Influence of renal function on long-term graft survival and patient survival in renal transplant recipients.

OBJECTIVES
Renal function post kidney transplantation is an outcome of interest for both clinicians and regulators evaluating immunosuppressive treatments post-transplantation. The current review sought to provide a synopsis of currently available literature examining the relationship between post-transplantation renal function and long-term graft survival and patient survival.


METHODS
A systematic literature review was performed using the PubMed, EMBASE and Cochrane Library databases. The search strategy was designed based on high level Medical Subject Heading (MeSH) terms and designed to capture studies published in English to 2012 and identified a total of 2683 unique hits; for inclusion studies were required to have >100 patients. Following two rounds of screening, a total of 27 studies were included in the final review (26 of which were identified via the literature review and one study was identified via searches of the reference sections of included studies).


RESULTS
The consensus among studies was that lower post-transplantation GFR, in particular 12 month GFR, was consistently and significantly associated with an increased risk for overall graft loss, death-censored graft loss and all-cause mortality in both univariate and multivariate analyses. The magnitude of the association between reduced GFR and outcomes was greater for death-censored graft loss versus overall graft loss and for graft loss in comparison with overall patient mortality. The predictive utility of GFR alone in predicting long-term outcomes was reported to be limited.


CONCLUSIONS
Lower GFR and greater rates of decline in GFR post-transplantation are associated with an increased risk for graft loss (overall and death-censored) and all-cause mortality; however, the predictive utility of GFR alone in predicting long-term outcomes is limited.


Introduction
Renal transplant is the preferred treatment modality for patients with end-stage renal disease (ESRD) and following successful renal transplantation patients regain renal function, which typically stabilizes at approximately 60% that of donor renal function 1 . Thereafter, a gradual decline in renal function is typically reported 2 , the rate of which may be influenced by numerous variables. These include both donor-and recipient-related characteristics, delayed graft function, and acute rejection. Moreover, calcineurin inhibitor (CNI)-based immunosuppression regimens exhibit dose-related nephrotoxicity, which has been described as the 'Achilles heel' of modern immunosuppression and may be a contributing factor in renal function decline 3 .
The absolute levels of renal function and the rate of renal function decline in turn have implications in terms of both patient and graft survival 4 . Indeed, large scale studies have demonstrated that serum creatinine levels at 1 year post-transplantation are strong predictors of graft survival, patient survival and cardiovascular mortality 5,6 . Renal function is commonly measured either by serum creatinine measurement alone or creatinine clearance based estimations of glomerular filtration rate (GFR). Indeed, stages of chronic kidney disease are defined according to GFR level and KDIGO guidelines recommend that kidney function be assessed by GFR, with serum creatinine measurements alone being considered insufficient.
In addition to absolute levels of renal function, evidence from recent studies has led to the suggestion that the rate of renal function decline is also an influential determinant of both graft and patient survival. In the multinational Patient Outcomes in Renal Transplant (PORT) study the decline in GFR over 3-12 months post-transplantation was associated with both overall graft failure and death-censored graft failure 7 . Additionally, evidence from a US-based study showed a number of factors were associated with a steeper decline in post-transplantation renal function including increased pre-transplant dialysis time, older donor age, diabetes as a primary diagnosis and African American ethnicity 4 .
To more fully elucidate the relationship between posttransplantation renal function and change in renal function, specifically post-transplantation GFR, and long-term outcomes (overall graft survival, death-censored graft survival and patient survival) a systematic literature review was performed to identify studies investigating these associations in adult, single kidney transplant recipients. Studies reporting pharmacokinetic or pharmacodynamic endpoints only, those conducted in recipients of combined kidney and pancreas (or other solid organ) transplants and those conducted exclusively in pediatric populations were excluded. No time limit was applied to the searches with regard to publication date. For inclusion in the final review studies were also required to include !100 patients (studies with !500 patients were considered to be large scale studies). No exclusion criteria were applied with regard to the methods used to measure/estimate GFR (full details of the search terms used are provided in Supplementary Appendix 1).

Methods
Literature searches identified a total of 2683 unique hits (following removal of duplicates). After two rounds of screening (initial screening of titles and abstracts and subsequent full text review of shortlisted publications) a total of 27 publications were identified for inclusion in the final review, 26 of which were identified through the literature searches and one study was identified via searches of the reference sections of included studies ( Figure 1).
Of those studies assessing overall graft survival a total of eight were considered large scale studies with a sample size in excess of 500 patients 7,8,[10][11][12][13]15,16 . Additionally, a total of 5 out of the 11 studies that assessed overall graft survival were conducted exclusively in European populations 9,12,[15][16][17] . Included studies investigated GFR (and change in GFR) at various time points posttransplantation, ranging from 1-3 months to 12 months post-transplantation. The clear consensus among the nine studies that assessed 12 month GFR was that reduced GFR was consistently associated with an elevated risk for overall graft loss 7-10,13-17 . Additionally, in these studies the magnitude of this association became notably stronger with GFR levels530 mL/min/1.73 m 2 . However, the magnitude of this effect also varied notably between studies. For example, the HR for graft failure for patients with GFR 30À59 mL/min/1.73 m 2 ranged from 1.50 in a study by Remport et al. 18 (using !60 mL/min/1.73 m 2 as the reference population) to 8.07 in the study by Marcén et al. 9 (using !90 mL/min/1.73 m 2 as the reference population). Studies reporting overall graft survival included two large scale studies that used data derived from the USRDS database, both of which reported a significant association between lower 12 month GFR and elevated risk for allcause graft loss, with a trend for accelerating risk as GFR declined 10,13 . Findings from the PORT study concurred with those from the USRDS studies and also reported a significant increase in risk for graft loss with lower GFR at 12 months post-transplantation, using GFR 60-90 mL/ min/1.73 m 2 as the reference population 7 . Moreover, in studies performing multivariate analysis, reduced 12 month GFR remained a significant predictor for overall graft loss in all 7-10,13-16 but one study 17 . The association between reduced GFR and overall graft survival also remained consistent across different patient populations and recipients of both standard and extended criteria donor organs.
Notably, reporting of mortality varied substantially between studies with a small number of studies simply reporting mortality rates for patient groups with different levels of renal function over the follow up period, whilst other studies reported HRs for mortality according to GFR level at a given time point or change in GFR over a given time. A total of seven studies reported HRs for overall mortality according to GFR, in all seven studies, lower GFR was associated with an elevated mortality risk 7,10,18,27,[31][32][33] .
A total of seven studies specifically reported on the relationship between 12 month GFR and mortality [7][8][9][10]17,19,20 . Overall, the findings from these studies suggest a strong association between 12 month GFR and mortality. These studies included two large scale analyses from the USRDS database incorporating data from 487,000 patients. A 2012 study by Schnitzler et al. examined the association between estimated glomerular filtration rate (eGFR) at 12 months (using the abbreviated MDRD formula) and longterm outcomes 10 . Their results showed that at an eGFR 570 mL/min there was a significantly increased risk for mortality over a period of 9 years, with mortality risk increasing in an accelerating fashion as eGFR decreased. Additionally The majority of studies identified here that examined the association between eGFR and patient survival were large retrospective database analyses. However, Remport et al. report the results of a prospective single center study collecting 5 year outcomes; the results of this study concurred with the findings of large scale retrospective database studies and showed that the risk of mortality increased with decreasing eGFR (abbreviated MDRD). Moreover, in multivariate analysis Remport et al. showed that for every 10 mL/min/1.73 m 2 decrease in eGFR the HR (95% CI) for mortality was 1.27 (1.12-1.44) p50.001 18 .

Discussion
Overall, the consensus findings of the studies identified in the present literature review was that lower GFR in the first year post-transplantation was strongly associated with poorer graft and patient survival and remained an independent risk factor for graft failure and mortality following adjustment for multiple other risk factors. However, it should be noted that there was substantial heterogeneity between studies in terms of the time point at which GFR was assessed and the follow-up period, which complicates the comparison of findings across studies. Additionally, some studies also assessed the change in GFR between given time points in addition to or instead of GFR at absolute time points.
The nature of the relationship between decreased GFR and increased risk for graft failure and mortality was found to be non-linear with studies consistently reporting a trend for progressively increased risk of graft failure and mortality with decreasing GFR, typically with patients with GFR below 30-40 mL/min/1.73 m 2 having significantly elevated risks for graft failure and mortality. Additionally, the magnitude of the association between graft failure and lower GFR was greater than for mortality and GFR, with the magnitude of the relationship between GFR and death-censored graft survival being greater than for overall graft survival.
The present review was specifically focused on the association between post-transplantation GFR, rather than serum creatinine, and long-term outcomes. Stages of CKD are defined according to GFR level and it is one of the most widely used endpoints in clinical trials and observational studies in renal transplant patients. Although GFR is considered a surrogate endpoint, it is often not possible to measure endpoints including graft failure and mortality. Additionally, the current KDIGO guidelines recommend that kidney function should be estimated using serum creatinine or creatinine clearance based GFR formulae, as measurement of serum creatinine alone is considered insufficient as elevated serum creatinine levels may be due to a loss of muscle mass rather than poor kidney function per se 34 . Additionally, use of creatinine clearance has been shown to overestimate GFR, as serum creatinine is influenced by factors other than GFR; use of measured creatinine clearance has been reported to overestimate GFR by 19% 35 . Similarly, creatinine clearance determined using the Cockcroft-Gault formula overestimates GFR by 16% 35 . Although several different formulae are available to estimate GFR, and a recent systematic review showed them to have similar levels of accuracy, the most commonly used methods are the MDRD, developed by Levey et al. 35 , followed by the Cockcroft-Gault formulae 36 . More accurate measures of GFR may be obtained via measurement of clearance of markers such as inulin or iothalamate; however, this is generally not recommended in routine clinical practice owing to logistical challenges and patient inconvenience. As such, measures of estimated GFR remain the best available option for monitoring GFR in routine clinical practice.
Notably, the majority of studies included in the present review were retrospective in design and therefore associated with the limitations inherent in this type of study design. However, a number of retrospective analyses were performed using data collected prospectively for large scale databases such as the USRDS and that of the PORT study, which provide longitudinal data that permit the investigation of long-term trends and factors that influence both patient and graft survival. Despite the retrospective nature of most articles the review identified a small number of studies (n ¼ 3) that examined the potential role in projecting future outcomes for an individual patient [10][11][12] . The consensus among these studies was that, although GFR was found to be strongly and significantly associated with both graft failure and mortality, the predictive utility of GFR was limited when predicting outcomes at individual level. Both He et al. and Kaplan et al. assessed the utility of GFR formulae in predicting mortality and graft failure using receiver operator characteristic (ROC) curves. The C statistic for mortality ranged from 0.51 to 0.63 in the analysis by He et al., although the corresponding value in the analysis by Kaplan et al. was notably higher at 0.729 (a C statistic of 0.7-0.8 is considered to show acceptable discrimination, a value of 0.8-0.9 indicates excellent discrimination and values !0.9 indicate outstanding discrimination) 37 . Additionally, the sensitivity and specificity of the models were highly dependent on GFR, with sensitivity being greatest at high GFR levels and specificity being greatest at low levels of GFR 11,12 . The predictive value of a risk factor is dependent on the incidence of the outcome of interest, as such GFR may have a higher predictive value in high risk patient sub-populations. Although the studies cited here suggest limited predictive utility in terms of mortality and graft failure, the prognostic value of many risk factors in chronic diseases is often limited 38 . This has been, in part, attributed to the fact that a large number of disease processes are multifactorial. Whilst GFR at 1 year post-transplantation is strongly linked to both graft failure and mortality, numerous factors influence GFR and the rate of GFR decline over time including delayed graft function, acute rejection, primary renal diagnosis and panel reactive antibodies 4 . Similarly, a large number of donor and recipient-related factors including age, primary renal diagnosis and presence of diabetes are also strong independent risk factors for graft failure and mortality. Additionally, the interplay between risk factors for reduced GFR, the influence of GFR on other factors such as cardiovascular disease and the role of other factors such as presence of diabetes means that it is difficult to fully delineate the cause and effect nature of GFR decline on clinical endpoints. As such, it should be noted that when assessing the impact of GFR on both patient and graft survival, the role of other confounding patient characteristics of the study population should be taken into account.

Conclusion
In summary, the consensus among studies examining the association between GFR and long-term clinical outcomes in renal transplant recipients showed that lower GFR is strongly associated with an elevated risk for graft failure and mortality. However, this relationship has been demonstrated mainly in retrospective studies; the few studies that examined the utility of GFR in predicting graft failure and mortality suggest that the predictive utility of GFR is limited, possibly due to the number of factors that may potentially affect GFR and may also independently influence the risk of graft failure or survival. Future prospective studies are required to further investigate the nature of the relationship between eGFR and long-term clinical outcomes; however, the long-term follow up required means that for many investigators this may not be feasible.

Transparency
Declaration of funding This study was funded by Bristol Myers Squibb.
Declaration of financial/other relationships A.K. is a current employee of Bristol Myers Squibb, which provided funding for this manuscript. J.S.-P. and W.V. are current employees of Ossian Health Economics and Communications, which has received consulting fees from Bristol Myers Squibb.