We conducted a longitudinal retrospective single center study including all SPKT performed at Hospital Clínic Barcelona from January 1st, 2000 until December 31st, 2023 (n = 385). Patients with ≥1 previous transplant of any type (n = 19), pre-transplant calculated Panel Reactive Antibody (cPRA) >20% and/or Donor Specific Antibodies (DSAs) (n = 54), and those with a kidney or pancreas graft primary non-function for technical reasons (n = 7) were excluded. In total, 305 SPKT recipients were included.

According to our immunology laboratory, a bead in the Single Antigen assay was considered positive when the mean fluorescence intensity (MFI) was ≥1,000. However, this threshold was subject to minor patient-specific adjustments based on the background MFI observed in non–donor-specific beads, which could result in a slightly higher or lower effective cut-off.

Data was collected until 31st December 2024. The clinical and research activities being reported were consistent with the Declaration of Istanbul on Organ Trafficking and Transplant Tourism. The study protocol was approved by the Research Ethics Committee (HCB/2025/0613).

Induction immunosuppression therapy was used in all patients, with two doses of basiliximab of 20 mg at day 0 and at day +4 after surgery between January 2008 until July 2013. Before January 2008, and after July 2013, rabbit anti-human lymphocytes polyclonal antibodies (either Thymoglobulin^® 1.25 mg/kg/day or ATG^® 2.5 mg/kg/day, for 4 consecutive days) was administered as induction therapy. The first dose was administered intraoperatively, and the subsequent three doses on consecutive days following surgery. Dosage was adjusted according to leukocyte and platelet counts: it was reduced by 50% if the leukocyte count was <3.000/mL and/or the platelet count was <75,000/mL. If the leukocyte count fell below 1.500/mL and/or the platelet count below 50.000/mL, the dose was postponed (not discontinued) until the following day to try to reach a total cumulative dose of 5 mg/kg (10 mg/kg for ATG).

Maintenance immunosuppression protocol was based on triple therapy with calcineurin inhibitor (cyclosporine A until 2005, and thereafter tacrolimus), mycophenolate or mTOR inhibitors, and steroids (methylprednisolone in the immediate post-transplant period, followed by oral prednisone). The first dose of the calcineurin inhibitor was administered immediately before surgery. Administration was not postponed in cases of kidney Delayed Graft Function (DGF), and dosage adjustments were made solely based on trough levels. Therefore, all patients received the same regimen regardless of kidney DGF occurrence and induction therapy.

Anticoagulation included subcutaneous enoxaparin 20 mg bid starting 8 h post-surgery and was maintained until patient discharge (in the absence of thrombotic/hemorrhagic complications), and acetylsalicylic acid 50 mg/day starting at 12 h post-surgery until discharge, when it is increased up to 100 mg/day.

Infections were considered when requirement for hospital admission. Cytomegalovirus (CMV) prophylaxis with valganciclovir was administered to all patients for 1 month post-transplant or three if a donor/recipient mismatch for CMV was present. Infection was defined as any replication in CMV load post-transplant, regardless of the presence of CMV disease. BK nephropathy was defined an increase in BK viremia >10.000 UI/mL, regardless of the presence of biopsy proven BK nephropathy.

De novo neoplasms were considered as any neoplasia diagnosed during the post-transplant period, including solid tumours, post-transplant lymphoprolipherative disorders (PTLD) and excluding non-melanoma skin cancer.

Primary outcomes were recipient survival and death-censored kidney or pancreas graft survival at 1, 5, 10, and 20 years after transplantation, and graft rejection during follow up. Our secondary outcomes were defined as number of infections requiring hospital admission, CMV infection, BK virus nephropathy, and new onset neoplasms.

Patient survival was calculated from the date of transplantation to the date of death from any cause. Patients alive at the last follow-up were censored at that date. Pancreas graft failure was defined as any of the following: a) graft removal, b) C-peptide <1 ng/mL or c) total daily insulin dose >0.5 U/Kg.

Kidney graft failure was defined as return to dialysis or re-transplantation. Kidney DGF was defined as the need for at least one session of hemodialysis during the first week following SPKT.

Graft survival was analyzed using death-censored estimates to evaluate the effect of induction therapy on graft failure, particularly from immunological causes, independent of patient mortality. Nevertheless, to reduce the risk of bias derived from potential competing risks between recipient death and graft failure, a competing risk analysis was also performed.

All rejection episodes (for both pancreas and kidney) were biopsy-proven. Diagnostic criteria were based on the Banff classification in use at the time of diagnosis for either pancreas or kidney grafts. In cases of pancreas T cell–mediated rejection (TCMR), patients received three doses of methylprednisolone (500 mg/day) followed by five doses of thymoglobulin (1.25 mg/kg/day). For pancreas antibody-mediated rejection (ABMR), treatment consisted of three doses of methylprednisolone (250 mg/day), two doses of rituximab (400 mg/day), and five sessions of plasma exchange. For kidney graft rejection, the same therapeutic protocols were applied, except in cases of TCMR grade I, in which thymoglobulin was not administered.

Data are presented as mean (standard deviation, SD) for continuous variables and median [interquartile range (IQR)] for the non-continuous ones. The corresponding tests used were t-test, Mann-Whitney test, Chi-square or Fisher’s Exact test as appropriated. Competing risk analysis for graft survival was performed using the Fine–Gray subdistribution hazard model.

Inverse probability of treatment weighting (IPTW) was used to account for covariate imbalance between basiliximab and thymoglobulin groups. IPTW was estimated from a propensity score from a logistic regression model to receive basiliximab as the induction agent. The model included factors associated with the donor and either of the outcomes: dialysis duration before transplant, diabetes duration before transplant, time on the waiting list, HLA mismatches between donor and recipient, type of maintenance immunosuppression, prednisone withdrawal, recipient age at transplantation, cold ischemia time for kidney graft, Pancreas Donor Risk Index (PDRI), pancreas transplantation era, recipient smoking habit, cPRA before transplant.

A stabilized weighting method was performed by multiplying the IPTW by the proportion of recipients treated with basiliximab and thymoglobulin. Check for adequate balance of covariates after IPTW analyses was performed by calculation of standardized differences and an absolute difference greater than 0.2 represented a meaningful imbalance. All subsequent analyses were performed on the weighted, covariate-balanced population. Kaplan-Meier was used to estimate patient and graft survival and compared using a log-rank test. Logistic regression was used to calculate odds ratio for graft rejection, infections and neoplasms, and Cox proportional regression was performed to estimate patient and graft hazards.

All variables analyzed presented less than 10% of missing values. Given the low percentage, imputation methods were not applied, and analyses were conducted with the available data.

Statistical analysis was performed using IBM SPSS Statistics 30.0 (SPSS, Inc; Chicago, Illinois) software for MacOS and Python programming language (Python Software Foundation, 2024) in MacOS. All tests were two-tailed and a significance of 0.05 was used. Graphs were generated using the Python programming language in MacOS.

A total number of 305 SPKT recipients were included in the study (Table 1). In 172 (56%), thymoglobulin was used as the induction agent, while in 133 (44%) basiliximab was administered as the induction therapy. The mean follow up time for the whole cohort was 12.08 ± 6.84 years. Recipient age at SPKT was similar between both groups. Type 1 Diabetes (T1D) was predominant in both groups, in which diabetes duration was also similar. Most of the patients were on dialysis at SPKT in both groups, although time on dialysis before transplant was higher in the basiliximab one, as well as time on the waiting list. Smoking habit was more frequent in the basiliximab group, as well as patients at high risk of CMV infection.

Table 2 summarizes donor characteristics. Age at donation was similar between both groups. No differences were observed for PDRI score among the studied groups. Donors after Circulatory Death (DCD) were more frequent in the thymoglobulin group. Pancreas and kidney Cold Ischemia Time (CIT) were longer in the basiliximab one.

After IPTW adjustment, no significant differences were observed between both groups neither for recipient nor for donor characteristics. Supplementary Table S1 shows standardized differences for donor and recipient characteristics before and after IPTW adjustment.

Patient survival at 1, 5, 10, and 20 years after SPKT was 98.8%, 98.1%, 94% and 86.2% in the thymoglobulin group, respectively. For basiliximab group, survival was not significantly different, being 99.2%, 94.6%, 92.2% and 80.5% for the same time periods, respectively (Log Rank P = 0.31) (Figure 1A). Unadjusted Cox regression analysis showed that basiliximab was not associated with an increased risk of patient death compared to thymoglobulin [HR 1.47 (0.69–3.14), P = 0.32]. A similar scenario was observed after IPTW adjustment, with no difference for patient death comparing both groups [HR 1.01 (0.43–2.34) for basiliximab group, P = 0.99] (Table 3). The main cause of recipient death in both groups were infections (50% vs. 37% for thymoglobulin and basiliximab groups, respectively), followed by neoplasms (34% vs. 26% for thymoglobulin and basiliximab, respectively), with no differences between groups (P = 0.55) (Supplementary Table S2).

In the thymoglobulin group, pancreas death-censored graft survival at 1, 5, 10, and 20 years after SPKT was 95.9%, 93.1%, 88%, 86.1%, respectively. For basiliximab one, pancreas graft survival was similar, being 95.5%, 90.7%, 88.2%, 83.3% for the same time periods, respectively (Log Rank P = 0.83) (Figure 1B). No differences were observed for overall pancreas graft survival (Log Rank P = 0.57, Supplementary Figure S1A). Unadjusted Cox regression analysis showed no increased risk of pancreas failure with basiliximab compared to thymoglobulin as induction [HR 1.08 (0.55–2.10), P = 0.83]. These results were maintained after IPTW weighting [HR 1.57 (0.75–3.28) for basiliximab group, P = 0.24] (Table 3). A similar scenario was observed when performing a competing risk analysis, with a HR 0.93 [0.48–1.83], P = 0.84 for the basiliximab group.

Kidney graft survival rates in the thymoglobulin group at 1, 5, 10, and 20 years post-SPKT were 97.1%, 95.5%, 89%, and 84.7%, respectively. Similarly, kidney graft survival in the basiliximab group was 98.5%, 94.6%, 90.4%, and 89.6% at the corresponding time points (Log Rank P = 0.56) (Figure 1C). No differences were observed for overall kidney graft survival (Log Rank P = 0.99, Supplementary Figure S1B). According to unadjusted Cox regression analysis, there was no significant increase in the risk of kidney graft failure with basiliximab when compared to thymoglobulin [HR 0.80 (0.38–1.70), P = 0.56]. This finding remained consistent following IPTW adjustment [HR 1.49 (0.84–2.63) for the basiliximab group, P = 0.17] (Table 3). A similar scenario was observed when performing a competing risk analysis, with a HR 1.24 [0.57–2.70], P = 0.58 for the basiliximab group.

Throughout the entire follow-up period, 34 pancreas rejection episodes (20%) occurred in the thymoglobulin group and 32 episodes (24%) in the basiliximab group, with no statistically significant difference between them (P = 0.40) (Table 4). Rejection occurred after a median time of 4 [1–23] and 6 [1–13] months for the thymoglobulin and basiliximab groups, respectively (P = 0.69).

In both groups, the most frequent type of pancreas rejection was TCMR, with 29 (17%) and 32 (24%) cases for thymoglobulin and basiliximab, groups, respectively. There was no statistical support for an association between rejection type and treatment group (P = 0.06). However, a tendency toward a different rejection pattern was observed for the pancreas graft: in the basiliximab group, all cases were TCMR, whereas in the thymoglobulin group, 3% of cases were ABMR. Pancreas graft rejection was the cause of graft loss in 14 cases in the thymoglobulin group and 11 cases in the basiliximab group, representing 82% and 61% of all graft losses (41% and 34% of treatment failure), respectively (P = 0.16).

Kidney graft rejection rate was similar between thymoglobulin and basiliximab groups, observing 11 (7%) and 14 (11%) kidney graft rejection episodes (P = 0.19). Rejection occurred after a median time of 8 [1–53] and 8 [2–44] months for the thymoglobulin and basiliximab groups, respectively (P = 0.93).

The most frequent type in both groups was TCMR (4% and 9% in the thymoglobulin and basiliximab groups, respectively). No statistically significant association was found between rejection type and treatment group (P = 0.10) (Table 4). Kidney graft rejection was the cause of graft loss in 2 cases in the thymoglobulin group and 1 case in the basiliximab group, representing 13% and 8% of all graft losses (18% and 7% of treatment failure), respectively (P = 0.63).

When assessing the risk of graft rejection, basiliximab was not associated with a higher risk of rejection, either for pancreas [OR 1.28 (0.74–2.22), P = 0.37] or kidney graft [OR 1.72 (0.76–3.93), P = 0.20] compared to thymoglobulin. These results were consistent after IPTW weighting, with an OR of 1.49 [0.84–2.63] (P = 0.17) for pancreas and 1.31 [0.54–3.15] (P = 0.55) for kidney graft and basiliximab compared to thymoglobulin (Table 5).

The rate of post-transplant infections (except for CMV and BKV) that required patient admission was similar between thymoglobulin and basiliximab (35% vs. 38%, respectively. P = 0.55). Infections occurred after a median time of 35 [1–109] and 34 [22–80] days for thymoglobulin and basiliximab, respectively (P = 0.57) Nevertheless, when specifically considering CMV infection and BK nephropathy, thymoglobulin group exhibited a significantly higher rate of CMV infection (29% vs. 14% for thymoglobulin vs. basiliximab, respectively. P = 0.002) and BK nephropathy (12% vs. 4%, P = 0.01) (Table 6). CMV infection occurred after a median time of 3 [2–5] and 3 [1–5] months (P = 0.42), while BK infection occurred after a median time of 9 [6–30] and 28 [23–30] months for thymoglobulin and basiliximab, respectively (P = 0.40). Induction with basiliximab was significantly associated with a reduced risk of CMV infection [OR 0.39 (0.21–0.70), P = 0.002] and BK nephropathy (OR 0.28 [0.10–0.77], P = 0.01) compared to thymoglobulin (Table 7). This association was maintained after IPTW adjustment (OR 0.41 [0.23–0.72], P = 0.002 for basiliximab and CMV infection; OR 0.31 [0.12–0.80], P = 0.02 for basiliximab and BK infection).

The incidence of new onset neoplasms was similar between both groups (6% vs. 10% for thymoglobulin and basiliximab groups, respectively, P = 0.21). Median time to neoplasm diagnosis was 73 [39–90] and 170 [95–201] months for thymoglobulin and basiliximab groups, respectively (P = 0.08). In this case, no significant association was identified between basiliximab and neoplasm development compared to thymoglobulin, either in the unadjusted [OR 1.72 (0.76–3.93), P = 0.20] or IPTW-weighted analysis [OR 1.24 (0.48–3.21), P = 0.66] (Tables 6, 7).