Donor genetic variants in interleukin-6 and interleukin-6 receptor associate with biopsy-proven rejection following kidney transplantation

Rejection after kidney transplantation remains an important cause of allograft failure that markedly impacts morbidity. Cytokines are a major player in rejection, and we, therefore, explored the impact of interleukin-6 (IL6) and IL-6 receptor (IL6R) gene polymorphisms on the occurrence of rejection after renal transplantation. We performed an observational cohort study analyzing both donor and recipient DNA in 1271 renal transplant‐pairs from the University Medical Center Groningen in The Netherlands and associated single nucleotide polymorphisms (SNPs) with biopsy-proven rejection after kidney transplantation. The C-allele of the IL6R SNP (Asp358Ala; rs2228145 A > C, formerly rs8192284) in donor kidneys conferred a reduced risk of rejection following renal transplantation (HR 0.78 per C-allele; 95%-CI 0.67–0.90; P = 0.001). On the other hand, the C-allele of the IL6 SNP (at position-174 in the promoter; rs1800795 G > C) in donor kidneys was associated with an increased risk of rejection for male organ donors (HR per C-allele 1.31; 95%-CI 1.08–1.58; P = 0.0006), but not female organ donors (P = 0.33). In contrast, neither the IL6 nor IL6R SNP in the recipient showed an association with renal transplant rejection. In conclusion, donor IL6 and IL6R genotypes but not recipient genotypes represent an independent prognostic marker for biopsy-proven renal allograft rejection.


Results
Determinants of biopsy-proven rejection following kidney transplantation. A total of 1271 kidney transplant donor-recipient pairs were included in this study with the donor and recipient characteristics listed in Table 1. The mean follow-up after transplantation was 5.2 years ± 5.0 with a maximum follow-up period of 16.7 years. During follow-up, 33.8% of the recipients developed biopsy-proven rejection. In the first year, 390 recipients developed acute biopsy-proven rejection, whereas 40 recipients presented with biopsy-proven rejection thereafter. Of all the assessed characteristics, the following were significantly associated with biopsy-proven rejection (Table 1); recipient age, recipient sex, warm ischemia time, the total number of HLA mismatches, and the occurrence of delayed graft function (DGF).

A common genetic variant in interleukin-6 receptor protects against renal allograft rejection.
To identify relevant genetic biomarkers in the context of rejection, we explored whether the Asp358Ala variant in IL6R (rs2228145 A > C, previously rs8192284) associated with biopsy-proven rejection. The observed genotype frequencies in recipients (n = 1270; AA, 34.3%; AC, 51.4%; CC, 14.2%) and donors (n = 1269; AA, 35.6%; AC, 50.6%; CC, 13.6%) were comparable (P = 0.75). Yet, the frequencies of the IL6R polymorphism in both recipients and donors were significantly higher than reported by the 1000 genomes project (P < 0.0001), but  15 . Kaplan-Meier survival analyses showed that the IL6R polymorphism in the donor was significantly associated with a reduced risk for acute rejection during follow-up ( Fig. 2A; log-rank P = 0.003). After the complete follow-up, the incidence of biopsy-proven rejection was 38.9% in the reference AA genotype group, 32.8% in the heterozygous AC genotype group, and 24.9% in the homozygous CC genotype group, respectively. In univariate analysis, the IL6R SNP in the donor was associated with a hazard ratio of 0.78 per C-allele (95%-CI 0.67-0.90; P = 0.001) for biopsy-proven rejection. Kaplan-Meier survival analyses for the IL6R polymorphism in the donor were re-estimated for patients transplanted in the 1990s and 2000s as immunosuppression has evolved through time and this could impact the risk of rejection. In these subgroups, the significance was maintained in patients transplanted in the 1990s ( Figure S1; P = 0.016), while a Figure 1. Illustration of the IL-6 signaling pathways and examined IL-6-related SNPs. IL-6 signalling can occur in three distinct manners. (For schematic purposes, only one of each IL-6 receptor and gp130 molecule is shown.) (A) Classic IL-6 signalling takes place when IL-6 directly interacts with membrane-bound IL-6R (mIL-6R) and membrane-bound gp130 (mgp130) to phosphorylate intracytoplasmic, gp130-specific secondary messengers. (B) IL-6 trans signalling instead occurs when solubilized IL-6R (sIL-6R) captures IL-6 and forms a complex with mgp130 to precipitate intracytoplasmic secondary signals. Notably, this mode of signalling can be inhibited by solubilized gp130 (sgp130), which only has affinity for the IL-6/sIL-6R complex and neutralizes it by preventing its binding to mgp130. (C) Finally, IL-6 trans presentation takes place when a secondary cell bearing mIL-6R captures IL-6 and presents it to mgp130 on a target cell to spur intracytoplasmic secondary signals in the target cell. To appreciate the potential role of IL-6-related SNPs in kidney transplant recipients and donor transplant kidneys, we assessed the association between transplant survival outcomes and (D) the IL6R SNP rs2228145 A > C and (E) the IL6 SNP rs1800795 G > C, respectively. The IL6R SNP causes a missense mutation in IL-6R, while the IL6 SNP is in the promoter region and causes an intronic variant.  Figure S1; P = 0.067). Nevertheless, in accordance with our previous results, the C-allele of the IL6R SNP in the donor remained protective against biopsy-proven rejection.
In the recipients, Kaplan-Meier survival analyses revealed no associations between the IL6R SNP and rejection during follow-up ( Fig. 2B; log-rank P = 0.15). Subgroup analysis for recipient sex or donor type did not change these results. To assess whether donor-recipient mismatch for the IL6R SNP increases the risk of biopsy-proven rejection, kidney transplant pairs were divided into four groups according to the presence or absence of the C-allele in the donor and recipient. Although Kaplan-Meier survival analyses revealed a significant difference in graft survival among the four groups, the C-allele of the IL6R SNP in the donor remained associated with a lower risk for BPAR ( Figure S2, P = 0.004). However, the effect of the C-allele in the donor was greater in recipients who did not themselves possess this allele. A multivariable analysis was performed to adjust for potential confounders (Table 2), including donor characteristics (model 2), recipient characteristics (model 3), and transplant variables (model 4).
In Cox regression analysis, the IL6R SNP in the donor remained significantly associated with biopsy-proven rejection independent of potential confounders. Finally, we performed a multivariable analysis with a stepwise forward selection procedure using all variables associated with rejection in univariable analysis (Table 3). In the final model, the IL6R SNP in the donor, recipient age and sex, the total number of HLA mismatches, and DGF were included. After adjustment, the IL6R SNP in the donor was associated with rejection with a hazard ratio of 0.73 per C-allele (95%-CI 0.62-0.86, P < 0.001). Hence, altogether these results demonstrate that a common functional variant in the IL6R gene in the donor associates with a lower incidence of biopsy-proven acute rejection after kidney transplantation.
A common genetic variant in interleukin-6 is a risk factor for renal allograft rejection. Next, we explored whether the IL6 SNP at position-174 in the promoter (rs1800795 G > C) affects the incidence of biopsy- Log-rank test was used to compare the incidence of biopsyproven rejection between the groups. Table 2. Associations of interleukin-6 and interleukin-6 receptor polymorphisms in the donor with rejection following kidney transplantation. Data are presented as hazard ratio with 95% confidence interval (CI) and P-value. Model 1 crude model. Model 2 adjusted for model 1 plus donor characteristic's: donor age, donor sex, donor blood type, and donor origin. Model 3 adjusted for model 1 plus recipient characteristic's: recipient age, recipient sex, recipient blood type and dialysis vintage. Model 4 adjusted for model 1 plus transplant characteristic's: cold and warm ischemia time, the total HLA-mismatches, and the occurrence of delayed graft function (DGF). a Previously rs8192284. b For the IL6 SNP (rs1800795-C), model 2 could not be adjusted for donor sex since this was a subgroup analysis for male donors. www.nature.com/scientificreports/ Table 3. Competitive analysis of the associations of clinical factors with rejection following kidney transplantation. Multivariable cox regression was performed with a stepwise forward selection. Only variables that with a P-value below 0.10 in the univariate analysis were included. Data are presented as hazard ratio with 95% confidence interval (CI) and P-value. In the final, the IL6R SNP (rs2228145-C, previously rs8192284) In the donor, total number of HLA-mismatches, recipient age, the occurrence of delayed graft function (DGF) and recipient sex (female compared to male) were included, whereas dialysis vintage and warm ischemia time were not.  In additional analysis, the IL6 SNP in the recipient was not tied to the risk of biopsy-proven rejection within the first year either (P = 0.19). Furthermore, we assessed whether donor-recipient mismatch for the IL6 polymorphism impacted the risk of BPAR. Kidney transplant pairs were divided into four groups according to the presence or absence of the C-allele of IL6 SNP in the donor and the recipient. Kaplan-Meier survival analyses revealed no significant difference in rejection-free survival of kidney allografts among the four groups ( Figure S2; P = 0.24). We next performed a subgroup analysis for sex, since sex is known to impact immunity, renal disease as well as transplantation outcome 16,17 . Kaplan-Meier curves revealed a significant association between the IL6 SNP and rejection in male donors (Fig. 3C, P = 0.015). After the complete follow-up, the incidence of biopsy-proven rejection was 27.1% in the reference GG genotype group, 37.3% in the heterozygous GC genotype group, and 40.7% in the homozygous CC genotype group, respectively. No association was seen between the IL6 SNP and rejection in female donors (Fig. 3D, P = 0.33). Univariate analysis revealed that the IL6 SNP in male donors was associated with a hazard ratio of 1.31 per C-allele (95%-CI 1.08-1.58; P = 0.006) for biopsy-proven rejection. Multivariable analysis was performed to adjust for potential confounders (Table 2), including donor characteristics (model 2), recipient characteristics In Cox regression analysis, the IL6 SNP in male donors remained significantly associated with biopsy-proven rejection independent of potential confounders. Thus, it appears that a functional variant in IL6 in male donors is linked to a higher incidence of biopsy-proven rejection after kidney transplantation.

Hazard ratio (per allele) 95% CI P-value Hazard ratio (per allele) 95% CI P-value
Prediction of biopsy-proven rejection. The combined presence of the IL6 and IL6R SNP was common ( Fig. 4A,B), as well as the presence of the same SNP in the donor and recipient of a transplant pair (Fig. 4C,D). We, therefore, speculated that assessing the combined presence of multiple polymorphisms, in a genetic risk score, in the donor-recipient pairs could yield more information than examining the polymorphisms individually. To explore the combination of IL6 and IL6R SNPs as predictors of biopsy-proven rejection, we created a genetic risk score of the two variants in both the donors and recipients. Weight was added to each SNP according to their hazard ratio, creating a negative score for protective polymorphisms and a positive score for hazardous ones. Overall, a genetic risk score above zero indicates the presence of more hazardous SNPs in a donor-recipient Table 4. Associations of interleukin-6/interleukin-6 receptor genetic risk score with rejection following kidney transplantation. Data are presented as hazard ratio with 95% confidence interval (CI) and P-value. Model 1 crude model. Model 2 adjusted for model 1 plus donor characteristic's: donor age, donor sex, donor blood type, and donor origin. Model 3 adjusted for model 1 plus recipient characteristic's: recipient age, recipient sex, recipient blood type and dialysis vintage. Model 4 adjusted for model 1 plus transplant characteristic's: cold and warm ischemia time, the total HLA-mismatches, and the occurrence of delayed graft function (DGF).  Figure 5. Hazard ratios for the IL-6/IL-6R genetic risk score among subgroups. Forest plot of sub-analyses of the IL-6/IL-6R genetic risk score, demonstrating that the hazard ratios for biopsy-proven rejection were consistent in different subgroups. The only exception was the donor origins of the kidney allografts. The association between the IL-6/IL-6R genetic risk score and rejection was not seen in kidney transplants from living donors. No significant interaction was seen between the IL-6/IL-6R genetic risk score and the different clinical variables of the subgroups. To assess the clinical applications of the IL-6/IL-6R genetic risk score, we studied the predictive value of this genetic profiling in more detail. The genetic risk score was significantly associated with biopsy-proven rejection in both the crude model (HR, 1.24; 95%-CI 1.12-1.36; P < 0.001 per SD increase) and after adjustment for donor characteristics, recipient characteristics, and transplant variables (Table 4). Furthermore, the hazard ratio of the IL-6/IL-6R genetic risk score was consistent in subgroups analyses and remained significant (Fig. 5), except in living donors after stratification for donor type (P = 0.22).
The confidence intervals of all subgroups showed substantial overlap with the overall hazard ratio at the top, demonstrating the consistency of the findings across subgroups. Next, we tested if the IL-6/IL-6R genetic risk score was a better predictor for rejection than the IL6R SNP in the donor by multivariable regression with a stepwise forward selection was performed (Table 5). In the final model, the IL-6/IL-6R genetic risk score was included whereas the IL6R SNP in the donor was not. After adjustment, the genetic risk score in the donor was associated with biopsy-proven rejection with a hazard ratio of 1.27 per SD increase (Fig. 6, 95%-CI 1.14-1.42; P < 0.001).
The performance of the IL-6/IL-6R genetic risk score for the prediction of rejection was also assessed ( Table 6). The genetic risk score alone had a Harrell's C of 0.57 (95%-CI 0.54-0.60). Moreover, when added to a model of the IL6R SNP in the donor (c-statistic, 0.51; 95%-CI 0.488-0.539), the IL-6/IL-6R genetic risk score significantly increased the Harrell's C values (c-statistic increase, 0.054; 95%-CI 0.023-0.086; P = 0.001). Next, additional variables were included and the discriminative accuracy to predict graft loss of the model improved. The Harrell's C of the models with the donor characteristics and the transplant variables significantly improved with the addition of the genetic risk score, while only a trend was seen in the model with recipient characteristics. In addition, the IL-6/IL-6R genetic risk score significantly improved the predictive value of the models according to the integrated discrimination improvement index (IDI). Even in the fully adjusted models, the IDI value Table 5. Competitive analysis of the associations of clinical factors with rejection. Multivariable cox regression was performed with a stepwise forward selection. Only variables that with a P-value below 0.10 in the univariate analysis were included. Data are presented as hazard ratio with 95% confidence interval (CI) and P-value. The IL-6/IL-6R genetic risk score, total number of HLA-mismatches, recipient age, occurrence of delayed graft function (DGF), recipient sex and warm ischemia time were included in the final model. a Previously rs8192284.

Discussion
Rejection after kidney transplantation remains an important cause of allograft failure that greatly impacts morbidity [18][19][20] . Use of novel therapies to reduce allosensitization is vital, but new drug development for kidney transplantation is limited 21 , necessitating the repurposing of existing anti-inflammatory drugs approved for other indications. Among these are IL-6-blocking therapies, which have already been approved for the treatment of auto-immune diseases 22 . Human genetics offer an opportunity for target validation. Moreover, drug targets informed by human genetic evidence are more than twice as likely to lead to approved therapeutics 23,24 . The main finding of our study is a significant association between a common functional IL6 and IL6R polymorphisms in the donor and the risk of biopsy-proven rejection following kidney transplantation. Extending these findings, a genetic risk score based on both SNPs in the donor and recipient was shown to be a major determinant of biopsy-proven rejection. Moreover, this IL-6/IL-6R genetic risk score significantly improved risk prediction for rejection beyond currently used clinical risk factors. In conclusion, our study provides genetic evidence for the potential efficacy of targeting the IL-6 pathway in renal transplantation and encourages the study of IL-6 receptor inhibitors in kidney donors in randomized controlled trials. To our knowledge, our study is the first to demonstrate that IL6R polymorphisms in the donor impact the risk of rejection after kidney transplantation. To sum up, we found that for each copy of the C-allele in the donor the relative risk of biopsy-proven acute rejection decreased by 27% (90%-CI 14-38%). In accordance with our results, the same allele (rs2228145-C) has been associated with decreased risks of rheumatoid arthritis and coronary heart disease 25,26 . A recent study by Bovijn and colleagues found that this IL6R variant was also associated with a lower risk of SARS-CoV-2 infection, as well as a lower risk of hospitalization for COVID-19 27 . Studies on the functional consequences of this variant have helped to understand the molecular mechanisms through which this IL6R genotype protects against a wide spectrum of diseases [28][29][30] . This non-synonymous polymorphism accounts for over 50% of the total variance in sIL-6R levels and each copy of the C-allele increases plasma levels of sIL-6R. Although these effects of this IL6R variant on sIL-6R levels may appear contradictory, further investigation revealed that the C-allele simultaneously reduces membrane-bound IL-6R on monocytes and CD4 + T cells (up to 28% reduction per C-allele) 30 . Importantly, reduced surface expression of IL-6R on leukocytes resulted in diminished IL-6 receptiveness, as observed by a reduction in phosphorylation of the key transcription factors STAT1 and STAT3 following stimulation with IL-6 30 . Additional in vivo evidence of the anti-inflammatory effects of this IL6R variant has been demonstrated by the association with lower levels of C-reactive protein, fibrinogen, IL-8, and TNF-α in various studies [26][27][28]31 . Our analysis convincingly shows that the IL6R variant is associated with a lower risk of renal allograft rejection. We postulate that this decrease in rejection rate is due to the amplification of sIL-6R combined with circulating, endogenous, soluble gp130 acting as a buffer to neutralize IL-6 ( Fig. 1B), thereby suppressing inflammation as well as decreasing allosensitization, invariably diminishing the risk of allograft rejection.
The impact of the IL6 ~ 174G/C polymorphism (rs1800795) on the risk of rejection among transplantation patients has previously been investigated by several studies 13,[32][33][34][35][36] . A recent metanalysis of seven studies including 369 cases and 679 controls concluded that the recipient IL6 genotype was not significantly associated with rejection, whereas a trend was seen for donor IL6 genotypes 14 . In our transplant cohort of 1271 patients we did not find such an association with recipient IL6 genotype either. Yet, when we performed a subgroup analysis for the sex of the donor, we observed a significant association between the IL6 SNP and rejection in male donors. This sex-related difference might also explain the conflicting results by previous studies. Similar to our observations, others have reported a clear sex dimorphism in the association of this IL6 polymorphism with the vulnerability to illnesses [37][38][39] . Overall, the associations with this IL6 variant were stronger and predominantly found in men. Our study, therefore, highlights that sex should be taken into account in transplantation Table 6. Additive value of the IL-6/IL-6R genetic risk score in with rejection. Data are presented as Harrell's concordance statistic (Harrell's C) with 95% confidence interval (CI) and integrated discrimination improvement (IDI) with P-value. a Change in C-statistics compared to the model without the IL-6/IL-6R genetic risk score. Model 1 crude model. Model 2 rs2228145-C (previously rs8192284) in the donor. Model 3 adjusted for model 1 plus donor characteristic's: donor age, donor sex, donor blood type, and donor origin. Model 4 adjusted for model 1 plus recipient characteristic's: recipient age, recipient sex, recipient blood type and dialysis vintage. Model 5 adjusted for model 1 plus transplant characteristic's: cold and warm ischemia time, the total HLA-mismatches, and the occurrence of delayed graft function (DGF).

Change (95% CI) a P-value IDI (%) P-value
Without the IL-6/IL-6R genetic risk score with the IL-6/IL-6R genetic risk score www.nature.com/scientificreports/ as well as for cytokine-targeted therapies. Furthermore, in conformity with our results, the C-allele of this IL6 SNP has also been associated with increased risks of rheumatoid arthritis and cardiovascular disease in recent meta-analyses [40][41][42] . For rs1800795, the G-allele was initially said to increase IL-6 levels, however, recent work revealed that the C-allele of this SNP leads to higher IL-6 expression in fibroblasts but not in leukocytes 43 . Brull et al. found different kinetic profiles for IL-6 increase after surgery based on IL6 genotypes, which could explain previous conflicting results 44 . Nevertheless, the overall increase in IL-6 was more profound in CC homozygotes. Further in vivo evidence of the pro-inflammatory effects of the C-allele is demonstrated by the association with higher serum levels of IL-6, C-reactive protein, and fibrinogen in multiple studies [45][46][47] . Altogether, our study adds to a growing body of evidence that connects local production of IL-6 to the allosensitization of recipients to renal allografts 48,49 . Tocilizumab, a humanized monoclonal antibody targeting the IL-6R, has been assessed for the treatment of acute rejection, chronic ABMR, and transplant glomerulopathy following renal transplantation [50][51][52] . Initially, a phase I/II trial was performed in 10 patients prior to kidney transplantation that were unresponsive to desensitization with intravenous Ig and rituximab 50 . Tocilizumab, combined with intravenous Ig, led to a significant reduction in DSA levels and appeared safe. Five patients were transplanted and six-month protocol biopsies showed no acute rejection or transplant glomerulopathy. Next, an open-label single-center trial was undertaken in 36 patients with chronic ABMR that were non-responsive to intravenous Ig and rituximab 51 . Patient and renal allograft survival was 91% and 80% at 6 years, respectively, and this was found to be superior to historical controls. Furthermore, stabilization of renal function was seen after 2 years. Finally, tocilizumab was investigated for the treatment of acute rejection in an observational study of 7 kidney transplant recipients 52 . Renal function stabilized or improved in all patients, but one patient had a potential hypersensitivity reaction, and another patient developed cytomegalovirus esophagitis. A multicenter randomized control trial is currently underway 53 . Considering IL-6 inhibitors are already being tested in clinical trials, what, then, is the role for genetic studies here? The results from our study provide important considerations for the design of these clinical trials by identifying the following key issues for targeting IL-6 in renal transplantation: (1) Site of action; (2) Sex differences; and, (3) Patient selection. We found that genetic polymorphisms of the IL-6 signaling pathway in donors, but not recipients, were associated with the risk of biopsy-proven rejection after kidney transplantation. These results indicate that not circulating IL-6 in the recipient but local IL-6R expression and IL-6 production by the donor kidney are key drivers of alloimmunity to the kidney transplant. This potentially suggests that therapies aimed at blocking IL-6/IL-6R interactions should focus on the donor kidney as the site of action. Finally, in this study, we constructed an IL-6/IL-6R genetic risk score based on two SNPs in the donor and recipient. From a prognostic perspective, a genetic risk score could be of interest to identify patients at risk of allograft rejection. Furthermore, a genetic risk score could be of interest to identify renal transplant pairs that would benefit from anti-IL-6 treatment. However, considering the discriminative performance of our prediction model, additional factors (such as IL6 and IL6R haplotypes) should be added to increase the predictive performance before it could be used in clinical practice.
Several limitations of our study warrant consideration. First, the associations found in this study are expected to be causal. However, since our study is prospective, but observational in nature, it cannot be proven by our results. Second, we only performed an analysis of individual functional SNPs and not for IL6 and IL6R haplotypes. Third, measurements of IL-6 and sIL-6R were not performed in our cohort due to the lack of serum samples, and genotypes could therefore not be correlated to systemic levels. Fourth, we could not investigate whether the association between the IL6 and IL6R variants differed for TCR or ABMR, due to the lack of a standardized assay over the years for the determination of DSA. Lastly, the Banff 2007 classification was used for the histopathological diagnoses, which is an older version. The Banff classification has undergone several revisions since 2007, predominantly related to the criteria for ABMR. The introduction of C4d-negative ABMR in the Banff 2013 classification has significantly impacted the number of ABMR diagnoses overall 54 . Accordingly, our cohort lacks C4d-negative cases of ABMR. On the other hand, major strengths of our study are the large sample size, robust statistical analysis (incl. subgroup analysis), and the hard and clinically relevant endpoint, namely biopsy-proven rejection.
In conclusion, we found that IL6 and IL6R variants in the donor associate with the risk of developing biopsyproven rejection after renal transplantation. These findings imply potential efficacy of targeting IL-6 signalling in renal transplantation. Ongoing, randomized controlled trials with IL-6 or IL-6R inhibitors are needed to identify the best settings, including the timing of intervention and patient selection, in which these agents might be effective.

Methods
Subjects. We enrolled patients who underwent single kidney transplantation at the University Medical Center Groningen in the Netherlands between March 1993 until February 2008. From the 1430 renal transplantations, 1271 recipient and donor pairs were included in the cohort as previously described [55][56][57] . Subjects were excluded due to technical complications during surgery, lack of DNA, re-transplantation or loss of follow-up. This study is in accordance with the declaration of Helsinki and all patients provided written informed consent. www.nature.com/scientificreports/ of IL6 contains several SNPs, of which the rs1800795-174 G > C is the most widely studied for its influence on acute rejection of renal allograft 14 . In addition, we chose the IL6R rs2228145 A > C (formerly rs8192284) SNP, which has previously shown to impair IL-6R signaling and influence the risk of diverse inflammatory diseases 30 . Genotype clustering and calling were performed using the Illuminus clustering algorithm 58 . The overall genotype success rate was 99.5% and 6 samples with a high missing call rate were excluded from subsequent analyses.
Study end-points. The primary end-point in this study was biopsy-proven rejection (all biopsies were reevaluated according to the Banff 2007 classification) after transplantation.
Genetic risk score. We created a genetic IL-6 and IL-6R risk score that assigns points for the presence of a risk-decreasing or a risk-increasing allele in the donor and recipient. However, to take into account the strength of the association of the SNPs with rejection, the point for the presence of an IL6 or IL6R SNP is multiplied by the regression coefficient (= logarithm of the hazard ratio) creating a weighted risk score. A regression coefficient is negative when an SNP is protective and the regression coefficient is positive when an SNP is hazardous. The total sum of the IL6 or IL6R SNPs in both the donor and recipient creates the genetic risk score. Next, we determined whether the genetic risk score improved the prediction of rejection compared to only the IL6R SNP in the donor.
Statistical analysis. Statistical analyses were performed using SPSS version 25. Data are displayed as median [IQR] for non-parametric variables; mean ± standard deviation for parametric variables and the total number of patients with percentage [n (%)] for nominal data. Differences between groups were examined with the Mann-Whitney-U test or the Student t-test for not-normally and normally distributed variables, respectively, and categorical variables with the χ2 test. Log-rank tests were performed between groups to assess the difference in the incidence of biopsy-proven rejection. Univariable analysis was performed to determine the association of genetic, donor, recipient, and transplant characteristics with rejection. The factors identified in these analyses were thereafter tested in a multivariable Cox regression. Additionally, multivariable cox regression with a stepwise forward selection was performed. Harrell's C statistic was used to assess the predictive value of the SNPs when added to the reference model. The additional value of the genetic risk score was assessed by the integrated discrimination improvement (IDI). The IDI indicates the difference between model-based probabilities for events and non-events for the models with and without the genetic risk score. Tests were 2-tailed and regarded as statistically significant when P < 0.05.