although Lepeytre and colleagues found the peak of poorer outcomes in female recipients to occur in the 15- to 24-year age category, our study found this to be later, at 25 to 44 years of age.
In this population-cohort analysis of 25 140 patients, we found that, for recipients of male donor kidneys, the association between recipient sex and graft survival was significantly age-dependent. Specifically, female recipients who were less than 45 years old tended to have shorter graft survival than male recipients, with the size of this effect declining with age and being reversed in recipients who were 45+ years old. In those 45+ years old, female recipients tended to have longer graft survival than males. In contrast, neither the overall effect of recipient sex nor the recipient age × sex interaction was found to be significant in female donor transplants.
The work from Lepeytre and colleagues, which evaluated 159 417 recipients of deceased donor kidney transplants within the Scientific Registry of Transplant Recipients between 1995 and 2013, was the first to specifically investigate the influence of recipient age on sex differences in graft survival.9 Our analyses aimed to validate these findings and so used a similar statistical methodology, with the same categories of recipient age. The conclusions of both studies were broadly similar, although there were some key differences. Within the subgroup of male donor transplants, Lepeytre and colleagues found graft survival to be inferior in female recipients, relative to males, across the whole range of age categories, whereas our study found marginally superior graft survival in female recipients aged 45+ years. For female donor organs, Lepeytre and colleagues found a significant recipient age × sex interaction, with poorer graft survival in female versus male recipients, peaking in the 15- to 24-year age group. However, our study found no significant interaction term, with the peak of poorer graft survival in female recipients occurring later at 25 to 44 years.
One potential explanation for these discrepancies is the lower statistical power in our study as a result of our smaller sample size. Although our study included a large number of patients from a national registry, the sample size analyzed by Lepeytre and colleagues9 was over 6 times larger, thus allowing for the detection of smaller effect sizes. This was a particular issue for the younger recipient age categories, as they consisted of a small proportion of the cohort as a whole, resulting in wide confidence intervals and low statistical power. Another explanation for the differences could be because of variations in demographics of the 2 cohorts. For example, recipients in our study were predominantly of White ethnicity (78%), whereas the proportion of White recipients in Lepeytre and colleagues ranged between 42% and 55% across the recipient age categories. Such differences may be confounders in the relationship between recipient age/sex and graft outcomes and so could have contributed to the discrepancies in the findings between the 2 studies. Finally, as previously discussed, graft survival outcomes in the United States beyond the first year are inferior to outcomes in the United Kingdom. The region-specific causes for this discrepancy may also have an impact on the relationship between recipient age/sex and graft outcomes, perhaps explaining the discrepancies in the results between the 2 studies.
Many studies have considered the role of both sex-specific histocompatibility antigens and sex-related differences in kidney size to explain the inferior graft outcomes observed in donor-recipient sex-mismatched transplants. However, our finding of age-specific differences in the effects of sex on graft survival in recipients of male donor kidneys indicates that several other contributing factors may exist. Sex differences have been observed in both kidney transplant11 and non-kidney transplant settings,12 and it has been postulated that sex hormones may have an influence on age-specific sex differences in graft survival. In female recipients of male donor kidneys, some females develop antibodies against male-specific H-Y antigens that can be associated with acute rejection.13 Estrogen receptors are present on most immune cells, and, although evidence for the specific effect of estrogen on immune cells is conflicting, estrogen is generally considered to be immunostimulatory.14 Lepeytre and colleagues9 hypothesized that estrogen exposure in women of childbearing age may exacerbate the immune response against H-Y antigens of male donor kidneys. In our analysis, although not statistically significant, effect sizes were consistently inferior for female recipients of childbearing age compared with male recipients.
A potential confounder to this explanation could be better adherence. Existing literature suggests that adolescent and young adult transplant recipients are at the greatest risk of nonadherence compared with any other age15-17; however younger adult females (aged 17-24 years) have been shown to have lower nonadherence rates than their counterpart males in the randomized TAKE-IT study using electronic pillboxes.18 Thus, we can speculate that any negative effect of estrogen exposure on graft survival could be obscured by a bigger positive effect of medication adherence within this age group. Pharmacokinetics of immunosuppression also differ between sexes,11 and this is a potentially confounding factor that we did not account for in our analysis, as data were not available for immunosuppression burden (along with medication adherence or sex hormone levels). Such detailed phenotypes require well-characterized clinical cohorts to be investigated further, likely as part of a prospective study.
This work is based on data from the national kidney transplant data set, maintained by the UK Transplant Registry, which contains data on a large number of outcome variables with good completeness of data. However, there are several limitations in our analysis that should be acknowledged to contextualize our research findings. First, retrospective population cohort studies may suffer from bias and confounding factors that are inherent with all observational studies of this nature. Furthermore, our analysis used registry data specific to the United Kingdom to demonstrate an interaction between recipient age and donor sex among recipients with male donors. Although caution should be taken in the translation of our research findings to population cohorts elsewhere, this is more pertinent when comparing data from the Unites States to elsewhere and so justifies the need for this replication work. Finally, we can only establish association rather than causation; therefore, further experimental studies are required to better understand the role of biological or sociological effects on observed sex differences related to kidney graft survival.
Our study demonstrated that graft survival differences for male and female kidney transplant recipients depend on donor sex and/or recipient age. With kidneys from male donors, younger female recipients have shorter graft survival than males, but the opposite effect is observed in older recipients. No significant interaction effect between recipient sex and age was observed for female donor organs. This work returns findings broadly similar to those by recent work from Lepeytre and colleagues9 by demonstrating an interaction between donor-recipient sex and recipient age, although observed effect sizes were different among certain subgroups. We believe further work is warranted to investigate these disparities and explore more personalized approaches to age- and sex-adapted immunosuppression.
Volume : 19
Issue : 3
Pages : 190 - 203
DOI : 10.6002/ect.2020.0112
From the 1College of Medical and Dental Sciences, University of Birmingham; the
2Institute of Translational Medicine, Queen Elizabeth Hospital; the
of Nephrology and Transplantation, Queen Elizabeth Hospital; and the 4Institute
of Immunology and Immunotherapy, University of Birmingham, Birmingham, United
Acknowledgements: The authors thank NHS Blood and Transplant for provision of data.
Data that support the findings of this study are available from NHS Blood and Transplant upon request. The authors have not received any funding or grants in support of the presented research or for the preparation of this work and have no declarations of potential conflicts of interest.
Author contributions: GM, ZGM, JH, JN, and AS contributed to conception or design, or analysis and interpretation of data, or both; GM, JN, and AS contributed to drafting or revision of manuscript; JH, JN, and AS provided intellectual content of critical importance to the
work described; and GM, ZGM, JH, JN and AS provided final approval of the version to be published.
Corresponding author: Adnan Sharif, Department of Nephrology and Transplantation, Queen Elizabeth Hospital, Edgbaston, Birmingham, B15 2WB, UK
Phone: +0121 371 5861
Table 1. Association Between Donor Factors and Recipient Sex
Table 2. Association Between Recipient and Transplant Factors and Recipient Sex
Table 3. Distribution of Cases by Donor/Recipient Sex and Recipient Age
Table 4. Associations Between Donor Factors and Recipient Sex in Male Donors
Table 5. Associations Between Donor Factors and Recipient Sex in Female Donors
Table 6. Associations Between Recipient/Transplant Factors and Recipient Age in Male Donors
Table 7. Associations Between Recipient/Transplant Factors and Recipient Age in Female Donors
Table 8. Death-Censored Graft Loss in Female Versus Male Recipients by Donor Sex and Current Recipient Age
Table 9. Factors Considered for Inclusion in the Multivariable Models for the Adjusted Analyses
Table 10. Adjusted Analyses of Death-Censored Graft Loss (part 1)
Table 11. Adjusted Analyses of Death-Censored Graft Loss (part 2)
Table 12. Summary of the Sensitivity Analysis of Death-Censored Graft Loss in Female Versus Male Recipients by Donor Sex and Current Recipient Age)
Table 13. Overall Graft Loss in Female Versus Male Recipients by Donor Sex and Current Recipient Age
Table 14. Adjusted Analyses of Overall Graft Loss (part 1)
Table 15. Adjusted Analyses of Overall Graft Loss (part 2)
Figure 1. Kaplan-Meier Curves of Death-Censored Graft Survival by Recipient Age and Sex for Recipients of Organs From Male Donors
Figure 2. Adjusted Hazard Ratios for Death-Censored Graft Failure in Female Versus Male Recipients by Current Age and Sex