Predictive factors of graft dysfunction and long-term kidney allograft failure

Abstract

Kidney transplantation is considered the treatment of choice for many patients with endstage chronic kidney disease; however, despite advancements in short-term allograft survival, long-term survival has not paralleled this improvement. Due to the inevitable ischemic damage and associated reperfusion injury, delayed graft function (DGF) is a common complication after kidney transplantation, which may negatively affect graft survival. Because serum creatinine (SCr) and other traditional markers of kidney injury are insensitive and delayed in the detection of the early stages of kidney damage and DGF, there has been a keen interest in the identification of novel biomarkers for the early detection of allograft dysfunction that could expedite treatment and improve long-term patient and graft survival. Biomarkers are characteristics that can be objectively measured in a biological sample. In clinical settings, biomarkers enable the diagnosis of a dysfunction or disease and, in some cases, they are used to monitor a treatment or to make a prognosis regarding the future outcome of a patient. The analysis of predictive factors of graft dysfunction and long-term kidney allograft failure focusing on novel biomarkers was the major motivation for this work. Thus, the general aim of this thesis was to investigate the potential of different biomarkers to reliably diagnose and predict early graft dysfunction and their effect on long-term kidney allograft failure as well as to gain insight into the underlying mechanisms of graft dysfunction.

Patients and Methods:The study involved three cohorts of patients: two retrospective cohorts that included kidney transplant recipients selected from a database that contained transplant and follow-up information on kidney transplants performed between 1983 and 2008 (for the first retrospective cohort) or 2012 (for the second retrospective cohort); and one prospective cohort that included 40 patients undergoing kidney transplantation between December 2009 and June 2010. The first retrospective cohort was used to validate the one-year SCr as a surrogate endpoint of long-term graft survival, and the second retrospective cohort was considered to analyze the impact of DGF (defined by the need for dialysis during the first week after kidney transplantation) on graft and patient survival using a competing risks approach. The studies based on the prospective cohort had a longitudinal observational design, which was initiated at the time of transplantation; this cohort was used to examine nine potential candidate biomarkers for the early diagnosis of DGF (one biomarker in urine and eight biomarkers in blood): cystatin C (CysC), neutrophil gelatinase-associated lipocalin (N.GAL), leptin and adiponectin, malondialdehyde (M.D.A), superoxide dismutase (S.OD), glutathione reductases (GR), peroxidases (GPx) and total antioxidant status (TAS). Five samples per patient were collected within the first week: 3 to 6 h prior to transplant surgery (pre-transplant); on the subsequent morning at approximately 8 to 12 h after graft reperfusion (day-1); and then on the second (day-2), fourth (day-4) and seventh (day-7) days after transplant, which resulted in five samples per patient. A linear mixed effects model was used to evaluate the longitudinal changes of the potential new biomarkers of early graft dysfunction over the first week after kidney transplantation and to identify the factors associated with these changes. The performance of the candidate biomarkers in the prediction of DGF was examined using receiver-operating characteristic (R.OC) curves. Survival analysis methods, including a survival analysis that accounted for competing risks were used to identify the predictive factors of long-term graft survival.

Results: Of the large number of variables that were considered, the SCr levels at 1, 6 and 12 months following kidney transplantation, as well as the changes between 1 and 6 months and between 6 and 12 months were independently associated with late graft failure. The R.OC curves identified urinary NGAL, MDA and CysC on the first postoperative day as moderately (NGAL) and highly (MDA and CysC) accurate in the prediction of DGF. Both urinary NGAL (at days 4 and 7) and MDA (day-7) were independently associated with one-year graft function, adjusting for variables that typically affect graft function, including acute rejection episodes and re-admissions during the first post-transplant year. Leptin at day-1 was slightly better than SCr in the prediction of the need for dialysis within the first week post-transplant, whereas adiponectin, SOD, GR, GPx and TAS were not. A triple-biomarker approach that used SCr, CysC, and MDA measured 8 to 12 h after kidney transplantation, was the most informative combination, which resulted in an increased ability (AUC=0.96) to distinguish patients with graft damage who would require dialysis within the first week. The application of a subdistribution regression model for competing risks indicated that DGF by itself and independent of acute rejection had a detrimental effect on long-term graft survival, but not on patient survival.

Conclusions: Independent of acute rejection, DGF per se was significantly associated with poor-graft survival, but not with patient survival. Urinary NGAL and serum CysC and MDA were early, noninvasive, and accurate predictors of both the need for dialysis within the first week of kidney transplantation and one-year graft function. A triple-biomarker approach using SCr, CysC and MDA were highly predictive of DGF. Combining biomarkers from different pathophysiologic pathways appears to be a rational and reliable strategy to optimize sensitivity and specificity and obtain additive diagnostic and prognostic information.