All patients who underwent pancreas transplantation (simultaneous pancreas-kidney (SPK), pancreas after kidney (PAK), and pancreas transplant alone (PTA) between 1st January 2010, and 30th April 2022, at our institution were included in the study. Data were extracted from the French prospective DIVAT cohort of transplanted patients. ¹

Complete available data are presented in Table 1. Donor and recipient characteristics, as well as peri-transplant parameters, were prospectively collected. Pancreas failure was defined by either the persistence of insulin requirement, allograft removal, or retransplantation. Kidney failure was defined by either a return to dialysis or retransplantation. Rejection episodes were diagnosed based on pancreatic biopsy findings or if no biopsy was available, pancreas rejection was considered in cases of dysfunction (hyperglycemia + increase in lipase levels) with a biopsy-proven diagnosis of kidney rejection [19]. This strategy aimed to minimize unnecessary invasive biopsies, especially for the pancreatic allograft. Rejection episodes were categorized according to the Banff classification. Cellular rejection was usually treated with steroid pulses or r-ATG (Thymoglobulin), while humoral rejection was managed with plasma exchanges, intravenous immunoglobulins, and sometimes associated with CD20 depleting therapy. Donor specific antibodies (DSA), assessed pre-transplant, in case of rejection, and at 1 year post-transplantation were determined by Luminex^® assay and considered positive when mean fluorescence index values were superior to 1000. Infectious complications, including CMV viremia (either asymptomatic or associated with CMV disease), BK virus (BKV) viremia (either asymptomatic or associated with BKV nephropathy), fungal infections, and severe bacterial infections, were recorded. Prospective follow-up of pancreatic and kidney allograft functions included fasting glycemia, fasting C-peptide, HbA1c levels, estimated glomerular filtration rate (eGFR, using the CKD-EPI formula), collected every 3 months during the first year and then annually. Follow-up and data collection ceased upon transplant failure or death.

The management of pancreas transplantation was consistent across all categories (SPK, PAK, and PTA) and remained globally unchanged during the study period, except for the addition of Etanercept. The surgical technique remains globally unchanged during the study period, with digestive anastomosis performed in all cases for exocrine diversion. Induction therapy consisted mostly in rabbit antithymocyte globulin (rATG) for five alternate days, or either basiliximab in some rare cases, along with two pulses of 500 mg methylprednisolone. From April 2017, pancreas transplant recipients received an additional course of Etanercept at a similar dosage than for islet recipients: 50 mg on day 0 (intravenous), followed by 25 mg (subcutaneous) on days 3, 7, and 10. All patients underwent screening for latent tuberculosis and hepatitis viruses before Etanercept administration. Maintenance immunosuppressive therapy included a calcineurin inhibitor (mainly tacrolimus) and mycophenolate mofetil or mycophenolic acid, with oral prednisone tapered and withdrawn from postoperative day 7. Our anticoagulation protocol involved per-operative administration of intravenous aspirin (250 mg) and heparin (25 UI/kg) at the time of clamping, followed by preventive anticoagulation using low molecular weight heparin within the first days post-surgery, typically for 10 days. In the absence of allograft thrombosis, detected on purpose or by systematic CT-scan on day 10, preventive heparin was replaced by long-term administration of antiplatelet therapy. Finally, our strategy for treating pancreatic rejection episodes remained largely consistent throughout the study period (i.e., steroid pulses for cellular rejection, with rATG used in cases of grade II or grade III cellular rejection or steroid resistance, and plasma exchange, IV Ig and Rituximab for treatment of humoral rejection).

The characteristics at transplantation were described using frequency and proportion for categorical variables and mean and standard deviation for continuous variables. To assess the impact of anti-TNFα treatment on a specific phenotype over time, survival curves were generated using the Kaplan-Meier estimator. Statistical comparisons were conducted using the log-rank test. For univariate analysis, the Student’s t-test or Mann-Whitney test was employed, while multivariate analysis used the Cox model. The anti-TNFα variable was consistently included in the statistical models to evaluate its effect on the different studied outcomes. Initial variable selection was performed retaining only those with a p-value of less than 0.2 according to the Wald test for inclusion in the final Cox model [20]. In addition, five variables were forced selectively into the Cox model for pancreas survival due to their known association with complete thrombosis (pancreas cold ischemia time, and donor-related variables: age, BMI, vascular cause of death, and history of hypertension). Similarly, induction therapy (r-ATG or Basiliximab) was forced into the Cox model for pancreatic rejection. Subsequently, a stepwise forward selection process was conducted, whereby variables were added to the model if their inclusion improved the Bayesian information criterion. The final model comprised the forced variables along with any additional selected variables. Of note, patients with missing data on the variables of interest were excluded from the final analysis. The hazard proportionality assumption was tested from the Schoenfeld residuals [21]. The absence of multicollinearity of the model was verified using the Variance Inflation Factor. To visualize the results, adjusted survival curves were generated to observe the impact of anti-TNFα use over time while holding other variables constant. While one-year endpoints were assessed to accurately determine the impact of anti-TNFα, we also conducted a three-year analysis to gain insights into its long-term effects. Even if some confounding factors may arise well after the induction treatment; these are part of the causal pathway of the initial treatment (i.e., they result from it) and should be considered as part of the evaluation process.

The analysis was conducted using R version 4.1.3, with statistical significance defined as a p-value of less than 0.05.

All data were extracted from the Nantes DIVAT database. This study received data privacy approval from CNIL (09-17-2004, number n°891735, Réseau DIVAT:10.16.618). The patient’s non-opposition regarding access to their medical records, collection and data processing is mandatory under French law. All data were anonymized before analysis. The use of Etanercept in pancreas transplant recipients was approved by the local human ethics committee (n°23-115-09-211). The quality of the DIVAT data bank is validated by an annual audit. 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.

During the study period, 236 pancreas transplant recipients were included, among whom 87 received anti-TNFα and 149 received standard of care (SOC). The complete characteristics of the population are described in Table 1. Briefly, 77.1% received simultaneous pancreas-kidney (SPK) transplants, 13.6% received pancreas transplant alone (PTA), and 9.3% received pancreas after kidney (PAK) transplants, with a mean age of 40 years. The mean donor’s age was 33 years, with a mean BMI of 23, and 39% of them died from cardiovascular events, without any significant differences observed among groups. Of note, patients receiving anti-TNFα were more often transplanted with shorter pancreatic and kidney cold ischemia times (563 vs. 635 min, p < 0.0001 and 688 vs. 794 min, p < 0.0001 respectively). 10.6% of patients presented with preformed donor-specific antibodies (DSA) at the time of transplantation. Induction therapy consisted of a T-cell depleting agent in 92.4% of the cohort, followed by maintenance therapy comprising a calcineurin inhibitor (mainly tacrolimus: 99.1%) and an antiproliferative agent, either mycophenolate mofetil or mycophenolic acid (99.6%). Oral steroids were administered to 97.9% of patients, followed by rapid tapering during the first weeks post-transplantation.

At 3 years post-transplantation, the overall pancreatic allograft survival rate was 80.1%. The main causes of failure were allograft thrombosis (68.1%), allograft rejection (17.0%), and surgical complications (10.6%). Numerically, there were 15 allograft failures in the anti-TNFα group (17.2%, of whom 13/15 were complete thrombosis) and 32 in the SOC group (21.5%, of whom 19/32 were complete thrombosis) at 3 years. After adjusting for confounding variables and factors associated with allograft failure due to thrombosis, the adjusted hazard ratio (HR) for pancreas survival was 0.92 (95% CI = 0.49; 1.73, p = 0.79) for patients receiving anti-TNFα therapy, Table 2. The cumulative adjusted probability of pancreatic allograft survival is presented in Figure 1A. Among SPK recipients, the adjusted HR for kidney allograft survival was 0.50 (95% CI = 0.10; 2.49, p = 0.40) for patients receiving anti-TNFα therapy compared to the SOC group, Supplementary Table S1. The cumulative adjusted probability of kidney allograft survival is presented in Figure 1B.

We further investigated pancreatic and kidney allograft function censored for allograft loss (Supplementary Figure S1). Regarding the pancreas, no differences were found in fasting glycemia, fasting C-peptide levels, and HbA1c levels during the first 3 years post-transplantation in the anti-TNFα group vs. SOC. Similarly, in the subgroup of SPK recipients, eGFR were globally comparable even if we observed a higher eGFR slope between 3 months and 3 years among patients from the SOC group vs. anti-TNFα (respectively −12.1% and −2.3%).

At 3 year post-transplantation, there were 5 pancreatic rejection episodes (5.7%) diagnosed in the anti-TNFα group (3 proven by pancreatic biopsy) and 26 (17.4%) in the SOC group (17 proven by pancreatic biopsy). The complete description of these rejection episodes is provided in Table 3. The occurrence of a pancreatic rejection episode led to further allograft loss in around 60% of cases. In the multivariate analysis, after adjusting for confounding factors—particularly induction therapy—adjunctive treatment with anti-TNFα was significantly protective against the occurrence of pancreatic rejection during the first year post-transplantation (HR = 0.23, 95% CI = 0.07–0.76, p = 0.01; Table 4; Figure 2A). Importantly, this protective effect persisted over time and remained significant up to 3 years post-transplantation (HR = 0.36, 95% CI = 0.14–0.95, p = 0.04; Table 5; Figure 2B). Notably, among the 18 patients who received non-depleting induction therapy and no anti-TNFα, the incidence of pancreatic rejection at 3 years was 11.1%, which aligns with the rejection incidence in patients who received a T-cell depleting induction without anti-TNFα. This may be linked to a higher level of maintenance immunosuppressive burden administered during the first year in these patients (Supplementary Figure S2). Finally, occurrence of DSA at 1 year was comparable between groups (16.4% vs. 10.4%, p = 0.55). The protective effect of anti-TNFα on pancreatic rejection was particularly notable as maintenance therapy was significantly reduced in the anti-TNFα group compared to the SOC group, especially regarding tacrolimus trough levels and steroid use during the first months, Figure 3.

Conversely, anti-TNFα did not significantly impact the risk of kidney rejection (HR = 0.72, 95% CI = 0.31; 1.66, p = 0.44), as shown in Figures 2C, D and Supplementary Tables S2, S3 . Nevertheless, we observed a shift in the kidney Banff classification, with a trend toward fewer TCMR and ABMR and more Borderline lesions among SPK patients treated with anti TNFα, Supplementary Figure S3 .

During the first year post-transplantation, we did not observe an increased risk of infectious complications following the administration of anti-TNFα. Regarding the occurrence of severe bacterial infections, the adjusted HR was 0.69, 95% CI = 0.50; 0.95, p = 0.02 for patients receiving anti-TNFα, as shown in Figure 4A, and Supplementary Tables S4, S5 . Concerning the occurrence of fungal infections, the adjusted HR was 0.53, 95% CI = 0.26; 1.07, p = 0.08 for patients receiving anti-TNFα, as depicted in Figure 4B and Supplementary Tables S6, S7 . The risk of CMV viremia was similar among patients receiving anti-TNFα compared to others (adjusted HR = 0.89, 95% CI = 0.37; 1.24, p = 0.21), Figure 4C and Supplementary Tables S8, S9 . Finally, the risk of BKV viremia was also similar following the administration of anti-TNFα (HR = 0.58, 95% CI = 0.31; 1.07, p = 0.08), Figure 4D, Supplementary Tables S10, S11. No cases of tuberculosis or viral hepatitis replication were observed among patients having received anti-TNFα therapy. Finally, anti-TNFα therapy did not impact patient survival (Supplementary Figure S4).