EXPERIMENTAL AND CLINICAL TRANSPLANTATION

Objectives: Delayed graft function is a significant prognostic indicator after renal transplantation. We hypothesized that delayed graft function is not a single entity, and different patterns of delayed graft function reflect various underlying pathological processes.

Materials and Methods: An analysis of 762 renal transplants was performed, showing serum creatinine was charted serially for the first 30 days after transplant. Measurements were obtained: time on hemodialysis; time to peak creatinine; time for creatinine to half; time for creatinine to within 10% of baseline.

Results: Four patterns of delayed graft function were identified. There was no association between pattern of delayed graft function, and 1-year graft survival or serum creatinine at 1 year. Time for creatinine to >15 days was associated with a higher creatinine level at 1 year than it was with patients with time for creatinine to half < 5 days (300.6 ± 54.3 vs 211.3 ± 26.0 μmol/L; P < .01). Patients with 1-year creatinine concentrations > 180 μmol/L had longer time on hemodialysis and time for creatinine to half than did those with 1-year creatinine concentrations ≤ 180 μmol/L (9.2 ± 1.3 μmol/L vs 7.0 ± 0.7 μmol/L; P = .03; and 11.6 ± 1.7 μmol/L vs 6.0 ± 0.4 μmol/L; P < .001). Time for creatinine to half of 6.5 days (sensitivity 67.3%; specificity 79.4%; area under the curve, 0.70) was the best predictor of a 1-year creatinine concentration ≤ 180 μmol/L.

Conclusions: Delayed graft function is not a single entity; rather; it is the most common presentation of a heterogenous variety of pathologies. Its rate of resolution of renal function is the best predictor of long-term graft outcome.

Key words : Transplant, Renal, Delayed graft function, Patterns, Outcomes

Delayed graft function (DGF) is a common and well-recognized complication in the immediate posttrans-plant period, affecting between 20% and 50% of deceased-donor renal transplants.^1,2 It is an important prognostic factor in allograft outcome and is associated with acute rejection episodes and reduced short and long-term graft survival.^3-5 Delayed graft function is more common in extended criteria donor (ECD) and donation after cardiac death (DCD) kidneys.^6-7 With a universal organ shortage, lengthening transplant waiting list, and the need to expand the donor pool to include ECD and DCD kidneys, DGF is likely to be a clinical dilemma in the foreseeable future.

A lack of a consensus definition of DGF complicates interpretation and comparison of the literature. One recent systematic review identified 18 different definitions of DGF.⁸ The most commonly used definition is “the need for dialysis in the first week after transplant”^9,10; however, a myriad of other descriptions exist based on either the need for dialysis, or creatinine and urine output trends.⁸

We postulated that DGF may not be a single entity and that several patterns of DGF with different clinical significance may exist. It is likely that several different underlying processes and injuries continue to the generic clinical condition of DGF. This study aimed to demonstrate that different patterns of DGF exist, and determine if there are definitions and patterns of DGF, which may be predictive of adverse graft outcomes.

A retrospective analysis of all adult patients undergoing renal transplant during the 10-year period between 2000 and 2010 at the Western Infirmary, Glasgow, Scotland, was undertaken (n = 762). The Western Infirmary, Glasgow is a tertiary referral center serving a population of 2.6 million in the West of Scotland. Pediatric renal transplants (< 18 years old) were excluded.

Demographic data were collected on donor (age, sex, cause of death, cold ischemic time) and recipient (age, sex, cause of renal failure, nature and duration of dialysis, previous transplants). Outcome measures including graft loss at 1 year, creatinine at 1 year, mortality at 1 year, biopsy-proven acute rejection (BPAR), and primary nonfunction were recorded. Definitions were used to identify patients with DGF as outlined below. The study was approved by the Institutional Clinical Effectiveness Committee All of the protocols conformed with the ethical guidelines of the 1975 Helsinki Declaration.

Definitions of delayed graft function
Patients with DGF were identified using 10 different definitions commonly quoted in the literature (Table 1).^10-19 The sensitivity and specificity of each DGF definition for predicting BPAR and graft loss at 1 year was obtained from standard formulae. A paired McNemar test was applied to compare these figures. Additionally, mean creatinine at 1 year was compared between the definitions using the t test. P < .05 was considered statistically significant.

Patterns of delayed graft function
All patients with DGF, defined as either the need for hemodialysis at any point in the first 7 days after a transplant or a failure of the serum creatinine to half within the first week, had their serum creatinine charted serially over the 30 days after a transplant to determine the pattern. Measurements were obtained from these graphs to describe the trends in serum creatinine as follows: length of time on hemodialysis (t^HD); time to peak creatinine (t^peak); time for creatinine to half (t^½); time for creatinine to fall within 10% of baseline (t^10%); gradient of decline in creatinine (Crgrad) (ie, maximum creatinine-best creatinine in first 30 days/time from maximum-to-minimum creatinine) maximum creatinine (Crmax); best creatinine in the first 30 days posttransplant (Crmin) (Figure 1). To ensure validity, all measurements were made by 2 independent reviewers (CC, MMcD), and where discrepancy arose, this was resolved by a third reviewer (EA). Patients with primary nonfunction were excluded. The chi-square and Mann-Whitney U tests were used to compare graft and patient outcomes at 1 year. Receiver operator characteristics curves were used to determine the optimal value for each of these measurements to predict favorable graft outcomes. For this purpose, favorable graft outcome was defined as a serum creatinine < 180 μg/L (2 mg/dL) at 1 year after the transplant. Statistical analyses were performed with SPSS software (SPSS: An IBM Company, version 19.0, IBM Corporation, Armonk, NY, USA).

Seven hundred sixty-two patients were underwent renal transplant between 2001 and 2010. One hundred ninety patients (24.9%) had DGF when both patients need hemodialysis in the first week after a transplant or a failure of the creatinine to half in the first week after a transplant. Thirty-nine patients (5.1%) remained on dialysis beyond the first 30 days after transplant. Thirty-six patients (4.7%) had primary nonfunction or early graft loss. These patients were excluded from further analysis. Table 2 outlines demographics for patients with DGF and the overall population.

Definitions of delayed graft function
Table 3 outlines the total number of patients who would be diagnosed with DGF using each of the 10 definitions previously described in Table 1. The sensitivity and specificity of the different definitions of DGF at predicting graft loss at 1 year ranged from 63.5% to 73.0% and 67.1% to 85.7%. Similarly, the sensitivities and specificities of the DGF definitions at predicting BPAR were 34.6% to 46.3% and 80.5% to 84.8% (Table 3). There was no significant difference in mean serum creatinine at 1 year depending on the terminology used to define DGF (range, 178.9 ± 6.3 μmol/L to 186.2 ± 6.3 μmol/L; P = .76).

There were no significant differences between the 3 “dialysis-based” DGF definitions’ sensitivities or specificity for BPAR (P = .67; 0.54) or graft loss at 1 year (P = .92; P = .29). There was a statistically significant difference between the predictive value of the “creatinine-based” and “dialysis-based” definitions of DGF for both graft loss at 1 year (P = .004) and BPAR (P = .001) with the “dialysis-based” definitions having a higher sensitivity and specificity for graft loss and the “creatinine-based” definitions having a higher sensitivity for BPAR.

Patterns of delayed graft function
Four distinct patterns of serum creatinine change with time were observed (Figure 2): (1) a prolonged hemodialysis (HD): a prolonged period of hemodialysis then a fall in creatinine; (2) a single hemodialysis (HD1): a single session of hemodialysis required then be creatinine fell; (3) a slow decline (SD): immediate reduction in creatinine; however, creatinine took a long time to reach baseline and > 1 week for the creatinine to half; and a (4) slow rise then a gradual decline (SR): creatinine initially rose then declined slowly with > 1 week for creatinine to half. No dialysis required.

Extended criteria donor kidneys were more likely to require hemodialysis (type 1, HD), while DCD kidneys more commonly demonstrated type 3 (SD) and type 4 (SR) DGF, where the creatinine was slow to fall but no dialysis was required (Table 4).

There was no association between the pattern of DGF and 1 year graft or patient survival or serum creatinine at 1 year (Table 5). Similarly, tHD, tpeak, and t^10% demonstrated no association with outcomes at 1 year. t^½ greater than 15 days was associated with a higher serum creatinine at 1 year than patients with t^½ less than 5 days (300.6 ± 54.3 μmol/L vs 211.3 ± 26.0 μmol/L; P < .01). Crgrad > 50 was associated with lower creatinine at 1 year than Crgrad ≤ 50 (280.5 ± 33.4 μmol/L vs 195.9 ± 34.2 μmol/dL; P < .001) (Table 5).

Patients with serum creatinine > 180 μmol/L at 1 year had longer t^HD and t^½ than those with serum creatinine at 1 year ≤ 180 μmol/L (9.2+/1.3 vs 7.0 ± 0.7; P = .03 and 11.6 ± 1.7 vs 6.0 ± 0.4; P < .001) (Table 6). Crmin also was higher in those patients with serum creatinine > 180 μmol/L at 1 year (209.7 ± 8.8 μmol/L vs 139.7 ± 3.4 μmol/L; P < .001) (Table 6). t^HD of 6.5 days (sensitivity 58.6%, specificity 55.5%, area under the curve 0.56) t^½ of 6.5 days (sensitivity 67.3%, specificity 79.4%, area under the curve 0.70) and Crgrad of 87.2 (sensitivity 69.4%, specificity 71.2%, area under the curve 0.73) are best predictive of a serum creatinine ≤ 180 μmol/L at 1 year (Table 7).

The association between DGF and adverse graft outcome is well described in the literature.^3-5 However, our results indicate that the sensitivity and specificity of DGF to predict adverse graft outcome vary depending in the definition of DGF used, with “dialysis-based” definitions better predictive of medium-term graft outcomes. The absence of a criterion standard definition of DGF complicates comparing existing studies, and standardizing is required to permit future therapeutic interventional trials.⁸

An increasing length of time required for hemo-dialysis (t^HD) and slower rate of creatinine decline after transplant (t^1/2 and Cr^grad) were associated with higher serum creatinine at 1 year. t^HD and t^½ of 6.5 days are the best predictors of an adverse graft outcome at 1 year. To our knowledge, only 1 other paper that the evaluates the rate of allograft recovery as a predictor of long-term graft outcome.¹⁶ Interestingly, the conclusion of Giral-Classe and colleagues that a requirement for posttransplant hemodialysis > 6 days (but not < 6 d) was associated with worse long-term graft survival is remarkably similar to the cutoff identified in our study. Furthermore, those authors also found (in keeping with our results) that, in patients who did not require hemodialysis posttransplant, DGF as defined by a Cockcroft-Gault creatinine clearance ≤ 10 mL/min of > 6 days also was associated with adverse long-term graft outcomes.¹⁶

We have introduced the novel concept of differing patterns of DGF depending on the rate of rise and/or decline in serum creatinine after a transplant and have identified 4 distinct patterns of DGF. As a result of these observations, it is postulated that DGF is not a single entity, rather a common presentation for a range of different pathologies including acute tubular necrosis, acute rejection, and chronic arteriopathy.

Rather than 4 distinct patterns of DGF, it is speculated that we may actually be witnessing a continuum of graft recovery with progression through several potential phases, including the need for hemodialysis, a slow rising creatinine and then declining creatinine depending on the severity of injury. If this is the case, our findings would indicate that it is the “recovery phase” that is, the speed and gradient of decline in creatinine and the ultimate creatinine level at the end of the recovery phase that are best predictors of long-term graft outcome.

We hypothesize that the various phases of recovery may reflect the various potential injuries (or the severity of injury) incurred by the kidney during the organ recovery and reperfusion processes, that is, the need for hemodialysis may be influenced by the degree of cold ischemia²⁰ or acute kidney injury in the donor, while the rate of decline of creatinine may be influenced by warm ischemia, the agonal phase in DCD donation and ischemia-reperfusion injury. The differing patterns of DGF observed in patients receiving DCD and ECD kidneys would support this assertion, with recipients of ECD kidneys being more likely to require hemodialysis after a transplant and recipients of DCD kidneys less likely to require hemodialysis but having a slower decline in creatinine to ultimate baseline function. Ultimate baseline creatinine is likely to overwhelmingly be influenced by donor-derived features in line with a donor-risk index.^21,22

Future work will focus on more accurate correlation between the pattern or phases of DGF observed and the specific perioperative insults. It may be that interventions and therapies (eg, normothermic reperfusion) are found to alter the pattern of DGF observed.

This study is limited by the fact that it is retrospective and from a single center. Despite evaluating nearly 800 renal transplants, there were relatively small numbers of patients with DGF. Although standard practice now, it was not routine during the follow-up to take routine time zero biopsies; therefore, pathological correlation to support our assertion that different insults may cause different patterns of DGF is lacking. Finally, donor characteristics continue to change with extension of acceptable criteria for extended criteria donors and an exponential increase DCD donation. Given that we have demonstrated distinct patterns of DGF in these patient groups, it is likely that the patterns of DGF seen in clinical practice with continue to evolve.

In conclusion, this work demonstrates different patterns of DGF associated with differing donor types and with varying graft outcomes. In particular, the rate of resolution of renal function once the creatinine has begun to fall (Crgrad) seems the best predictor of medium-term graft outcome. These findings support the assertion that DGF is not a single entity; rather, the common presentation of a heterogenous variety of pathologies. It may be that a new definition of DGF is required to reflect the significant influence that rate of recovery of renal function posttransplant has on ultimate function.

Volume: 13 Issue: 1 February 2015

FULL TEXT