Preimplant Histologic Acute Tubular Necrosis and Allograft Outcomes

Discussion

This multicenter study is the largest cohort to date to examine associations between reported AKI by preimplant histology and allograft outcomes. We demonstrated that preimplant biopsy-reported ATN, as provided with actual organ offers, is modestly associated with the development of DGF only in DCD kidneys. More generally, however, our findings suggest that these ATN reports provide little utility for determining the overall risk of DGF or premature death-censored graft failure in deceased-donor kidney transplantation. In addition, we noted substantial variability in pathology reports with evidence to suggest that acute structural injury is frequently underreported in preimplant kidney biopsies.

In aggregate, prior studies of histologic AKI by preimplantation biopsy have reported inconsistent associations with allograft outcomes (Table 2) (17–23). None of the studies that we identified provided magnitudes of association or confidence intervals for histologic AKI, and only one study analyzed multicenter data (21). Three studies noted a significant association between histologic AKI and DGF (17,21,22), but two others did not (18,19). In their analysis of 92 kidneys, Sulikowski et al. also described worse longer-term allograft function (reported as “higher concentration of serum urea at 12 and 36 months”) for preimplant biopsies with more severe ATN (22), but the two largest prior cohorts (200 and 371 kidneys, respectively) failed to demonstrate a significant association between this exposure and graft survival (20,23). In fact, the latter study by Munivenkatappa et al. which proposed the Maryland Aggregate Pathologic Index to predict graft failure based on preimplant biopsy findings, did not retain ATN for scoring given its lack of association at a mean follow-up of 38±30 months. Apart from these studies of direct associations, however, other types of evidence have suggested the importance of early allograft injury.

Several investigators have studied newer techniques (e.g., quantifiable protein or other biomarkers) that may ultimately improve upon current approaches to preimplant biopsies and modify the way we assess acute structural kidney injury. Schröppel et al. demonstrated that kidney injury molecule-1 (KIM-1) expression in pretransplant biopsies corresponded with donor renal function, but concomitant injury histology was not described (24). Zhang et al. assessed KIM-1 staining in biopsies obtained on average 16–79 weeks after transplant (25). KIM-1 staining was present in all 25 cases that showed histologic evidence of tubular injury, and it was even detected in 7 of 25 cases without histologic injury. Mishra et al. found staining intensity for neutrophil gelatinase–associated lipocalin (NGAL) in 1-hour reperfusion biopsies from 25 children to be associated with donor type (deceased versus living), cold-ischemia time, and peak postoperative creatinine (26). Our investigator group previously reported the first biomarker evidence of a detrimental effect of AKI (detected perioperatively) on 1-year outcomes in a prospective, multicenter deceased-donor kidney transplant cohort (27). We showed that higher first postoperative day urine NGAL and IL-18 concentrations were associated with >5-fold higher adjusted odds for the return to dialysis or an eGFR<30 ml/min per 1.73 m² at 1 year. Furthermore, results from a recent trial showed that therapies successful in ameliorating early allograft injury, as assessed by a reduction in the rate of DGF, can improve 1-year allograft survival (28). Thus, there is mounting evidence to suggest that early AKI is affecting outcomes in kidney transplantation, but that the value of measuring this injury depends on the method of detection. We believe more studies are needed to validate KIM-1, NGAL, IL-18, and other promising biomarkers, with appropriate comparison to histology (via fixed and stained core-needle biopsies read by blinded renal pathologists), to measure AKI at procurement in hopes of improving our ability to predict important allograft outcomes.

Considering the essentially negative results of this study, it is reasonable to question whether ATN, or AKI in general for that matter, truly causes important allograft outcomes. Preimplant ATN can only reflect one aspect of the injury involved (i.e., ischemic injury) and cannot capture the contribution from recipient reperfusion. As such, preimplant biopsies may simply be too early in the process to adequately address the importance of AKI in transplant. Alternatively, currently utilized biopsies may have significant limitations in this context because of several practical concerns related to rushed interpretations by nonrenal pathologists based on frozen sections from wedge biopsy specimens. One might also consider whether kidneys selected for transplant, and thereby included in this cohort, had somewhat minor ATN in the setting of otherwise favorable characteristics such that the overall spectrum of acute injury in these kidneys was relatively inconsequential. Our finding of a positive association in DCD kidneys may actually support this notion regarding other influential characteristics.

Because of the time inherent in allocating and transporting a deceased-donor kidney, some degree of ischemic structural injury may be unavoidable, which may be even more severe in DCD settings given longer periods of warm ischemia before the kidney can be flushed and procured. We were interested to see that ATN was not more frequently reported in DCD kidneys, and that biopsy-reported ATN was associated with DGF only in this subgroup. Others have highlighted the importance of DCD status when evaluating the effect of treating ischemia via machine perfusion on the development of DGF (29). Although reported ATN per se was not associated with premature death-censored graft failure in our cohort, DCD status itself was, a finding that conflicts with prior studies (30,31). This inconsistency may be related to DCD acceptance patterns, in that DCD kidneys accepted for transplant and included in most cohort studies otherwise tend to be of relatively high quality. Evidence for this differential donor quality can be seen by the lower terminal creatinines in our DCD kidneys compared with non-DCD kidneys. Nonetheless, our cohort consisted of kidneys that were specifically chosen to undergo preimplant biopsy, which likely excluded some high-quality DCD kidneys that were not chosen for biopsy.

Although other methods exist for assessing kidney injury at procurement, the current findings are important to transplantation because histopathology can be considered the gold standard measure for this risk factor. As previously noted, however, the quality of currently utilized preimplant biopsies for detecting acute injury may be limited for several reasons. Preimplant biopsies require rapid turnaround for results within the time constraints of organ allocation. As such, these specimens are typically frozen sections from wedge biopsy samples read by on-call pathologists rather than immediately fixed and subsequently stained core-needle biopsies read by dedicated renal pathologists. In addition, the focus of the preimplant pathology evaluation is often on chronic disease rather than AKI. Thus, histologic ATN can be overlooked in many cases, and our data regarding variability in ATN reporting between OPOs as well as misclassification based on a manual review of a small number of biopsies from one OPO provide strong evidence for this. To address this potential bias, we performed a sensitivity analysis in which we excluded all kidneys from the one OPO with no reports of ATN, which did not affect our study conclusions. Most importantly, however, our conclusions speak to the utility of currently utilized pathology reports for identifying ATN during organ offers and not the theoretical value of truly quantifying all renal tubular injury.

As for other limitations, our cohort was defined by available preimplant biopsy reports for kidneys that were ultimately transplanted in accordance with current allocation practices. Thus, we could not evaluate the effect of biopsy-reported ATN on allograft outcomes in kidneys that were discarded for any reason (typically older, higher-risk donors) or in kidneys not chosen to undergo preimplant biopsy (typically younger, healthier donors). These important selection biases affect any observational study of preimplant kidney biopsy, and a properly designed randomized controlled trial will be required to adequately address this uncertainty for the transplant community. In addition, we utilized the UNOS database to determine outcomes given the inclusion of kidneys from different OPOs; however, we view our multicenter study design as a strength with regard to generalizability, and we analyzed additional key donor and allograft characteristics not available from UNOS as potential confounders.

In summary, we found a modest association between preimplant biopsy-reported ATN and the development of DGF for DCD kidneys, but no significant associations overall between reported ATN and the outcomes of DGF or premature death-censored graft failure. The potential effects of more rigorous ATN reporting remain unclear; however, our findings indicate that ATN, as described in typical deceased-donor pathology reports at the time of organ offer, is unlikely to be a good classifier of important recipient outcomes on its own. Innovative clinical trial data are needed to inform the transplant community about the true utility of preimplant kidney biopsy and how biopsy should affect organ acceptance/rejection patterns. Ancillary studies to such a trial could also compare newer methods of defining/detecting AKI against traditional methods such as histology as well as evaluate the overall clinical significance of structural ischemic kidney injury, especially in relation to other increasingly common and important factors such as DCD status.