Evaluating anti-thymocyte globulin induction doses for better allograft and patient survival in Asian kidney transplant recipients

Anti-thymocyte globulin (ATG) is currently the most widely prescribed induction regimen for preventing acute rejection after solid organ transplantation. However, the optimal dose of ATG induction regimen in Asian kidney recipients is unclear. Using the Korean Organ Transplantation Registry, we performed a retrospective cohort study of 4579 adult patients who received renal transplantation in South Korea and divided them into three groups according to the induction regimen: basiliximab group (n = 3655), low-dose ATG group (≤ 4.5 mg/kg; n = 467), and high-dose ATG group (> 4.5 mg/kg; n = 457). We applied the Toolkit for Weighting and Analysis of Nonequivalent Groups (TWANG) package to generate high-quality propensity score weights for intergroup comparisons. During four-year follow-ups, the high-dose ATG group had the highest biopsy-proven acute rejection rate (basiliximab 20.8% vs. low-dose ATG 22.4% vs. high-dose ATG 25.6%; P < 0.001). However, the rates of overall graft failure (4.0% vs. 5.0% vs. 2.6%; P < 0.001) and mortality (1.7% vs. 2.8% vs. 1.0%; P < 0.001) were the lowest in the high-dose ATG group. Our results show that high-dose ATG induction (> 4.5 mg/kg) was superior to basiliximab and low-dose ATG induction in terms of graft and patient survival in Asian patients undergoing kidney transplant.

Clinical outcomes. A total of 1947 kidney transplant recipients had kidney allograft biopsies due to an elevated serum creatinine (n = 773), aggravated proteinuria (n = 40), scheduled protocol (n = 1081), and other reasons (n = 53). Cumulative incidences of biopsy-proven acute rejection (BPAR) were significantly higher in the high-dose ATG group than in the other groups during a four-year follow-up period, although the increase nearly plateaued two years after transplantation (Table 2) (Fig. 1). The rates of acute antibody-mediated rejection (ABMR) and chronic ABMR were the highest in the high-dose ATG group. In contrast, acute T-cell mediated rejection (TCMR) within 1 year post-transplantation was the highest in the low-dose ATG ( Table 3).
The rate of delayed graft function was the highest in the high-dose ATG group (3.40% vs. 2.95% vs. 4.28%, P = 0.009) ( Table 2). In contrast, the cumulative rate of overall graft failure was the lowest in the high-dose ATG group (Fig. 2a). Also, the rate of death-censored graft failure (DCGF) was the lowest in the high-dose ATG group, albeit without statistical significance (P = 0.08; Fig. 2b). The cumulative rate of mortality was the lowest in the high-dose ATG group and the highest in the low-dose ATG group (Fig. 3). Since significant differences in the mortality among groups is identified, the graft survival was re-analyzed using Fine-Gray Subdistribution Hazard Ratio (HR) analysis (Fig. 4). The results of the Fine-Gray Subdistribution HR analysis for DCGF and mortality show a significant increase in mortality risk for the low-dose ATG group compared to both the Basiliximab and high-dose ATG groups, with the highest risk observed when comparing low-dose ATG to high-dose ATG. Additionally, the risk of DCGF was significantly lower in the high-dose ATG group compared to Basiliximab, while there was no significant difference between low-dose ATG and Basiliximab. These findings suggest that patients receiving low-dose ATG may have a higher risk of mortality, and that high-dose ATG may provide better protection against DCGF compared to Basiliximab.
Cytomegalovirus infection was the most common in the high-dose ATG group (Fig. 5a), and hematologic abnormalities including neutropenia and thrombocytopenia were higher in the ATG groups than in the basiliximab group (Fig. 5b). Although there was no significant difference in the incidence of post-transplant malignancy until two years after transplantation, the incidence in the low-dose ATG group was higher thereafter (Fig. 5c).
Furthermore, we evaluated clinical outcomes based on induction regimens used in patients considered at high immunological risk. High immunological risk patients were defined as those who underwent desensitization before kidney transplantation due to the presence of pre-transplant donor-specific antibodies, ABO-incompatible www.nature.com/scientificreports/ KT, or human leukocyte antigen (HLA)-incompatible KT. Out of the 4579 enrolled recipients, 1072 were classified as high-risk immunologically. Among them, 779 recipients were administered basiliximab, while 133 and 160 recipients received low-and high-dose ATG, respectively. Over a four-year follow-up period, the cumulative incidences of BPAR were significantly higher in the low-and high-dose ATG groups compared to the basiliximab group, but no significant difference was observed between the low-and high-dose ATG groups (Table 4).
Conversely, the cumulative rates of overall graft failure and mortality were significantly lower in the high-dose ATG group compared to the low-dose ATG group, while no significant difference was found in the cumulative rate of DCGF between the groups.

Discussion
Using the KOTRY database, we found that the mean, median, 25th percentile, and 75th percentile of the ATG dose in Korean KT recipients were 4.85 mg/kg, 4.50 mg/kg, 4.26 mg/kg, and 5.45 mg/kg, respectively. We also found that high-dose ATG induction (> 4.5 mg/kg) showed superior outcomes in terms of graft and patient survival compared with basiliximab and low-dose ATG (< 4.5 mg/kg), although BPAR was the most common in the high-dose ATG group. To our knowledge, this is the first study to examine the currently used dose of ATG as an induction regimen and to compare the clinical outcomes according to the doses of ATG using a national kidney transplant database in an Asian population. Previous randomized controlled trials compared clinical outcomes according to the doses of ATG as an induction regimen, which ranged from 5 to 10 mg/kg [19][20][21][22][23][24][25] . On the other hand, recent studies examined the efficacies of lower doses of ATG on the depletion of immune cells as well as clinical outcomes. Kho et al. compared the counts of peripheral immune cells after kidney transplantation among 1.5, 3.0, and 6.0 mg/kg ATG induction groups and showed that only the 6.0 mg/kg ATG induction group had a significantly lower T cell count compared with  www.nature.com/scientificreports/ recipients without an induction regimen 11 . According to one randomized pilot study, 2.25 mg/kg ATG induction with early steroid tapering was effective not only in preventing acute rejection but also in depleting T cells 3 . A recent study by Mehta et al. showed that patients in the regular dose ATG (≥ 5 mg/kg) group had more favorable outcomes in terms of early (within six months posttransplant) rejection compared with the low-dose (< 5 mg/kg) group 26 . On the other hand, another study by Linhares et al. reported that the 3 mg/kg and 5 mg/kg ATG induction groups did not show significant differences in the incidence of efficacy failure (first BPAR, graft failure, or mortality) and in safety outcomes at twelve months posttransplant 27 ; however, the study by Linhares et al. was limited due to its single-center, retrospective design, small number of patients, short follow-up duration, and the fact that only those who underwent repeat kidney transplantation were included.   www.nature.com/scientificreports/ Even though the incidence of BPAR was the highest in the high-dose ATG group in our study, recipients in this group showed superior outcomes in terms of graft and patient survival. Considering that the high-dose ATG group had the highest incidence of BPAR during the first 6 months, it is likely that the high-dose ATG group had more patients with higher immunologic risk although we tried to create high-quality propensity score weights with the TWANG package. Although it is known that ATG is effective to prevent and treat TCMR, there are several reports that ATG is also effective against ABMR by suppression of natural killer (NK) and follicular helper T cells. Several studies using microarray transcriptomic analysis showed that NK and follicular helper T cells are actively involved in the pathophysiology of ABMR and graft failure after kidney transplantation [28][29][30][31][32] . Recently, we reported the results of a prospective, randomized, pilot study showing that NK cells in recipients with ATG 6.0 mg/kg induction remained suppressed whereas NK cells in those with ATG 6.0 mg/kg induction had recovered to the pre-transplant levels 33 . Therefore, it is considered that those with high-dose ATG induction had superior outcomes in terms of graft failure and mortality in spite of higher incidences of ABMR.
Interestingly, not only the rate of developing malignancy but also the rates of graft failure and mortality were the highest in the low-dose ATG group. Higher incidences of cytomegalovirus infection, neutropenia, or thrombocytopenia in the high-dose ATG group did not have a significant impact on graft failure and mortality. The unique observation of mortality preceding death-censored graft failure in the low-dose ATG group highlights the need for further investigation to understand the underlying factors contributing to this phenomenon.
This study has several limitations. We could not assess the presence of donor-specific antibodies before transplantation, which is necessary for estimating the immunologic risk of each patient. In addition, data on calcineurin inhibitor trough levels were not collected during the follow-up period, and Banff scores for each BPAR were not recorded in the registry. However, we could investigate the incidences of TCMR and ABMR separately. Due to the limited follow-up duration in this nationwide registry, we could not evaluate the long-term clinical outcomes according to the induction regimen. Lastly, the protocols for desensitization and treatment of TCMR and ABMR could not be determined at each center, which could affect graft survival and patient survival after an episode of BPAR.
Nevertheless, it is noteworthy that this study investigated the contemporary practice of KT induction regimens in an Asian population and compared the clinical outcomes according to the induction regimen and ATG dosage using data from a nationwide kidney transplant cohort. Notably, low-dose ATG induction had the worst outcomes even than basiliximab induction in terms of graft survival and patient survival. www.nature.com/scientificreports/ In conclusion, our current study in an Asian nationwide cohort showed that high-dose ATG induction (> 4.5 mg/kg) had superior outcomes to basiliximab and low-dose ATG induction in terms of graft and patient survival.

Materials and methods
Study population. KOTRY is a nationwide transplantation cohort for five solid organs (kidney, liver, heart, lung, and pancreas) consisting of 59 participating centers in South Korea. Longitudinal data collection was initiated in 2014. Details about the design and methods were previously reported 16 . Briefly, KOTRY collects data on demographics, comorbidities, laboratory measurements, concentration and dosage of immunosuppressants including induction and maintenance regimen, allograft rejection, allograft failure, infectious events, cardiovascular complications, malignancies, recipient mortality, donor comorbidity, and outcomes of live donors. The KOTRY dataset includes a total of 5652 kidney transplant recipients from 30 centers between 2015 and 2019, from which 931 were excluded due to follow-up loss except for death and 142 were excluded because both ATG and basiliximab were administered as induction regimen (Fig. 6). Therefore, a total of 4579 recipients were included in this study. Among them, ATG was administered to 924 recipients whereas basiliximab was used for 3655 recipients. The mean, median, 25th percentile, and 75th percentile of the ATG dose were 4.85 mg/kg, 4.50 mg/kg, 4.26 mg/kg, and 5.45 mg/kg, respectively (Fig. 7). Of the 924 recipients with ATG induction, 467 were classified as low-dose ATG group and 457 as high-dose ATG group based on the median value of ATG dose (4.50 mg/kg). Written informed consent was achieved from all recipients and live kidney donors before KT and donor nephrectomy, respectively. This study was approved by the institutional review board of Asan Medical Center (approval number: 2021-0032) and was performed according to the Declaration of Helsinki and the Declaration of Istanbul. No organs/tissues were procured from prisoners. Clinical outcomes. The primary outcomes were the rates of BPAR, DCGF, overall graft failure, and mortality. BPAR was diagnosed according to the Banff 2015 criteria 17 . Secondary outcomes were the level of serum creatinine and the incidence of cytomegalovirus infection, bacterial infection, neutropenia or thrombocytopenia, or malignancy. CMV infection was defined as end-organ infection by CMV or persistent detection of CMV viremia (> 3.0 log copies/mL) in polymerase chain reaction tests. Neutropenia was defined as white blood cell count less than 1,500 /uL whereas thrombocytopenia was considered if the platelet count was less than 150,000 /uL. The primary and secondary outcomes were assessed at six months after transplantation and at one-year intervals thereafter. www.nature.com/scientificreports/  www.nature.com/scientificreports/ Statistical analysis. Continuous variables are presented as mean ± standard deviation (SD). One-way analysis of variance (ANOVA) was used to compare continuous variables among the three groups. Primary and secondary outcomes at each period were compared between groups using ANOVA. Categorical variables are summarized as absolute and relative frequencies, and differences among categorical values were analyzed using the Chi-squared test. In order to create high-quality propensity score weights, the Toolkit for Weighting and Analysis of Nonequivalent Groups (TWANG) package was applied to estimate the probability that a recipient developed primary and secondary outcomes for calculating weights and generalized boosted regression 18 . A total of fifteen recipient characteristics (age, sex, diabetes, previous transplant, ABO-incompatible KT, ABDR mismatch, DR mismatch, and calcineurin inhibitor) and donor characteristics (age, sex, body mass index, relation to the recipient, current smoker, serum creatinine, and cold ischemic time) were included in the weighting algorithm. After the matched samples were identified, generalized linear mixed models were estimated by including all recipient and donor characteristics to adjust for any possible differences among the groups. P values < 0.05 were considered statistically significant. Statistical analyses were performed using R software version 3.1.2 (R Foundation for Statistical Computing, Vienna, Austria).

Data availability
Raw data were generated at Korea Centers for Disease Control and Prevention Agency. Derived data supporting the findings of this study are available from the corresponding author on request.