The Kidney Allocation System Does Not Appropriately Stratify Risk of Pediatric Donor Kidneys: Implications for Pediatric Recipients

Abstract

Kidney Allocation System (KAS) was enacted in 2014 to improve graft utility, while facilitating transplantation of highly-sensitized patients and preserving pediatric access to high-quality kidneys. Central to this system is the Kidney Donor Profile Index (KDPI), a metric intended to predict transplant outcomes based on donor characteristics but derived using only adult donors. We posited that KAS had inadvertently altered the profile and quantity of kidneys made available to pediatric recipients. This question arose from our observation that most pediatric donors carry a KDPI over 35 and have therefore been rendered relatively inaccessible to pediatric recipients under KAS. Here we explore early trends in pediatric transplantation following KAS, including 1) use of pediatric donors, 2) use of Public Health System (PHS) high infectious risk donors, 3) wait time, and 4) living donor transplantation. We note some concerning preliminary changes following KAS implementation, including the allocation of fewer deceased donor pediatric kidneys to children and stagnation in pediatric wait times. Moreover, the poor predictive power of the KDPI for adult donors appears to be even worse when applied to pediatric donors. These early trends warrant further observation and consideration of changes in pediatric kidney allocation if they persist.

Introduction

Kidney Allocation System (KAS) was implemented by the Organ Procurement and Transplantation Network (OPTN) on December 4, 2014 after a comprehensive 12-year effort(1). One of the stated goals of this significant undertaking was to improve the utility of organs while reducing differences in transplant access for specific populations, including children(2). In the spirit of maintaining pediatric priority, pediatric allocation was altered from “Share 35” (pediatric candidates are given priority for kidneys from donors less than 35 years of age) to preferential allocation of kidneys of the highest quality as determined by Kidney Donor Profile Index (KDPI) value < 35. The KPDI is central to the KAS: an estimate of graft utility derived from a comprehensive analysis of risk factors for graft loss, and transformed from the normative Kidney Donor Risk Index (KDRI). Although the KDRI is essential in the allocation of all kidneys under KAS, it was initially developed and validated in donors over 18 years of age(3).

We wished to explore potential unanticipated effects of KAS on pediatric transplantation, and thus posed five questions:

1. Is KDPI a good predictor of graft survival for pediatric donor kidneys transplanted into pediatric or adult recipients?

2. Are children receiving a similar proportion of pediatric donor kidneys since KAS adoption compared to before?

3. Are children receiving more Public Health System (PHS) high infectious risk kidneys after KAS, in light of the high prevalence of opioid use among adult donors with KDPI < 35?(4, 5)

4. Are children waiting longer for transplants after KAS?

5. What are trends in pediatric living donor transplantation now compared to pre-KAS in light of potentially altered access to deceased donor transplantation?

Materials and Methods

Data including all deceased donor, first-time kidney transplants between January 1, 2000 and March 31, 2016 were obtained from the Organ Procurement and Transplantation Network (OPTN). Patients receiving other organs simultaneously were excluded. We limited our analysis to first-time transplants as patients undergoing re-transplantation might carry other risk factors for graft failure such as sensitization, recurrent disease, comorbidities or questionable compliance. Given the unique anatomic challenges associated with the transplantation of very small and en bloc pediatric kidneys as well as the lack of uniformity in the decision to split such kidneys(6, 7), we chose to limit our analyses to single kidneys from donors age 3–17 (en blocs removed). For survival analyses, the study population was limited to those transplanted between January 1, 2000 and December 31, 2009, allowing for a minimum of 6.25 years follow-up to March 31, 2016. For all donors, the KDRI Rao value was calculated according to the equation used by the OPTN(8), adjusted to the 2015 scaling factor and mapped to KDPI(9).

We investigated the association between KDPI and graft survival (Question 1) using the Kaplan-Meier method as well as by multivariate logistic regression for using one-year graft failure as a binary outcome for pediatric recipients with a minimum of one year of follow-up. To consider the impact of KAS on the proportion of pediatric kidneys being transplanted into pediatric recipients (Question 2), we compared the year preceding KAS (12/4/2013 – 12/3/2014) to the year following its implementation (12/4/2014 – 12/3/2015), using chi-squared tests except when use of Fisher’s exact test was mandated by small sample sizes. Since pediatric priority originates from time of listing, we defined “pediatric recipient” as those listed before age 18. For Question 3 regarding high risk donors in the year pre- and post-KAS, we classified organs as PHS high risk as identified by UNOS(10) in the UNOS Standard Transplant Analysis and Research (STAR) file. We analyzed changes in median days on the waiting list, median days on dialysis (Question 4) and frequency of living donation (Question 5) using the appropriate STAR file variables across the time frames defined above. For these analyses, we included pre-transplant candidates continuing to accrue wait time. Medians were compared using the Wilcoxon-Mann-Whitney test.

Stata SE 14 was used for all analyses (StataCorp LP, College Station, TX). The University of Pennsylvania Institutional Review Board (IRB) deemed the research IRB review exempt as authorized by 45 CFR 46.101, category 2.

Results

Between our study period dates of January 1, 2000 and March 31, 2016, 136,605 first-time, kidney-alone, deceased donor transplants occurred in the U.S. The cohort used for graft survival analyses (minimum 6.25 years follow-up) included 78,882 transplants. Figure 1 demonstrates the KDPIs of all donors, grouped by age. As seen, the KDPI reached its nadir at donor age 18 and trended upwards towards 100 for the very young and very old. Below age 6, no donor had a KDPI less than 35 and thus would not be prioritized for a pediatric recipient.

Figure 1.

KDPI by Donor Age for Transplants from January 1, 2000 to March 31, 2016. KDPI, Kidney Donor Profile Index.

1) Is KDPI a good predictor of graft survival for pediatric donor kidneys transplanted into pediatric or adult recipients?

As shown in Figure 2, overall unadjusted graft survival was not different in recipients receiving donor kidneys age 3–17 with KDPI < 35 versus ≥ 35 (p = 0.94). A sensitivity analysis stratifying by recipient age (0–10 and 11–17) again showed no difference in graft outcome among these two groups. Among adult recipients of pediatric kidneys, graft survival was statistically inferior for KDPI ≥ 35 compared to < 35, but likely clinically insignificant: 6.2% and 4.5% difference in graft survival at 10 years for recipients aged 18–40 and > 40, p = 0.01, p < 0.001 respectively.

Figure 2.

Graft Survival for Recipients of Pediatric Kidneys (Donors Age 3–17, En Blocs Removed), KPDI < 35 versus ≥ 35. KDPI, Kidney Donor Profile Index.

Multivariate logistic regression accounting for adult versus pediatric donor, kidney laterality and PHS status demonstrated that KDPI < 35 versus ≥ 35 was a poor predictor of one-year graft outcomes for pediatric recipients (graft failure odds ratio of 1.28 for KDPI ≥ 35, P = 0.158). A separate analysis with the same covariates but using KDPI as a continuous variable rendered an odds ratio of 1.01 for KPDI (95% confidence interval [1.001, 1.018], P = 0.035), meaning that pediatric recipients on average incur a 1% increase risk of 1-year graft failure for each point KDPI increase when controlling for adult versus pediatric donor, laterality and PHS status. These covariates were non-significant in the analysis (P = 0.888, 0.287 and 0.683, respectively).

2) Are children receiving a similar proportion of pediatric donor kidneys since KAS adoption compared to before?

Table 1 demonstrates the percentage of age-stratified pediatric donors transplanted into pediatric recipients. These data show an overall decrease in pediatric donor utilization among pediatric recipients following KAS implementation, from 13.4% to 10.1%, p = 0.023. Most notable is the change in the utilization of donors age 6 to 11 years for pediatric recipients, which dropped from 14.0% in the year prior to KAS, to 7.8% in the year following, although this did not reach strict statistical significance (p = 0.056).

Table 1.

Donor Age	Pediatric Donors Transplanted into Pediatric Recipients
	Pre KAS	Post KAS	p value
0 – 2	2.2%	0.0%	0.045
3 – 5	2.0%	2.4%	1.00
6 – 11	14.0%	7.8%	0.056
12 – 17	21.3%	18.0%	0.225
Total (0–17)	13.4%	10.1%	0.023

3) Are children receiving more PHS high risk kidneys after KAS?

Figure 3 demonstrates the percentage of Public Health Service (PHS) high risk donors by KDPI since the implementation of KAS. Those kidneys prioritized for children by the KAS (KPDI < 35) carry a high proportion of PHS high-risk donors. In the 12 months following implementation of KAS (12/4/2014 – 12/3/2015) compared to the preceding 12 months (12/4/2013 – 12/3/2014), the proportion of children < 18 years receiving PHS high-risk kidneys increased to 13.7% (66 of 483) from 11.0% (53 of 482), although this did not achieve statistical significance (p = 0.06). Of note, the percentage of adults receiving PHS high risk kidneys also increased over this period, from 18.4% (1663 of 9039) in the year prior to KAS to 21.6% (1953 of 9041) in the year following (p < 0.001).

Figure 3.

Percentage of PHS High Risk Donors by KDPI and Donor Age After KAS (December 4, 2014 to March 31, 2016). KAS, Kidney Allocation System; KDPI, Kidney Donor Profile Index; PHS, Public Health Service.

4) Are children waiting longer for transplants after KAS?

There was no statistically significant change in wait time for children receiving deceased donors when comparing the year prior to the year following KAS implementation. In the year following KAS, pediatric recipients waited a median of 255 days, whereas the median wait time was 236 days in the year preceding KAS (p = 0.167). However, adult candidates experienced a drop in their median wait time over the same timeframe, from 991 to 886 days, p < 0.0001. With respect to dialysis time, both adult and pediatric candidates have seen their median dialysis exposure increase. Children’s median dialysis time increased from 601 to 709 days, p = 0.006, whereas adults’ dialysis time increased from 1660.5 to 1799 days, p < 0.0001. This is not unexpected, as KAS specifies that wait time should date to the initiation of dialysis.

5) What are trends in pediatric living donor transplantation now compared to pre-KAS?

The rate of pediatric living donor transplantation increased in the year following KAS implementation compared to the year prior, and adult transplantation saw the same trend. 243 living donor kidneys went to children in the year after KAS (33.5% of all transplants) compared to 214 living donor transplant in the year prior (30.7% of all transplants), p = 0.02. Likewise, adult recipients received 4697 (34.2% of all transplants) living donor kidneys in the year following KAS, up from 4438 (32.9% of all transplants) in the year preceding, p = 0.028.

Discussion

Kidney Allocation System is the result of a comprehensive effort across more than a decade, with the goal of equitably matching donor and recipient expected outcomes and improving access to transplantation among distinct groups including highly sensitized recipients, those with blood group B and children. Here we explore the early predictive ability of the KPDI for pediatric kidneys as well as changes in the pediatric transplant landscape since KAS.

In contrast to the competing Life Years From Transplant (LYFT) proposal (11), KAS sacrifices some longevity gains to maintain transplant access for older recipients(2). However, the implementation of KAS was not intended to disadvantage children and specifically sought to maintain pediatric priority for what were felt to be “better” quality organs. Children had previously received priority for donors less than 35 years as young donor age remains a positive predictor of successful graft outcomes(12). Under KAS, KDPI < 35% replaced age as a proxy for quality, but the application of KDPI failed to acknowledge its lack of discriminant validity in characterizing pediatric donors. Specifically, the significant weighting of height and weight in the KDPI formulation, based on the underlying assumption that lower height and weight are associated with reduced renal mass and therefore worse graft outcomes(13), assure that almost every pediatric donor under age 10 will have a KDPI over 35%.

Our exploratory analysis found the outcomes for pediatric donors age 3–17 with KDPI ≥ 35 unchanged from those with KDPI < 35 when transplanted into children (p = 0.94). Multivariate analysis of 1-year graft survival for pediatric recipients found the KDPI 35 dichotomy to be nonpredictive (odds ratio 1.28 for graft failure if KDPI ≥ 35, P = 0.158). In addition, the donors most easily accessible to pediatric recipients with low KDPI under KAS carry the highest PHS high risk prevalence among all donors, likely due to the epidemic of opioid abuse(5) and general increased engagement in risky behavior in young adults(14, 15). Rates of PHS high risk transplantation did increase for both adult and pediatric recipients after KAS, although this trend did not achieve statistical significance for children (p = 0.06). Many pediatric programs attempt to avoid PHS high risk donors, thus it is possible that a significant proportion of PHS high risk offers for children are declined categorically, perhaps limiting significance in the analysis and also possibly lengthening wait times for children.

Children have not benefitted from the shortened wait times after KAS implementation seen for adults, at least in this early analysis. However, there are admittedly numerous other factors at play in wait time shifts following KAS, including the new prioritized national and regional donor access for recipients with high PRAs. Further observation of wait time trends is warranted as many of the long-waiting adults with high PRAs are receiving organs. As with adults, rates of living donor transplantation increased following KAS. It is unknown whether the perception of longer pediatric wait time under KAS is driving this living donation trend, and what the trajectory of pediatric living donor transplantation will be in the future.

We acknowledge that these observed changes in PHS status, wait times, dialysis times and living donation are impacted by many other factors, and the causality of KAS pediatric allocation policy in these early trends is unclear. Additionally, our analyses are limited by the projection of pre-KAS outcomes into the post-KAS era. We acknowledge as well that as with any retrospective analysis, our evaluation has the inherent limitations of large national datasets and the interpretation of prior actions to predict future events.

It cannot be denied, however, that the centrality of KDPI in KAS has restricted the donor cohort available to children. This represents the flawed application of an adult donor metric to pediatric donors, for whom the KDPI was neither derived nor tested. As measures of donor and recipient quality become more sophisticated in the future, it is incumbent upon the transplant community to apply these models only to the population for which they were formulated and validated. Furthermore, KAS has limited the ability of the transplant surgeon to make customized choices for the recipients he or she knows well. There may be situations where a smaller sized kidney is desirable for a pediatric recipient(16) yet would be nearly unobtainable under KAS. KAS has functionally limited the donor pool for children without demonstrable benefit to their outcomes.

While an elegant solution might be to adopt a KDPI validated for pediatric donors as suggested by Parker et al., (17) such implementation through UNOS is likely to be logistically complex. Until such time, we suggest that pediatric donors be offered first to pediatric recipients, without KDPI restriction and at the typical allocation sequence for pediatric priority, allowing pediatric centers to appropriately risk stratify the possible advantages or disadvantages of pediatric donor kidneys for their own patients. Following this, allocation should proceed through the current algorithm. Any future policy must include an accurate assessment of the risk of pediatric kidneys across recipient populations, thus ensuring equitable allocation of this donor resource.

Abbreviations

IRB

Institutional Review Board

KAS

Kidney Allocation System

KDPI

Kidney Donor Profile Index

KDRI

Kidney Donor Risk Index

LYFT

Life Years From Transplant

OPTN

Organ Procurement and Transplantation Network

PHS

Public Health Service

STAR

Standard Transplant Analysis and Research