Is it time to drop the use of race from KDRI calculator?

Introduction

The term race has been used since the 18^th century to categorize individuals based on physical traits including skin color. Individuals with darker skin color were restricted to inferior social statuses based on the erroneous eugenic belief that they were biologically inferior. Today, the US census data continues to capture race to determine allocation of resources and political representation. As a result, race captures epidemiologic information such as social determinants of health and environmental exposures which affects the incidence of diseases and their outcomes. Nevertheless, it is important to note that not all members within a racial group experience social, environmental, and biological measures equally; and therefore, race is not a biological construct. Despite this fact, race continues to be widely used in medicine for clinical decision making and clinical research on the premise that it captures intrinsic biological differences between groups.

There is growing concern that race-based practice may perpetuate provider bias and amplify racial inequities in access and delivery of health-care. In the context of kidney transplantation, Black race of the deceased donor has been reported to increase the risk of graft failure by 20% which has led to incorporation of donor race into a calculator to assess organ quality. The national organ allocation system is currently designed to match organ to recipient longevity. Hence including donor race in calculators of organ quality has a direct impact on organ allocation, acceptance, and non-use. Given the recent advances in the field of genetics coupled with our growing multiracial society, it has been proposed that genetic ancestry markers be used instead of race to improve the performance of these calculators. In this paper, we will review the origin of kidney allograft quality calculator, the impact of removing donor race variable from the calculator and replacing it with ancestry markers and propose an action plan.

Origin of Kidney Donor Risk Index equation

Kidney Donor Risk Index (KDRI) was proposed in 2009 to provide a continuous score for quantifying the risk of graft failure after deceased donor kidney transplant. It was developed using data obtained from the Scientific Registry of Transplant Recipients (SRTR) by analyzing outcomes of 69,440 deceased donor kidney transplants occurring from 1995 to 2005.¹ It included a greater number of deceased donor characteristics affecting allograft outcomes than the binary Expanded Criteria Donor (ECD, determined using donor age, serum creatinine and history of hypertension and stroke) indicator of kidney quality in use at the time. ECD status was determined using. The original KDRI calculator used ten donor and four transplant characteristics, including Black donor race. A modified version of the KDRI using ten donor variables is commonly used due to availability of these variables at the time of organ offer. These include donor age, height, weight, race/ethnicity, hypertension, diabetes, cause of death, serum creatinine, Hepatitis C virus serostatus and donation after circulatory death (DCD) criteria donor. From this abbreviated version of KDRI, a Kidney Donor Profile Index (KDPI) score is calculated on a percentile scale which quantitates the quality of the deceased donor kidney relative to other deceased donor kidneys recovered for transplantation in the prior calendar year. The performance of KDPI is mediocre (the c-statistic of KDPI is 0.60 which is slightly better than chance alone). Kidney transplant outcomes are a complex interplay of donor, recipient, and transplant factors, as a result of which any calculator solely based on donor characteristics will not perform very well.

Despite its limitations, the OPTN has been providing KDPI with every organ offer since March 2012. KDPI is used allows stratification of organ offers under the revised Kidney Allocation System (KAS) implemented in December 2014 to match donor organ to recipient longevity. KDPI is also utilized by transplant physicians to assess organ quality and discuss organ offers with potential recipients. The KAS allocation sequence follows the KDPI categories of: < 20, 21–34, 35–85, and >85%. Inadvertently, introduction of KDPI has resulted in labeling effect wherein kidneys with KDPI 85%, referred to as high-risk kidneys, are harder to get accepted for transplantation and more likely to not be used.² In the current KDRI calculator, Black donor race has hazard ratio of 1.196 as a result kidneys from Black donors are assigned a higher KDRI and KDPI value and are disproportionately more likely to be labeled as “high-risk” for graft failure than kidneys from non-Black donors with similar characteristics.

Rationale for including race in the KDRI calculator.

Prior studies have reported that kidneys from Black deceased donors are associated with increased risk of graft failure compared to kidneys from non-Black donors.³ Selection of donor race in KDRI calculation was based on statistical association with graft failure and the biological basis for this inclusion remains ambiguous. Prior to KAS, kidneys from Black donors were more likely to be allocated to Black recipients due to shared blood type and human leucocyte antigen (HLA).⁴ The original KDRI calculator only considered donor and recipient race and did not account for differences in social determinants of health between Black and non-Black donors and recipients that could confound the association of Black donor race with outcomes. It is also possible that the Black donor race likely captures undiagnosed and untreated medical conditions resulting in subclinical kidney disease from lack of access to health-care, epigenetic changes from environment and lifestyle, and genotype - prominently Apolipoprotein L1 (APOL1).^5,6

Studies comparing race-based and race-free KDRI equations

To date, there have been three studies that have examined the impact of dropping Black donor race variable from the KDRI calculation on risk prediction and discrimination of allograft failure, as well as kidney non-use.^7–9 All three studies used SRTR data and donor-only version of KDRI equation but had significant methodological differences in their calculations. Table 1 shows the differences in methodology and results. Below are the main take aways from the three studies.

Table 1.

Comparing methods and results of three studies evaluating the impact of dropping race from KDRI equation

	Doshi et al, 2022	Chong et al, 2021	Miller et al, 2023
Source of Data	SRTR	SRTR	SRTR
Method	Zeroed the coefficient for Black donor race	Refitted KDRI model with and without race	Refitted KDRI model with and without race
	Used OPTN KDPI mapping tables from 2010	Used OPTN KDPI mapping tables from prior years	Recalculated KDPI mapping tables based on new KDRI values
Change in risk discrimination	C-statistic* decreased by 0.16%	Difference in mean AUC of 0.002	C-statistic decreased by 0.16%
Change in risk prediction by IBS**	Increased by 0.07%		Increased by 0.08%
Reclassification of black donors from KDPI >85% to ≤85%	60%	73%	57%
Change in kidney non-use for Black deceased donors	70 additional kidneys used per year		353 additional kidneys used in 2015–2021 cohort

Please note that methods are different and therefore absolute values cannot be compared.

^*A higher C statistic represents better risk discrimination.

^**IBS, Integrated Brier Score. A lower IBS represents better prediction accuracy.

Reclassification of high-risk (KDPI > 85%) kidneys from Black donors to lower-risk categories.

All studies reported that the use of race-free KDRI calculator decreases KDRI and KDPI values for kidneys procured from Black deceased donors. As anticipated, while there will be no changes in raw KDRI for non-Black donors, there will be an increase in KDPI values for some non-Black donors. Since KDPI is on a percentile scale, decrease in proportion of Black donor kidneys being labeled as high-risk (KDPI >85%) will result in corresponding increase in proportion of non-Black donor kidneys as being labeled as high-risk. Around 45% of kidneys with KDPI >85% are not utilized.² Removing donor race from KDRI calculator will reduce KDPI values of kidneys from Black deceased donors and lower the likelihood of non-use.

No impact on risk prediction and discrimination.

All three studies report that removal of Black donor race from the KDRI equation was associated with minimal change in the risk discrimination or prediction accuracy. This may be due to the nature of the calculator which only looks at the donor factors and not recipient factors and care after transplant which are also strong predictors of graft outcomes.

Possible increase transplant rates among Black candidates.

Miller et al. reported that removing race from KDRI equation resulted in fewer donors in southeastern United States to be labeled as high-risk (or KDPI >85%) and predicted lower rates of non-use in this region.⁷ The proportion of Black candidates with end stage kidney disease in this region is highest in the country (26%) and therefore may lead to increase transplants among Black candidates in this region.¹⁰ Since KAS, most Black recipients (79.5% from 2015 through 2021) received kidneys from non-Black donors. So, the impact of removing race from KDPI equation on transplant rates in Black candidates may be marginal.

Possible positive effect on organ donation among Black individuals.

Black individuals are disproportionately affected by chronic kidney disease, hence there are fewer potential Black kidney donors. Several studies have demonstrated that in addition to socio-economic factors, there is hesitancy to donate due distrust in the medical community.¹¹ Tran et.al. conducted 33 semi-structured qualitative interviews to understand challenges and opportunities in promoting blood donation among Black communities and reported perceived wastefulness as one of the barriers.¹² Information on removal of race from the KDRI equation may help build trust by reassuring the community that their donated organs are not labeled as high-risk because of their race.

Replacing race with ancestry genetic markers

Variants in APOL1 gene arose thousands of years ago among residents of sub-Saharan Africa to protect them from Trypanosome infection. A single copy of the APOL1 risk variant G1 or G2 provides protection from African sleeping sickness; however, 2 risk variants (G1G1, G2G2, or G1G2), also referred to as high-risk genotype, markedly increases the individual’s risk for kidney disease.⁶ The APOL1 risk variants are common in people with African ancestry but absent in those with European ancestry, and are found in different frequencies across the world based on migratory patterns.

The presence of 2 APOL1 risk variants in Black deceased donors is associated with about two-fold increased risk of graft loss when compared to kidneys from donors with 0/1 APOL1 kidney risk variants.¹³ Majority of the failed allografts show features of FSGS. Kidneys transplanted from Black donors carrying 0/1 APOL1 risk variant have outcomes similar to those from non-Black deceased donors.¹³ Therefore, labeling all kidneys from Black deceased donors as high-risk is not warranted. Eliminating the Black donor race variable from KDRI equation would provide better estimate for the remaining 85% with 0/1 APOL1 kidney risk variants.¹⁴

However, the association between donor APOL1 genotype and kidney transplant outcome is not straight forward. Not all transplanted kidneys carrying 2 APOL1 kidney risk variants fail prematurely, and majority (over 70%) have good long-term outcomes. Hence, similar to native kidneys, a second hit is also required for transplant allografts carrying high-risk genotype to have shortened survival. It is currently unclear if recipient APOL1 genotype also influences graft outcomes.^15,16

The ongoing APOL1 Long-Term Kidney Transplantation Outcomes Network (APOLLO) study findings will help answer some of these issues. The APOLLO study has been extended to Phase 2 to obtain follow-up data on recipient outcomes, particularly albuminuria and therefore the study findings will not be available in near future.

Summary and call for action

Current data suggests that removing the Black donor race variable from KDRI equation would decrease KDPI score of Black donor kidney and increase use of kidneys from Black donors without impacting risk prediction and discrimination of kidney allograft failure. It will instead bring racial parity to the “high-risk” labeling effect of kidneys based on KDPI values i.e., >85%. APOL1-based KDRI models may improve its precision, but the results of the APOLLO study will not be available in near future. Gill et. al. made a strong argument for adopting race-free KDRI immediately.¹⁷ Several national organizations including American Society of Nephrology, National Kidney Foundation and the National Academies of Sciences, Engineering, and Medicine have recommended removing race from the KDRI calculation.^18,19 SRTR held a consensus conference inviting all stakeholders (including transplant professionals, regulators, patients and deceased donor family members) to discuss transplant metrics, and the attendees supported a race-free KDRI calculation.²⁰ OPTN minority affairs committee is currently assessing the impact of removal of race from KDRI calculation and is likely to follow the due process which may take months to get final approval. Given the potential benefits of race-free KDRI and lack of foreseeable negative consequences, we call for an immediate action from OPTN to remove from KDRI calculation instead of adhering to a standard lengthy review process.

Abbreviations Page

APOL1

Apolipoprotein L-1

ECD

expanded criteria donor

FSGS

focal segmental glomerulosclerosis

HLA

Human Leukocyte Antigen

KAS

Kidney Allocation System

KDPI

Kidney Donor Profile Index

KDRI

Kidney Donor Risk Index

OPTN

Organ Procurement Transplant Network

SRTR

Scientific Registry of Transplant Recipients