Procurement Biopsies in Kidney Transplantation: More Information May Not Lead to Better Decisions

The ongoing shortage of organs available for transplant contributes to prolonged waiting times, death on the waiting list, and higher costs of care for patients with ESKD. Although the total number of deceased donors has risen nationally, the waiting list remains large, with nearly 100,000 candidates currently in need of kidney transplantation.¹ Despite evidence that many nonstandard organs are beneficial for appropriate recipients in need, the discard rate of recovered, potentially transplantable kidneys in the United States increased markedly in the late 1990s, and remains at approximately 20%.¹ The primary reason given for discarding a kidney removed for transplant is “quality,” a determination frequently made on the basis of frozen section biopsy results.² There is marked variation in biopsy practice across United States donation service areas, from 23% to 78% of kidneys recovered.³ Importantly, after adjustment for donor factors and organ procurement organization characteristics, performance of a kidney biopsy is associated with more than three times the odds of kidney discard.³

To identify kidneys at risk of primary nonfunction or early graft failure, frozen section analysis of kidney biopsy prior to transplant is selectively performed to identify chronic or irreversible organ damage. In 1995, a study of 65 kidneys with pretransplant biopsies reported that among eight kidneys with >20% glomerular sclerosis (GS), 87% experienced delayed graft function, and 38% developed graft failure within 6 months.⁴ This study was followed by a trend of increasing discard of kidneys with >20% GS that persists to this day. A recent analysis of United States transplant registry data (2005–2014) for kidneys with a kidney donor profile index >85% found that discards rose from 34% with 0%–10% GS to 77.4% with >20% GS.⁵ However, among kidneys transplanted, 5-year death-censored graft survival differences were modest: 75.8% for 0%–10% GS, 70.9% for >10% GS, and 74.8% for kidneys without biopsy. Thus, selected kidneys with high GS can offer benefit to patients on the waiting list with outcomes superior to dialysis.

The use of frozen section biopsy results to guide organ acceptance decisions has been widely and appropriately challenged. Frozen section biopsies are generally limited to review of a few sections using hematoxylin and eosin staining and may be evaluated by on-call pathologists who lack specialized renal pathology training. Although reasonable correlation has been demonstrated between on-call and specialty pathologists for grading GS (κ=0.86), a recent study found poor agreement on an aggregate overall kidney quality score (κ=0.41).⁶ Further, pathologic evaluation by nonspecialty trained pathologists did not correlate with outcomes, and critically, 20% of grafts graded as unacceptable for transplant by on-call pathologists were deemed transplantable upon retrospective review by trained pathologists. Importantly, inter-rater agreement is limited even among experts. In a seminal publication on the Banff histopathological consensus criteria for preimplantation kidney biopsies, Liapis et al.⁷ reported the inter-rater agreement for a sample of 19 preimplantation biopsies (frozen and permanent) reviewed by 32 trained pathologists. In frozen section core biopsies, the interclass correlation coefficient was <0.4 for all characteristics, including GS (where >0.75 is considered good). The interclass correlation coefficients for interstitial fibrosis, arteriolar hyalinosis, and tubular injury were all <0.1. The findings were only slightly better on permanent sections. These and other studies support the conclusion that the reliance on pathology findings alone, regardless of the individual reading the slide, leads to inappropriate discard.⁸

In an effort to address the limitations inherent in pretransplant biopsy results, multiple scoring systems have been developed, combining various pathologic findings (with or without additional donor factors) to better predict likelihood of early graft failure. The 2014 Maryland Aggregate Pathology Index (MAPI) was developed on the basis of wedge biopsies to combine GS, arterial wall-to-lumen ratio, scar/fibrosis in at least ten tubules, and arteriolar hyalinosis.⁹ Increasing scores were associated with similar 1-year survival but reduced 3-year survival (low MAPI, 84.3%; intermediate MAPI, 56.5%; and high MAPI, 50%; P<0.001), a finding that was replicated at a second unaffiliated center. However, only 5% of allografts had a high MAPI score. In a systemic review, Wang et al.¹⁰ reported that only half of the 15 studies examined demonstrated an association between biopsy scores and graft failure or graft function. In a subsequent study of 142 deceased donor allograft recipients, neither the MAPI or other scoring systems accurately predicted the risk of delayed graft function or graft loss.¹¹ To date, pathologic scoring systems have not been evaluated with large registry trials as detailed pathologic findings are not reported in field-defined data, limiting validation within large representative populations.

Providing transplant centers with images of photomicrographs for evaluation by the accepting center is another proposed solution, but it has also proven problematic in practice. An assessment of agreement between three renal pathologists in reading 100 photomicrographs of frozen section biopsy images found moderate agreement on GS (κ=0.68–0.77 in pairwise comparison) but poor agreement on other findings (e.g., fibrosis [κ=0.32–0.66] and acute tubular injury [κ=0.12–0.18]).¹² A more promising approach may be to apply artificial intelligence (AI) algorithms to allow robust assessment of multilayer sections. In a pilot comparison of AI reading with on-call results and subsequent “gold standard” consensus assessment by three renal pathologists, the AI systems were significantly more accurate in assessing GS (κ=0.89 versus κ=0.71 for on-call pathologists).¹³ Importantly, by reviewing multiple levels, the AI system dramatically reduced the proportion of grafts inappropriately deemed not transplantable, from 14% to 2%.

All current United States studies of biopsy results and outcomes are limited by selection bias. Only organs that were accepted for transplant were included in these analyses. Consequently, these studies cannot truly assess the accuracy of biopsies in assessing outcomes of all recovered organs. To overcome the effect of selection, a recent study compared organ allocation in the United States with that in France and Belgium—where kidneys are not routinely biopsied during allocation—and found no predictive value of biopsies over what is routinely available from donor medical records.¹⁴ In this analysis, 496 United States discarded kidneys were matched to transplanted organs from Europe, which had with graft survival rates of 93.1% at 1 year, 80.7% at 5 years, and 68.9% at 10 years.

Despite 20 years of data demonstrating that biopsies performed at the time of kidney recovery provide limited insight regarding the eventual outcome of renal allografts, >50% of kidneys removed for transplant still undergo a biopsy. Obtaining a biopsy is strongly associated with kidney discard rates, and it may not significantly augment prediction of outcomes beyond estimates using clinical criteria (e.g., age and kidney donor profile index). We believe that United States transplant centers should curtail or eliminate the use of biopsies for standard criteria donors and avoid discard on the basis of biopsy data alone. To further inform this debate, a randomized, controlled trial of biopsy use should be performed to determine if and when a procurement biopsy should be used in the decision to accept or decline an organ offer. A pilot trial to assess the feasibility of surgeon willingness to defer access to biopsy information at the time of offer acceptance is underway (ClinicalTrials.gov identifier NCT03837522). New approaches to procurement biopsy should also be tested in appropriate trials to determine if and when benefits of biopsy (e.g., improved quality-adjusted life-years on the waiting list) outweigh the harms (e.g., death on the waiting list and misuse of scarce resources). Pursuit of evidence to guide appropriate biopsy use and development of strategies to improve the quality of kidneys at higher risk for discard (intervention research), motivate acceptance, and expedite placement are vital priorities to reduce unnecessary discard and increase access to transplant for patients in need (Figure 1).

Figure 1.: Considerations for defining appropriate use of procurement biopsies in kidney transplantation. Shown are some considerations pertinent to the current state of United States procurement biopsy practices, a potential idealized state of biopsy usage, and research needs to bridge the gaps.

Disclosures

D. Axelrod reports consultancy agreements with CareDx, Inc. and Talaris; and is a member of the National Kidney Foundation Policy Committee. K. Lentine reports consultancy agreements with CareDx, Inc. and speakers bureau participation with Sanofi. The remaining author has nothing to disclose.

Funding

K.L. Lentine receives support from Mid-America Transplant Foundation grant NCT03837522.