;on behalf of the Korean Organ Transplantation Registry Study Group (KOTRY) Study GroupIntroduction
Kidney transplantation (KT) is the optimal renal replacement therapy for patients with end-stage kidney disease (ESKD) [1]. KT requires a donor; however, the number of available kidney donors, including deceased donors, is limited, and the number of patients with ESKD continues to rise [2]. Consequently, the number of patients on the waiting list for KT has been steadily increasing. Living-donor transplantation has several advantages over deceased-donor transplantation, including a higher survival rate. Nevertheless, a significant number of patients receive kidneys via deceased-donor transplantation [2].
The beneficial effects of human leukocyte antigen (HLA) matching in KT have been well-documented across different eras, from the pre-cyclosporine period to the calcineurin inhibitor era [3–11]. As early as 1987, the United States’ allocation policy, governed by the United Network for Organ Sharing (UNOS), mandated that well-matched kidneys be shared on a national basis. This evidence-based approach to HLA matching has led to a consistent policy of kidney sharing among transplant centers across the country. In the Republic of Korea, kidney allocation has been conducted in accordance with these policies since the establishment of KONOS (Korean Network for Organ Sharing) in the 2000s.
However, the accumulation of patient transplantation experience and advancements in immunosuppressive therapies have significantly improved the survival rates of transplanted kidneys. Recently, Casey et al. [12] reported that the survival advantage of zero mismatches in living-donor transplants has diminished. Therefore, using nationwide data, we aimed to compare the clinical outcomes of patients with zero HLA mismatches with those of patients without zero mismatches who underwent KT.
Methods
Study population
We extracted data from 10,391 patients who underwent KT between May 2014 and December 2022 as recorded in the Korean Organ Transplantation Registry (KOTRY). The KOTRY is a prospective observational cohort study that included transplant information from 42 transplantation centers in Korea [13]. KT recipients in the KOTRY database were enrolled at the time of transplant surgery and have been undergoing regular follow-ups since then. We excluded patients who underwent living-donor KT (n = 7,041). Among the 3,350 deceased-donor KT recipients, 361 patients received zero HLA-mismatched KT (zero group), while 2,989 patients received KT with >0 HLA mismatches (non-zero group) (Fig. 1).
Data collection
Demographic data of the donors and recipients were collected from the KOTRY database. The numbers of HLA mismatches at loci A, B, and DR were recorded. Pretransplant peak panel reactive antibody (PRA) levels were categorized as 0% to 50%, 51% to 80%, and 81% to 100%. As PRA is a critical factor for posttransplant biopsy-proven rejection (BPR), missing PRA values were treated as a separate category, accounting for approximately 30% of the study population. Donor factors such as age, sex, body mass index (BMI), history of hypertension, history of diabetes mellitus (DM), and serum creatinine level at donation were also collected. According to the KOTRY data collection interval, serum creatinine level, postoperative outcome information, immunosuppressant regimens, such as steroid, tacrolimus, and mycophenolate mofetil (MMF) use, were collected at discharge, 6 months, and annually thereafter. Graft function was evaluated using the estimated glomerular filtration rate (eGFR), calculated by the CKD-EPI (Chronic Kidney Disease Epidemiology Collaboration) equation [14].
Immunosuppression
All patients received induction therapy with basiliximab or antithymocyte globulin (ATG), determined by donor/recipient characteristics, postoperative progress, and institutional protocols. In Korea, most institutions use a maintenance immunosuppression regimen typically initiated with tacrolimus (0.1 mg/kg twice daily) or cyclosporine (5 mg/kg twice daily), prednisolone (tapered to 5–10 mg/day within 3 weeks), and either MMF (1,000–2,000 mg/day) or mycophenolate sodium (720–1,440 mg/day). The specific regimen was adjusted according to institutional protocols and individual patient conditions.
Outcomes
The primary outcome measure was death-censored graft failure. Secondary outcomes included patient death, BPR, admission owing to infection, cancer, new-onset DM after transplantation (NODAT), and delayed graft function (DGF). Infection was recorded when patients were admitted for the treatment of confirmed pathogens. However, the specific pathogen and the type of quantitative measurement (e.g., quantitative polymerase chain reaction for cytomegalovirus) were not recorded in KOTRY until January 2017. The main types of infections include urinary tract infection, bacterial pneumonia, bacteremia, viral infection, viral pneumonia, fungal infection, and Pneumocystis jiroveci pneumonia. Positivity for each pathogen was determined according to the standards of each center. Graft failure was defined as the return to dialysis or the need for retransplantation, with patient death censored during the analysis of graft failure. We defined BPR as cases where each institution reported rejection based on biopsy results. We defined NODAT based on cases where each center reported the occurrence of DM after transplantation to the KOTRY registry. To account for early posttransplant hyperglycemia, we excluded patients who were diagnosed with DM only during the early posttransplant period but were later reported as non-diabetic during follow-up. NODAT analysis was conducted only in patients without pretransplant DM. DGF was defined as the need for dialysis within 7 days after KT.
Statistical analysis
Baseline characteristics were compared between the zero group and non-zero group before and after propensity score matching (PSM) using the independent t test or chi-square test, as appropriate. Results are presented as mean ± standard deviation or numbers and percentages, depending on the type of variable. A 1:1 PSM was performed between the zero and non-zero groups using the nearest neighbor method, with a caliper set at 0.2 standard deviations. All available baseline factors were included as covariates. The transplant year, PRA level, dialysis duration, cause of ESKD, DM, donor age, donor DM, donor creatinine level at transplantation, ATG use, and cold ischemic time were used to generate the propensity scores. After PSM, covariate balance was evaluated by calculating the standardized mean difference between the groups [15] (Supplementary Table 1, available online). For each outcome, hazard ratios (HRs) were estimated in both matched and unmatched data using multivariable Cox regression analysis. HRs obtained after PSM were further adjusted for variables that remained significantly different between the two groups after matching. Since PSM reduced the sample size, we also conducted the same analysis without matching by adjusting for variables that were significantly different between the two groups to compare the results. The multivariable Cox regression analysis was adjusted for baseline characteristics that were statistically different between the two groups before PSM. For all outcomes, death was considered a competing risk, and regression analyses were performed using Fine and Gray’s model. Kaplan-Meier analysis was used to compare the cumulative probability of posttransplant events, and statistical significance was assessed using the log-rank test. All analyses were performed using the R statistical package (version 4.2.0; R Foundation for Statistical Computing) for Windows. Statistical significance was set at p < 0.05.
The Institutional Review Board of each center approved this study (CR321351), and written informed consent was obtained from all patients before enrollment. The study was conducted in accordance with the principles of the Declaration of Helsinki.
Results
Baseline characteristics
Supplementary Table 1 (available online) shows the baseline characteristics of the zero and non-zero groups before and after matching. Before matching, 361 patients (10.8%) received zero HLA mismatch deceased-donor KT, whereas 2,989 patients (89.2%) received >0 HLA mismatch deceased-donor KT. Age, sex, BMI, and cause of ESKD were similar between the groups. However, the zero group had a higher rate of retransplantation (8.6% vs. 7.6% in the zero and non-zero groups, respectively; p = 0.04), shorter dialysis duration (72.3 ± 59.2 months vs. 96.4 ± 59.8 months in the zero and non-zero groups, respectively; p < 0.001) and longer cold ischemic time (338.2 ± 142.1 minutes vs. 283.3 ± 135.8 minutes in the zero and non-zero groups, respectively; p < 0.001). The PRA was significantly different between the two groups (p = 0.03), with the 81st to 100th percentile PRA values being higher in the zero group than in the non-zero group (8.0% vs. 4.9%). Additionally, the proportion of KTs received between 2019 and 2022 was higher in the zero group than in the non-zero group (57.6% vs. 44.8%). Several donor risk factors, such as age at donation (47.4 ± 14.2 years vs. 50.0 ± 16.7 years, p = 0.04) and creatinine level at donation (1.4 ± 1.2 mg/dL vs.1.6 ± 1.4 mg/dL, p = 0.01), were lower in the zero group than in the non-zero group.
Through PSM, 276 patients from the zero group were matched with 276 patients from the non-zero group, achieving adequate balance. After matching, baseline covariates were generally well balanced between the two groups, except for cardiovascular disease history (14.9% vs. 22.1%, p = 0.04) and tacrolimus as initial immunosuppressant regimen (97.1% vs. 99.6%, p = 0.04) (Table 1). The number of HLA mismatches before and after matching is shown in the Supplementary Table 2 (available online).
Immunosuppression (induction and maintenance treatment)
The mean total ATG dose was similar between the two groups (4.8 ± 2.4 mg/kg vs. 5.0± 2.3 mg/kg in the zero and non-zero groups, respectively; p = 0.53) (Supplementary Table 3, available online). All patients who received basiliximab were administered two doses of 20 mg basiliximab on postoperative days 0 and 4. Supplementary Fig. 1 (available online) shows a comparison of maintenance immunosuppressant use between the two groups. Most patients in both groups received a combination of tacrolimus (TAC), MMF, and steroids at discharge, and no significant differences were observed between the regimens. At 6 months, the proportions of TAC, MMF, and steroid regimens declined, whereas TAC and steroid regimens became more common in both groups. The mean tacrolimus trough levels were similar between the two groups, measured at discharge (7.9 ± 3.0 ng/mL vs. 8.0 ± 2.5 ng/mL, p = 0.79), 6 months (6.6 ± 2.5 ng/mL vs. 6.5 ± 2.3 ng/mL, p = 0.83), 12 months (6.3 ± 2.2 ng/mL vs. 6.5 ± 2.3 ng/mL, p = 0.23), and 24 months (6.2 ± 2.4 ng/mL vs. 6.5 ± 2.6 ng/mL, p = 0.33). Likewise, mean MMF-equivalent doses did not differ significantly between two groups, measured at discharge (1,346 ± 463 mg vs. 1,281 ± 372 mg, p=0.41), 6 months (934 ± 316 mg vs. 875 ± 330 mg, p = 0.30), 12 months (922 ± 298 mg vs. 860 ± 327 mg, p = 0.28), and 24 months (862 ± 303 mg vs. 870 ± 334 mg, p = 0.90) (Supplementary Figs. 2, 3; available online).
Clinical outcomes (propensity score matching matched analysis)
DGF occurred in 46 patients, with 15 cases in the zero group and 31 cases in the non-zero group, showing a significant difference in incidence rates between the groups (p = 0.02).
During a median follow-up duration of 38.4 ± 28.8 months, 26 patients (4.7%) died. The most common cause of death was infection (n = 9, 34.6%). Details on the causes of death are provided in the Supplementary Table 4 (available online). The cumulative probability of patient death in matched patients was 2.2% at 1 year and 3.1% at 3 years in the zero group, compared to 3.7% at 1 year and 8.6% at 3 years in the non-zero group (Fig. 2A). In multivariable adjusted Cox regression analysis showed that non-zero mismatch was not significantly risk factor for patient’s death (HR, 2.09; 95% confidence interval [CI], 0.87–5.00; p = 0.10). A total of 21 patients (3.8%) experienced death-censored graft failure, and five patients (0.9%) died with a functioning graft. The cumulative probability of death-censored graft failure was 1.9% at 1 year and 3.6% at 3 years in the zero group, compared to 1.8% at 1 year and 4.3% at 3 years in the non-zero group (HR, 1.63; 95% CI, 0.72–3.69; p = 0.24) (Fig. 2B). The cumulative probability of infection was 22.1% at 1 year and 34.8% at 3 years after operation in the zero group, compared to 32.6% at 1 year and 42.8% at 3 years in the non-zero group (Fig. 2C) (HR, 1.38; 95% CI, 1.02–1.86; p = 0.03). The cumulative probability of NODAT was 12.2% at 1 year and 17.1% at 3 years after operation in the zero group, compared with 21.9% at 1 year and 25.1% at 3 years in the non-zero group (HR, 1.43; 95% CI, 0.82–2.50; p = 0.21) (Supplementary Fig. 4, available online). The cumulative probability of BPR was 2.3% at 1 year and 8.3% at 3 years after operation in the zero group, compared to 3.1% at 1 year and 12.8% at 3 years in the non-zero group (HR, 1.62; 95% CI, 0.96–2.76; p = 0.07) (Fig. 2D). Subgroup analysis showed that the non-zero group was associated with a higher risk of BPR among patients with longer cold ischemic time ≥300 minutes, high BMI ≥25 kg/m2, longer dialysis duration ≥60 months, and those who received basiliximab as an induction agent (Supplementary Table 5, available online).
The cumulative probability of antibody-mediated rejection (ABMR) was significantly higher in the non-zero group than in the zero group (HR, 4.48; 95% CI, 1.52–13.25; p = 0.001). However, the cumulative probabilities of chronic active ABMR and chronic T-cell–mediated rejection did not differ significantly between the two groups (Supplementary Table 6, available online). Similar results were reported in unmatched data with multivariable Cox regression analysis (Table 1).
The incidence of ABMR was higher in patients with HLA-DR mismatches compared to those with HLA-A or HLA-B mismatches (Supplementary Table 7, available online).
Graft function, measured by eGFR, remained similar throughout the follow-up period in both groups, with an approximate eGFR of 60 mL/min/1.73 m2 (Fig. 3).
Discussion
Using matched analysis from a Korean multicenter registry, this study comprehensively compared clinical outcomes between two HLA mismatch groups: zero mismatch and non-zero mismatch. Among Korean KT recipients, the zero group did not demonstrate superiority over the non-zero group in terms of graft failure, graft function, or all-cause death throughout the study period. However, the zero group exhibited a lower cumulative risk of BPR compared to the non-zero group. Subgroup analysis revealed that this trend was particularly prominent in longer cold ischemic time (>300 minutes), high BMI (≥25 kg/m2), longer dialysis duration (>5 years), and those who used basiliximab as induction therapy. Additionally, the zero mismatch group was significantly associated with a lower cumulative risk of infection and ABMR compared to the non-zero mismatch group.
In this era of modern immunosuppression, the effectiveness of HLA matching as a predictive tool for graft survival has been investigated. This aligns with recent research indicating that with current immunosuppressive protocols, the emphasis on HLA matching might be less critical than previously believed, and current kidney allocation guideline recommendations are inconsistent. In the United Kingdom kidney allocation scheme, HLA-A matching is no longer considered [16]. The United States kidney allocation system was modified to eliminate HLA-A and HLA-B similarity [17,18]. The European Renal Best Practice Transplantation Guidelines recommend HLA-A, -B, and -DR; however, they assign more weight to the HLA-DR locus [19]. The French system adds HLA-DQ matching to the allocation scheme [20]. Nevertheless, allocation policies with HLA-A, -B, and -DR matching are still considered the gold standard for general allocation agencies. Despite the theoretical advantages of zero HLA mismatch, the current zero mismatch allocation system requires modifications.
Despite its historical significance, there has been a noticeable decline in recent research focusing specifically on HLA mismatches and their implications in the context of KT. A UNOS registry analysis of 189,141 KT recipients was conducted to assess the importance of HLA matching; however, the study included patients who underwent transplantation before 2013 [21]. Similar findings were reported in an analysis of the Australian and New Zealand Dialysis and Transplant Registry (ANZDATA), which included 8,036 KT recipients and produced comparable results; however, this analysis also included only those who received KT before 2009 [22]. Furthermore, some meta-analyses showed that most of these studies were performed before 2016, and these analyses revealed that as the analysis becomes more recent, the efficacy of HLA mismatching appears to be gradually decreasing [9,10]. Therefore, our study provides valuable contemporary data that reflect current clinical practices, highlighting the need for ongoing research to better understand the evolving role of HLA mismatches in the era of advanced transplant medicine.
ABMR is the most common cause of immune-mediated allograft failure following KT [23]. Although the treatment guidelines are based on low-level evidence [24], advancements in treatment strategies have led to improved management and outcomes in these cases [25,26]. Nevertheless, while ABMR remains a significant concern, the effectiveness of contemporary therapies, including tailored immunosuppressive regimens, may mitigate its impact on the overall outcomes. Analysis of ANZDATA and UNOS data showed that a DQ mismatch is associated with acute rejection and graft survival, independent of HLA-A and B mismatches [27,28]. Interestingly, even in HLA-identical DR pairs, graft loss due to ABMR has been reported, indicating that HLA-A, -B, and -DR matching alone do not fully account for the occurrence of ABMR or BPR [29]. In our study, despite the occurrence of BPR and ABMR in the zero group, these phenomena may have been influenced by factors such as DQ mismatch. Moreover, mismatches at non-HLA loci and other immunological factors may play a significant role in graft outcomes, highlighting the need for further research to better understand these complex interactions [30]. This underscores the importance of individualized treatment approaches over traditional HLA matching.
This study had some limitations. First, owing to the medium duration of follow-up, our results do not capture the long-term (over 10 years) consequences of the comparison between the zero and non-zero groups. Second, the small sample size of the zero group and the presence of unmeasured confounders limited the robustness of our findings. Additionally, discrepancies in immunosuppressive treatment strategies among hospitals participating in the KOTRY study may have influenced the outcomes. Furthermore, some kidney biopsy results were missing, which necessitates caution when interpreting the occurrence of rejection types, such as ABMR and acute T-cell–mediated rejection. Finally, all study participants were Korean; therefore, our results may not be generalizable to other ethnicities.
In conclusion, although HLA matching has historically been the cornerstone of KT, its role in the modern clinical landscape is evolving. Our study suggests that rigid adherence to HLA matching may be less critical than previously thought, particularly given advancements in immunosuppressive therapies and the introduction of new allocation indices, such as the kidney donor profile index (KDPI). A more nuanced approach that incorporates these factors may be essential for optimizing transplant outcomes.
PDF Links
PubReader
ePub Link
XML Download
Full text via DOI
Download Citation
Supplement table 1
Print
