Risk Factors Influencing the Outcomes of Kidney Re-Transplantation

Background

Shortly after kidney transplantation was introduced, graft failure commonly meant patient death, since no kidney replacement therapy existed. Thus, to save lives, re-transplantation (TX2) was introduced in 1963; however, at that time, the 1-year patient survival was only 60% [1]. The Hannover Medical School transplant program is one of the largest in Germany, with 654 patients on the active waiting list (February 2020). Of these, 123 patients (19%) have been transplanted before and 22 (4%) more than once. In Germany, the interminable waiting list makes matters even more acute [2,3]. We sought to learn the particular hazards faced by re-transplanted patients. Repeat transplantation has been the subject of numerous publications since 1974, demonstrating increasingly better results in graft and patient survival [4–11]. In some reports, hardly any difference in graft or patient survival was found [10,11]. However, currently the results are conflicting and sometimes difficult to interpret. We conducted a retrospective analysis of re-transplanted patients compared to a matched control group of patients with first transplantation to identify particular risk factors for graft and patient survival, so that inherent problems might be addressed prospectively in future studies.

Material and Methods

STATISTICAL ANALYSIS:

We performed a retrospective, single-center, matched-pair investigation. Matching was performed on a 1: 1 basis where patients with a second transplantation served as cases (TX2 group) and patients who received their first transplant during the same period served as controls (TX1 group). Matching criteria were recipient age (±10 years), sex, living/non-living kidney donation, and transplantation date (±18 months). Descriptive analysis of the data included presenting relative and absolute frequencies for categorical data. Continuous variables are presented as arithmetic mean and standard deviation. Baseline characteristics of cases and controls were compared with a paired t test for continuous variables and McNemar’s test for binary outcomes. A time-to-event analysis was performed for cases and controls separately. Graft survival or patient survival served as an event. Times were censored for graft survival at the date of the last hospital visit if a patient was lost to follow-up or had a functioning graft. For patient survival, times were censored at the last hospital visit if a patient was lost to follow-up or had a functioning graft or a transplant failure. Kaplan-Meier curves were plotted for cases (TX2) and controls (TX1) separately. A log-rank test and a Cox regression model were used to compare relevant covariables. To account for the matched-pair design, a marginal Cox regression model was set up to compare covariables as well as cases and controls. To compare graft and patient survival between TX1 and TX2, we used a test as described in Klein and Moeschberger [12]. Multivariable Cox regression models are presented for a set of 8 covariables that were deemed of high relevance. The full model is presented as well as models after variable selection. We applied several approaches (backward selection, score-based best subset selection, and stepwise selection). All analyses were done with SAS 9.3 (SAS Institute Inc., Cary, NC, USA).

Results

KAPLAN-MEIER CURVES:

For the variables ≥50 or <50 years of age, HLA mismatches, initial graft function, post-transplant serum creatinine at hospital discharge, humoral rejection, cardiovascular disease, number of comorbidities, and severe infection, differences of the clinical course in graft and/or patient survival were tested by Kaplan-Meier curves. TX2 patients had a lower graft and patient survival associated with vs without severe infection (Figure 2) and with 3–5 vs 0–2 comorbidities (Figure 3). Graft and patient survival were also lower in TX2 patients with 4–6 vs 0–3 HLA mismatches (log-rank p=0.015 and 0.004, respectively). TX2 patients had a lower graft survival with serum creatinine ≥150 vs <150μmol/L at hospital discharge (Supplementary Figure 1, log-rank p=0.003). Despite the same number of humoral rejections, graft survival only of TX2 patients was lower with vs without humoral rejection (Supplementary Figure 2, log-rank p=0.013). Both cohorts had a lower graft and patient survival associated with vs without cardiovascular disease (Figure 4), as well as a lower patient survival in patients ≥50 vs <50 years of age (log-rank TX2 p=0.018 and TX1 p=0.027), and in patients with malignancy (log-rank TX2 p=0.038 and TX1 p<0.001). Both cohorts had a higher graft survival with initial graft function (log-rank TX2 p=0.002 and TX1 0.038).

COX REGRESSION ANALYSIS:

Regarding graft and patient survival, higher and lower hazard ratios (HR) for the different covariables are enumerated in Supplementary Tables 1 and 2.

Multivariable analysis for graft survival showed that only TX2 and not TX1 patients had a higher risk associated with early inferior graft function; while both cohorts had a higher risk together with cardiovascular disease, severe infection, and humoral rejection (Table 3A). Patient survival showed that only TX2 and not TX1 patients had a higher risk for death together with a high number of comorbidities, while both cohorts had a higher death risk together with cardiovascular disease and severe infection (Table 3B).

Discussion

We found that graft and patient survival in TX2 patients were inferior to TX1 patients (Figure 1). There were numerous explanations in terms of pre-transplant conditions. TX2 patients were confronted with longer times on dialysis, they had a greater immunological risk by their higher immunization rate, and had more HLA mismatches by avoidance of HLA-matches related to their high PRAs (Table 1). Thus, TX2 patients were more likely to receive rATG induction therapy (Table 1).

Inspecting outcomes in detail, TX2 patients generally had lower rates of graft function, more commonly suffered graft loss from rejection, and more often died of severe infection (Table 2). Multivariable Cox regression models for 8 important covariables showed that fundamentally both cohorts had some similar problems with graft and patient survival, namely a risk associated with severe infection and cardiovascular disease (Table 3A, 3B). However, TX2 and not TX1 patients additionally had a lower graft survival with early restricted graft function (Supplementary Figure 1) and had a higher death risk with a high number of comorbidities (Table 3B). This was confirmed in Kaplan-Meier analysis. In patients with cardiovascular disease, the course of graft and patient survival was inferior in both cohorts (Figure 4). However, in TX2 patients, this relationship was more robust, as seen in the distinctly inferior Kaplan-Meier curve of the TX2 cohort (Figure 4A, 4B). The same is true for patient survival in patients with a high number of comorbidities and a distinctly inferior Kaplan-Meier curve in TX2 patients (Figure 3A, 3B).

In the TX2 group, most infections were in the early transplant period (≤6 months) compared to the TX1 group (Table 4, Figure 4A). This state of affairs at least partially may be due to the more frequent use of rATG induction therapy. The TX2 group had a higher immunization rate, and patients were treated accordingly (Table 1). Rabbit ATG is very efficacious in the prophylaxis of rejection episodes in patients at risk of rejection [13,14]. T-cell or T-cell subset suppression even in low-dose treatment of rATG lasts 6 to 12 months until slight recovery [15,16]. It is not clear how long the propensity to infection lasts after the use of rATG, and the timely coincidence of rATG and severe infection during the first 6 months after transplantation is only an indirect sign of a possibly causal role of rATG [14,17].

Transplant nephrectomy of the preceding failed transplant was not an advantage regarding rejection in the second graft, which confirms recent studies [18,19]. TX2 patients had a distinctly lower BMI than TX1 patients (Table 1 and Table 3A, 3B), which has been reported before [6,11]. This finding may be due to the longer cumulative dialysis vintage (Table 1 and Table 3A, 3B) resulting in weight loss associated with mortality [20,21]. However, compared to dialysis, after transplantation we did not find a low BMI correlating to mortality, but within both cohorts we found an improved patient survival at least tested against a BMI above the limit of overweight and obesity (≥25; Table 3B). For obese transplant patients, in general an inferior graft and patient survival, compared to non-obese patients, has been reported [6,22,23].

We did not find an influence of the longer dialysis vintage on graft survival of the TX2 compared to the TX1 group (Table 1 and Table 3A, 3B). Indeed, in former publications such an impact has been convincingly reported and was regarded as one of the main negative influences on graft survival [24]. However, in more recent publications this was no longer found. This can be attributed in general to the better dialysis conditions and especially to the reduced need for blood transfusions because of the regular supply of the dialysis patients with erythropoietin and iron; thus, a main source of pre-transplant immunization has reduced [25,26].

Another reason for a continuously better graft survival of repeat transplantation is the development of better techniques of HLA class II typing as well as HLA-antibody detection [27–32]. Complement-dependent crossmatch had for a long time been the only method to prevent hyperacute rejection and to test for HLA antibodies [33,34]. HLA typing and antibody detection has improved the second graft survival rate since 1974 [1,5,9–11]. However, all-cause mortality is still influenced by a longer dialysis vintage [25,26,35]. We did not find a direct association between duration of dialysis vintage and mortality, but we saw that TX2 patients had a higher mortality caused by severe infection (Tables 2, 4), associated with a high number of comorbidities, and with cardiovascular disease (Figures 3, 4). Thus, the higher morbidity of TX2 patients causing mortality can be interpreted as an indirect consequence of the longer dialysis vintage as one of the basal conditions causing this morbidity.

We compare our results with those reported earlier. Before 2005, TX2 patients generally had worse outcomes except under selected favorable subgroup conditions [4,24]. However, these cohorts were usually not well defined and living-donor status and cyclosporine treatment were not always mentioned. Since 2007, the cohorts of second and first transplantation were mostly compared studying large-registry databases [5,6,7,9,10]. Since then, the transplant outcome gap between both cohorts has constantly narrowed but did not become similar. Registry data, with their multicenter sources, different modes of access to transplantation, and different immunosuppressive treatments, cannot be easily compared to single-center studies. A strength of the present study is the fact that we compared 2 rather uniform cohorts transplanted in a limited similar 10-year timeframe under controlled conditions by the same team and with a follow-up period of 10 years after the end of the study. TX2 results are inferior to TX1-matched controls (Figure 1). This finding contradicts a large single-center study of 3000 patients recently published [11]. However, in that study, living donation was common (71% of TX2 patients), which makes the study less comparable to ours. We were interested to learn that 2 or more re-transplantations have no worse graft and patient survival than 1 re-transplantation [36].

The factors influencing poorer graft and patient outcome may represent a more general inability of overcoming unfavorable conditions of the post-transplant course. This state of affairs brings to mind impaired resistance in TX2, compared to TX1 patients, a condition also called frailty. Frailty has been described and standardized in older patients as indicating physical and cognitive pre-aging and has been related to decreased physiologic reserve and resistance to stressors [37]. Chronic kidney disease has been found to be associated with a higher frailty score, increasing with the stage of renal insufficiency [38,39]. After renal transplantation, frail patients have a higher risk of death [40]. Here, we did not measure physical and/or cognitive parameters, which was not possible because of the retrospective character of the analysis. However, loss of body weight and a lower BMI, as observed in TX2 patients (Table 1 and Table 3A, 3B) are main features of frailty [37]. Conceivably, DNA methylation could be measured in such patients [41], and epigenetic age acceleration has been found to be correlated to other physical and cognitive frailty parameters. Nevertheless, this has mostly been evaluated in large cohorts [42].

The poorer performance of repeat transplantation is related to several factors regarding immunologic, graft-quality, and infectious problems, of which the most important for patient survival were a high number of comorbidities and severity of infections (Table 3A, 3B, Figures 2–4). Immunologic sensitization together with more HLA mismatches contributes (Table 1). This situation more often implies a higher immunosuppressive induction therapy, as was the rule in the TX2 patient cohort by rATG (Table 1), favoring early post-transplant infections. The combination of a higher immunological risk and a possibly generally higher underlying frailty of the TX2 group may be the causal factors for the inferior graft and patient survival of the TX2 patients compared to matched patients with a first transplant. Dealing with comorbidities in any clinical setting is not trivial. In this population, the task is daunting. Nevertheless, our data underscore that this state of affairs could be addressed with diagnostic controls as well as appropriate therapies.

The alternative to repeat transplantation is dialysis. As known from the literature, first transplantation as well as re-transplantation are both clearly better than dialysis; therefore, even preemptive second transplantation has been proposed [43–45]. The relative risk reduction by transplantation seems to be higher in re-transplanted patients because of their higher mortality on the waiting list [10]. To avoid that fate, these higher-risk patients deserve all the chances we can provide by repeated transplantation [46,47].

Conclusions

TX2 graft and patient survivals are inferior to TX1. However, we have not only to compare TX2 to TX1 but also compare them to dialysis patients as the alternative treatment option, and dialysis would mean an even higher mortality. The higher number of comorbidities is, beside immunologic and infectious problems, the main risk factor for inferior outcomes of TX2 patients. Therefore, we should try as far as possible to address comorbidities by preventing and treating them.