Risk factors for rejection after deceased donor kidney transplantation: a mono-institutional analysis of paired kidneys

Deceased donor kidney transplantation is an important therapeutic option for end-stage renal diseases. Adverse events including acute rejection after deceased donor kidney transplantation are not uncommon and result in poor transplant outcomes. Exploration of risk factors and patient stratication is increasingly signicant to improve graft survival. This study aim to evaluate and identify the risk factors for treated rejection of patients after deceased donor kidney transplantation. Clinical and immunological data of deceased donors and corresponding recipients between 2015 and 2018 in West China Hospital were retrospectively collected. The Kolmogorov-Smirnov test was used to indicate distribution of variables. Univariate comparisons of baseline characteristics were made with Chi-square, t and Mann-Whitney U tests. Logistic regression was constructed to analysis potential risk factors. Receiver operating characteristic curve and Jordan index were generated to determine optimal cut-off value for continuous variables.


Introduction
Worldwide, kidney transplantation from deceased donors is an important therapeutic option for patients with end-stage kidney diseases [1]. With improvement of surgical conditions and reduction of ischemic time in recent years, the long-term survival of recipients after deceased donor kidney transplantation (DDKT) appears to be possible. Thus, researches on complications related to graft failure and mortality after organ transplantation are essential [2]. Despite some risk factors associated outcomes after DDKT have been documented before, recent evidence reveals that short-term adverse events, especially acute rejection, are still the principal reason for increased incidence of graft loss, re-transplantation and even death [3,4]. On the other hand, it is undoubted that kidneys from deceased donors inevitably suffer various injuries in donation for many reasons [5]. It is reported that kidneys from donation after cardiac death (DCD) donors are susceptible to warm ischaemia, whilst kidneys from donation after brain-stem death (DBD) incur greater metabolic disturbance and in ammatory response [6,7]. Thus, intricate harmful factors may limited graft survival in DDKT and require unremitting investigation [8,9].
Actually, the association of rejection in DBD with inferior graft outcomes have been established prior to this study [6]. Meanwhile, the de nition, diagnostic criteria, determination and treatment of rejection have been re ned [10]. However, the outcomes of DDKT might be affected by speci c allocation, health care, surgical experience and complicated risk factors related to donors and recipients, causing inevitable confounding factors and statistical bias. Thus, analysis of risk factors basing on paired donor and recipient might be conducive to reduce the bias, make correct decision on allocation and adequate preparation for recipients. Taken together, the exploration of risk factors and patients strati cation are increasingly signi cant to improve graft survival [1]. This study aim to evaluate and identify the risk factors for treated rejection of DDKT basing on the data of paired kidneys.

Patients and data collection
We collected detailed data of both donors and recipients performed DDKT between 2015 and 2018 from the medical archives of West China Hospital. The end of follow-up period in this study is December 2020. After exclusion of 8 DDKT donors and 16 paired recipients according to exclusion criteria such as lack of major perioperative parameters, insu cient follow-up data, dual organ transplantation and donor age older than 65, we included 123 pairs and analyzed their information ( Supplementary Fig. 1). Baseline characteristics included recipient pro le, donor characteristics and transplantation-related features, which were showed in Tables 1-3. Recipient age were not restricted, and donor-recipient characteristics were matched correspondingly to analyze covariates better and control confounding variables. For each recipient, clinical and laboratory data within two years after transplantation were obtained. Bold gures indicate as statistical signi cance at P < 0.10.  Abbreviations: HBV hepatitis B virus; DBD donation after brain death; DCD donation after cardiac death.
Bold gures indicate as statistical signi cance at P < 0.10.

Clinical outcomes
The primary clinical outcomes was treated rejection (TR) de ned as treatment for rejection within 24 months after DDKT. Meanwhile, the baseline characteristics of 123 pairs and transplantation were assessed as covariant for exploring potential risk factors. DGF was de ned as receiving dialysis within 1 week following transplantation. Overall graft loss was regaining permanent dialysis after transplantation or death with functioning graft by any cause. Cold ischemic time (CIT) was de ned by the time from cold storage to reperfusion following implantation. Levels of serum creatinine in recipients at preoperation and 24 months after transplantation were available for evaluation of recipient renal function and graft performance, respectively.

Statistical analysis
Statistical analyses in this study were conducted using SPSS 23.0 (SPSS Inc, Chicago, USA). The Kolmogorov-Smirnov test was used to indicate distribution of variables. Univariate comparisons of baseline characteristics between transplants suffered TR versus no rejection were made with chi-square tests, t tests and Mann-Whitney U test, as appropriate. Additionally, variables were considered as statistical signi cance at P values less than 0.10, which might be conducive to seek possible correlation. Logistic regression model was constructed to analysis potential risk factors. Possibly signi cant characteristics of recipients, paired donors and transplantation in prior correlation analyses were incorporated into regression model as covariates. Variables with statistically signi cance in univariate analysis were chosen into multivariate analysis. In multivariate analysis, the stepwise regression method was selected to prevent multicollinearity. Receiver operating characteristic (ROC) curve and corresponding Jordan index were generated to determine optimal cut-off value for continuous variables included regression model. In univariate and multivariate analysis, P values were two-sided and statistical signi cance was de ned as P < 0.05.  Table 1-3.

Risk factors associated with rejection
Optimal cut-off values of signi cant continuous variables in preliminary correlations analysis were con rmed via ROC curve and corresponding Jordan index. After that, dichotomous variables were generated and entered into univariate and multivariate analysis to explore independent risk factors. It is revealed in univariate analysis that several variables of recipients and transplantation were strongly associated with TR (  Bold gures indicate as statistical signi cance at P < 0.05.

Discussion
For comprehensive analysis of outcomes from DDKT cohort, our study involved 123 donors and 246 recipients with follow-up period of 2 years at least after transplantation and detected several possible risk factors for TR. In this study, donor pro les were matched to corresponding recipients for analysis, reducing feasible selection biases in evaluation of relationship between risk factors and outcomes, and may provide pragmatic value in clinical practice.
Given the great infectious risk from over-immunosuppression caused by imbalance between immunosuppressive protocols and occurrence of rejection, appropriate strati cation of recipient is important to clinical practices. Not only could be induction regimen individually tailored for each recipient, but also immunosuppression medication be personalized basing on the pro le of immunologic hazard. Hence the assessment of risk factors for rejection would be bene cial to improve graft survival and long-term prognosis of patients.
Traditionally, re-transplantation, grafts from deceased donor and high level of panel reactive antibody (PRA) have been reportedly associated with increased risk of graft loss and rejection after transplantation [11,12]. In current cohort, these risk factors were also evaluated and none of them demonstrated signi cant relevance with rejection in multivariate analysis. However, apart from HLA-DQ mismatch as an independent predictor of rejection was con rmed ( Platelet, neutrophil and neutrophil-to-lymphocyte ratio (NLR) have been seemed as the surrogates for in ammatory severity which positively correlated prognosis in several diseases [13,14]. It's speculated that these parameters or ratios could re ect the systemic in ammation which might have adverse impacts on hematologic cell lines and subsequently result in alteration of their ratios [14,15]. Current study indicated a positive correlation of both PLT and ANC with TR (HR 2.202, 95%CI:1.051-4.617, P = 0.037 and HR 2.164, 95%CI:1.018-4.599, P = 0.045, respectively). Our hypothesis is that elevated preoperative PLT and ANC of recipients maybe represent robust in ammatory response or over-activated immune system by any cause, which may underlie the pathogenesis of rejection. Thus, for recipients with high preoperative PLT and ANC, aggressive regimen of induction and maintenance immunosuppression could be considered to decrease the risk of rejection in these patients.
Another unexpected nding in our analysis was that preoperative UA levels revealed independent association with TR. However, the comprehensive effect of UA on graft outcomes still remains controversial in published studies [16]. Although hyperuricemia could result in deterioration of renal disease by inducing endothelial dysfunction and in ammatory dysregulation, it is hard to identify that UA is a immediate cause of renal disease due to unclear causal link between elevated UA and impaired renal function [17]. It is unclear why UA could be an independent risk factor for TR in our investigation. However, to our knowledge, the association of decreased UA with reduced graft-versus-host disease (GVHD) in allogeneic stem cell transplantation (allo-SCT) has been veri ed by animal models, and levels of serum UA could be used as predictor for allo-SCT outcome [18]. We therefore assume that higher level of UA from reduced renal clearance might initiate non-infectious in ammation and contribute to immune reconstitution, which increased the risk of rejection. Moreover, it is not unusual that hyperuricemia would be concomitant with end-stage renal diseases. Thus, although further studies are needed to con rm our results, it is necessary to address the hyperuricemia during perioperative period.
In addition to retrospective design, this study has several inherent limitations. On one hand, despite characteristics of donors were matched to paired recipients and analysed with recipients data at once to control potential confounding variables, regression residual is an important and iterative confounding factor in observational studies. On the other hand, for the sake of reducing negative effects of multicollinearity, the stepwise method was adopted in regression model with matched donor-recipient, probably leading some variables were marginalized by those with more statistical weight. Finally, our data originated from single center, which may somewhat limit its feasibility and relevance in other settings. However, although this retrospective study based on a single-center cohort and need further validation, heterogeneity of large dataset from multi-center or even transnational registry could be signi cantly reduced in current analysis.

Conclusions
Our study found that several hemato-biochemical and transplantation-related parameters might be independent risk factors for treated rejection after DDKT. Actually, the exploration on these inexpensive, easily obtainable and potentially reversible indicators may contribute to stratify patients and develop personalized regimen in perioperative period for better graft outcomes. We hope our work could motivate further meta-analysis and clinical studies to provide more high-level evidence.

Declarations
Competing interests: