Introduction

Transplantation is associated with higher survival and quality of life compared with dialysis.1–3 However, the transplant rate in the United States is only 3.6 per 100 patient-years.4 The relative shortage of donor kidneys is the main factor limiting transplantation, and there are nearly 100,000 patients waitlisted for a transplant in the United States.5 Deceased donor kidneys with risk factors that portend a higher risk of allograft loss but still provide a mortality benefit compared with dialysis are frequently recovered but not transplanted because of concerns regarding allograft survival. For example, over 60% of kidneys recovered from donors aged 65 years and older are discarded.6

The Kidney Donor Profile Index (KDPI) is a summary metric based on deceased donor factors to estimate the risk of allograft failure using a scale from 0% (lowest risk) to 100% (highest risk), compared with transplant outcomes in the previous year.7 Kidneys with KDPI between 21% and 85% are expected to function for approximately 9 years, whereas kidneys with high KDPI ≥85% are expected to function for 5.5 years.8 Therefore, the use of high-KDPI kidneys is best justified when more rapid transplantation is enabled.9–11

KDPI is used to maximize the utility of the available organ supply by ensuring that kidneys with the longest projected survival (KDPI 0%–20%) are allocated to patients with the longest life expectancy.12 However, the allocation system does not prioritize the allocation of high-KDPI kidneys to patients who are at higher risk of death on dialysis who may benefit from more rapid transplantation.

Transplantation before the need for dialysis, also known as preemptive transplantation, is associated with longer allograft and patient survival and lower health care costs compared with transplantation after initiation of dialysis (nonpreemptive transplantation).13–18 However, there has been limited research examining the use of preemptive transplantation with high-KDPI kidneys, and it is uncertain whether the benefits of preemptive transplantation also extend to these kidneys.

We hypothesized that first, the risk of allograft failure in preemptive transplants using high-KDPI kidneys ≥85% would be lower than nonpreemptive transplants with the same KDPI, and second, the risk of allograft failure in these transplants would be comparable with that in nonpreemptive transplants using lower-KDPI kidneys.

Methods Data Source and Study Population

Data from the 2021 Scientific Registry of Transplant Recipients (SRTR) were used for this study.5 The SRTR includes data on all donors, waitlisted candidates, and transplant recipients in the United States, submitted by the members of the Organ Procurement and Transplantation Network (OPTN). The Health Resources and Services Administration, US Department of Health and Human Services, provides oversight to the activities of the OPTN and SRTR contractors.

This study included patients aged 18 years and older who underwent a kidney-only transplant between January 1, 2005, and December 31, 2017 (Figure 1). This cohort was chosen to permit at least 3 years of follow-up until December 31, 2020, in the 2021 SRTR database. Patients younger than 18 years were excluded as these patients receive priority for transplantation. Patients with a previous nonkidney transplant, those waitlisted for another organ transplant, and those for whom the KDPI could not be calculated owing to missing data were also excluded. The final study cohort included 12,331 preemptive recipients and 107,760 nonpreemptive recipients.

Figure 1:

Study cohort. KDPI, Kidney Donor Profile Index; SRTR, Scientific Registry of Transplant Recipients.

The study cohort was grouped according to KDPI that was standardized to the 2018 mapping values and categorized into groups of 0%–20%, 21%–50%, 51%–84%, and 85%–100%.7 These groupings were chosen to correspond to the current organ allocation policy that allocates kidneys with KDPI 0%–20% to waitlist candidates within the top quintile of expected post-transplant survival and the previous designation of expanded criteria deceased donor kidneys that corresponded to a KDPI ≥85%.12,19 The characteristics of the study cohort are described using medians with interquartile ranges for continuous variables or proportions for categorical variables. Statistical differences between the KDPI groups were evaluated using the chi-square test or ANOVA, as appropriate.

Association of Preemptive Transplantation with Allograft and Patient Survival

The unadjusted time to allograft loss from all causes, including death, death-censored graft loss, and death with a functioning graft among preemptive and nonpreemptive recipients was determined using the Kaplan–Meier method, and group differences were compared using the log-rank test.

Separate multivariable Cox proportional hazards analyses were performed to determine the association between preemptive transplantation and transplant outcomes. Model covariables included established risk factors of allograft failure, including recipient age at the time of transplantation, sex, race, body mass index, diabetes as the cause of kidney failure, history of vascular disease, pretransplant anti-HLA antibodies characterized by the pretransplant calculated panel reactive antibody, donor and recipient HLA mismatch at the HLA A and B, DR gene loci, insurance provider, recipient education, type of induction therapy, and transplant year. Because missing data for study variables were <1%, we created a category of “missing” to permit the inclusion of all study patients in the multivariable models. The proportionality assumption was assessed using Schoenfeld residual plots.

Outcomes of Transplants Characterized by KDPI and Treatment with Dialysis

The abovementioned multivariable models were repeated with transplant recipients characterized by donor KDPI grouping and prior treatment with dialysis with a reference group of nonpreemptive transplants using kidneys with KDPI 0%–20%. This reference group was chosen because of superior outcomes compared with other KDPI categories and because nonpreemptive transplantation is most common.

Mate Kidney Analysis

A mate kidney subcohort was identified in which one kidney was transplanted preemptively while the second was transplanted nonpreemptively. In this way, donor characteristics between the kidneys would be perfectly balanced between preemptive and nonpreemptive recipients. The characteristics of this mate kidney cohort are described above. The time to all-cause and death-censored graft loss and death with function was determined using the Kaplan–Meier method, stratified by KDPI among preemptive and nonpreemptive recipients, and group differences were assessed with the log-rank test. Separate multivariable Cox proportional hazard models in groups defined by KDPI were used to determine the association of preemptive transplantation with transplant outcomes with adjustment for differences in recipient factors only, as otherwise described above.

Secular Trends in Preemptive Use of High-KDPI Kidneys for Transplantation

The annual unadjusted incidence of preemptive transplantation as a proportion of the total number of deceased donor transplants was calculated, stratified by KDPI. Similarly, the total annual number of kidneys recovered but not transplanted was calculated and stratified by donor KDPI <85 or ≥85.

Pretransplant Kidney Function and Delayed Graft Function

We sought to understand the mechanisms by which preemptive transplantation might affect outcomes in high-KDPI transplants. We hypothesized that a higher level of pretransplant native kidney function may contribute to improved transplant outcomes, including a lower risk of delayed graft function (DGF), as defined as the need for dialysis in the first post-transplant week. Because native kidney function at transplantation is not captured in the SRTR, we used the eGFR at the time of waitlisting as a surrogate of pretransplant native kidney function.

Preemptive recipients were stratified by eGFR at the time of waitlisting (≤10 ml/min, >10 ml/min, or missing). The unadjusted time to all-cause graft loss was estimated using the Kaplan–Meier method among preemptive recipients, grouped by eGFR at the time of listing, and nonpreemptive recipients. An unadjusted logistic regression was used to compare the odds of DGF in those who received a kidney from a donor with KDPI ≥85% before and after treatment with dialysis. Finally, the Kaplan–Meier method was used to determine the time to all-cause and death-censored graft loss in preemptive and nonpreemptive recipients, stratified by the occurrence of DGF.

Sensitivity Analyses

We repeated our analyses with varying cohort definitions. First, we included patients with less than 6 months of dialysis treatment in the preemptive cohort. Second, we stratified the preemptive cohort into those who received a kidney transplant within 1 year or after 1 year of waitlisting. Finally, we stratified nonpreemptive recipients by dialysis duration of less than or more than 2 years.

All analyses were performed using STATA, Version 17.120 A P value of <0.05 was used to determine statistical significance. This study was approved by the University of British Columbia Research Ethics Board (H18-01381). The clinical and research activities being reported are consistent with the principles of the Declaration of Istanbul as outlined in the Declaration of Istanbul on Organ Trafficking and Transplant Tourism.

Results Study Cohort

The assembly of the study cohort is shown in Figure 1. Among the 12,387; 37,529; 46,964; and 23,211 patients who underwent transplantation with deceased donor kidneys with KDPI 0%–20%, 21%–50%, 51%–84%, and ≥85% during the study period, 12%, 12%, 9%, and 9% were preemptive recipients, respectively. Preemptive recipients were more likely to be White, female, older, and privately insured and have greater than high school education, but were less likely to have diabetes-related kidney failure (Table 1). Among transplants from a donor with KDPI ≥85%, the median KDPI was 92% in both preemptive and nonpreemptive recipients (Supplemental Figure 1).

Table 1 - Characteristics of preemptive and nonpreemptive transplant recipients of a kidney-only transplant in the United States between January 1, 2005, and December 31, 2017 Characteristic Preemptive
Transplants (N=12,331) Nonpreemptive
Transplants (N=107,760) % N % N Median age at the time of transplant (Q1, Q3) 58 (49–65) — 55 (45–63) — Sex, male (%) 54 5718 62 41,488 Race (%)  White 67 8213 39 41,491  Black 18 2270 35 27,954  Other 15 1848 26 1848 Median duration of pretransplant dialysis (Q1, Q3) N/A — 4 (2–6) — Diabetes as a cause of kidney failure 20 2407 30 32,560 History of vascular disease (%) 2 216 3 2791 History of angina (%) 6 750 7 8015 BMI, kg/m 2 (%)  <18 2 199 1 1510  18–25 28 3409 27 29,508  26–40 65 7979 66 70,985  41+ 6 744 5 5757 Insurance provider (%)  Private 54 6625 22 23,573  Medicare 45 5539 77 82,905  Veterans Association 1 145 1 1190  Other 0 22 0 92 KDPI (%)  0–20 13 1542 10 10,845  21–50 35 4331 31 33,198  51–84 36 4457 39 42,507  85+ 16 2001 20 21,210 eGFR at listing (%)  ≤10 and preemptive 7 844  >10 and preemptive 65 7948  Missing eGFR and preemptive 29 3539  Nonpreemptive 0 100 120,091 Year of transplantation (%)  2005–2008 29 3519 28 30,482  2009–2013 36 4480 37 39,790  2014–2017 35 4332 35 37,488 Median HLA mismatch (Q1, Q3) 4 (3–5) — 4 (4–5) — Cold ischemic time (h)  0–12 34 4180 29 31,221  13–24 44 5485 50 53,357  25+ 18 2211 19 20,464  Missing 4 455 3 2718 Induction therapy (%)  Depleting antibody therapy 84 10,385 88 94,695  Non–depleting antibody therapy 14 1681 10 11,247  No induction 2 265 2 1818 Highest level of education (%)  High school or less 40 4897 51 54,906  Bachelor degree 41 5101 36 39,322  Graduate degree 10 1237 5 5678  Unknown 9 1096 7 7854 Peak cPRA (%)  0–20 79 9784 78 84,453  21–80 12 1456 13 14,394  81–90 4 443 3 3349  91+ 5 626 5 5527

BMI, body mass index; KDPI, Kidney Donor Profile Index; cPRA, calculated panel reactive antibody.

Association of Preemptive Transplantation with Transplant Outcomes

The time to allograft loss from all causes, including death, was shorter among nonpreemptive transplant recipients in all KDPI groups (Figure 2), as was the time to death-censored graft loss and time to death with a functioning transplant (Supplemental Figures 2 and 3). Specifically, the probability of survival without graft loss from any cause at 3 years after transplant was 89.1% (95% confidence interval [CI], 88.5% to 89.6%) in preemptive recipients compared with 84.6% (95% CI, 84.4% to 84.8%) in nonpreemptive recipients. In multivariable analyses, preemptive transplantation was associated with a lower risk of all-cause graft loss (hazard ratio [HR], 0.77; 95% CI, 0.75 to 0.80), a lower risk of death-censored graft loss (HR, 0.75; 95% CI, 0.71 to 0.79), and a lower risk of death with function (HR, 0.85; 95% CI, 0.81 to 0.89) compared with nonpreemptive transplantation. The full model results are provided in Supplemental Table 1.

Figure 2:

Impact of preemptive transplantation on graft loss from all causes, including death, stratified by donor KDPI. Kaplan–Meier curves of graft loss from all causes, including death, in recipients receiving kidneys with KDPI 0%–20% (A), KDPI 21%–50% (B), KDPI 51%–84% (C), and KDPI ≥85% (D) after deceased donor transplantation, stratified by preemptive transplantation (solid line) and nonpreemptive transplantation (dashed line). The P values represent the statistical significance of the log-rank test for each analysis.

In a multivariable model (Table 2), transplants using higher-KDPI donor kidneys had a higher risk of all-cause graft loss. Within each KDPI group, the risk of graft loss was higher among nonpreemptive compared with preemptive recipients. Compared with the reference group of nonpreemptive transplants from donors with KDPI 0%–20%, the risk of all-cause graft loss in preemptive transplants using kidneys with a KDPI ≥85% (HR, 1.51; 95% CI, 1.39 to 1.64) was lower than that in transplants using donors with the same KDPI performed nonpreemptively (HR, 2.39; 95% CI, 2.21 to 2.58), but was similar to that among nonpreemptive transplants using kidneys with KDPI 51%–84% (HR, 1.61; 95% CI, 1.52 to 1.70).

Table 2 - Multivariable analysis of risk of graft loss from any cause in recipients of preemptive and nonpreemptive transplant by the donor Kidney Donor Profile Index Category of Kidney Transplant Donor KDPI % 0–20 21–50 51–84 ≥85 HR (95% CI) N (%) HR (95% CI) N (%) HR (95% CI) N (%) HR (95% CI) N (%) Nonpreemptive transplant Reference 3581 (33.0) 1.13 (1.08 to 1.18) 10,458 (31.5) 1.61 (1.52 to 1.70) 16,471 (38.8) 2.39 (2.21 to 2.58) 10,015 (47.2) Preemptive transplant 0.77 (0.75 to 0.80) 379 (24.6) 0.85 (0.80 to 0.89) 1050 (24.2) 1.11 (1.04 to 1.19) 1399 (31.4) 1.51 (1.39 to 1.64) 799 (39.9)

The reference group in this analysis is recipients of a nonpreemptive transplant from a donor with a Kidney Donor Profile Index (KDPI) of 0%–20%. The number and percentage of individuals with events in each category is presented. HR, hazard ratio; CI, confidence interval.

Model covariates included risk factors of allograft failure, including recipient age at the time of transplantation, sex, race, body mass index, diabetes as the cause of kidney failure, history of vascular disease, pretransplant anti-human leukocyte antigen antibodies characterized by the pretransplant calculated panel reactive antibody, donor and recipient HLA mismatch at the HLA A and B, DR gene loci, insurance provider, recipient education, type of induction therapy, and transplant year.

Mate Kidney Analysis

A total of 7232 donors had one kidney transplanted preemptively and the second transplanted nonpreemptively. As observed in the primary cohort, patients who received a preemptive transplant were more likely to be older, White, and female and have more than high school education and private insurance (Supplemental Table 2). The time to all-cause graft loss, death-censored graft loss, and death with a functioning graft (Figure 3, Supplemental Figures 4 and 5) was consistently shorter in nonpreemptive compared with preemptive recipients in all KDPI groups. In separate multivariable analyses within KDPI groups that adjusted for differences in recipient characteristics between mate kidney transplants, preemptive transplantation was consistently associated with a lower risk of all-cause graft loss in all KDPI groups. The full multivariable model results are provided in Supplemental Table 3.

Figure 3:

Impact of preemptive transplantation in mate kidney cohort on graft loss from all causes, stratified by KDPI. Kaplan–Meier curves of graft loss from all causes, including death, by KDPI group: KDPI 0%–20% (A), KDPI 21%–50% (B), KDPI 51%–84% (C), and KDPI ≥85% (D) among mate kidney transplants in which one kidney was transplanted preemptively (solid line) and another kidney transplanted after treatment with maintenance dialysis (dashed line). The P values represent the statistical significance of the log-rank test (A–D) for each analysis.

Secular Trends in Preemptive Transplantation and Utilization of Deceased Donor Kidneys by KDPI

The incidence of preemptive transplantation varied between 9.4% and 11.4% of all kidney transplants during the study period (Figure 4). Kidneys from donors with a KDPI ≥85% were less likely to be transplanted preemptively relative to kidneys from donors with KDPI 0%–84% (Figure 4). The overall number of discarded kidneys was higher in more recent years (Supplemental Figure 6A), and the proportion of discarded kidneys was highest among kidneys with KDPI ≥85% and was ≥50% throughout the study period (Supplemental Figure 6B).

Figure 4:

Unadjusted incidence rate of preemptive transplantation between 2005 and 2017, overall and stratified by donor KDPI (0–84%, ≥85%). The vertical dashed line represents the implementation of the new kidney allocation system in 2014.

Native Kidney Function and Delayed Graft Function

Among KDPI ≥85% preemptive recipients, those waitlisted with eGFR >10 ml/min had a longer time to all-cause and death-censored graft loss compared with both preemptive recipients waitlisted with eGFR ≤10 ml/min and those who received a transplant nonpreemptively (Supplemental Figure 7). The odds of DGF was lower in preemptive recipients waitlisted with eGFR >10 ml/min compared with preemptive recipients waitlisted with eGFR ≤10 ml/min and nonpreemptive recipients (Table 3). Those with DGF had a shorter time to all-cause and death-censored graft loss, and there were relatively smaller differences in either outcome between preemptive and nonpreemptive recipients among those with and without DGF (Figure 5). Together, these analyses suggest that the benefit of preemptive transplantation in recipients of high-KDPI kidneys ≥85% is largely because of avoidance of DGF.

Table 3 - Results of unadjusted logistic regression comparing the risk of delayed graft function in those who received a preemptive transplant, stratified by reported eGFR at the time of listing, with those who received a nonpreemptive transplant Category of Kidney Transplant All KDPI ≥85% OR (95% CI) N (%) OR (95% CI) N (%) Nonpreemptive transplant Reference 30,507 (28.3) Reference 6607 (31.2) Preemptive transplant, eGFR ≤10 (n=844) 0.26 (0.20 to 0.33) 79 (9.4) 0.22 (0.13 to 0.38) 15 (9.2) Preemptive transplant, eGFR >10 (n=7948) 0.15 (0.14 to 0.17) 458 (5.8) 0.16 (0.13 to 0.19) 91 (6.6) Preemptive transplant, eGFR unknown (n=3539) 0.32 (0.29 to 0.36) 401 (11.3) 0.33 (0.25 to 0.43) 60 (12.9)

The number and percentage of individuals with events in each category is presented. KDPI, Kidney Donor Profile Index; OR, odds ratio; CI, confidence interval.

Figure 5:

Association of delayed graft function (DGF) on post-transplant outcomes in recipients transplanted preemptively and nonpreemptively. Kaplan–Meier curves of graft loss from all causes, including death (A), and death-censored graft loss (B) in recipients receiving transplants preemptively and nonpreemptively, stratified by the occurrence of postoperative DGF. The P values represent the statistical significance of the log-rank test for each analysis.

In all sensitivity analyses, the risk of all-cause and death-censored graft loss was lower in recipients of a preemptive transplant, even in patients who received a kidney from a donor with KDPI ≥85% (Supplemental Figures 8–13, Supplemental Tables 4 and 5).

Discussion

This study demonstrates that the benefit of preemptive transplantation is maintained in transplants using deceased donor kidneys with characteristics associated with a higher risk of allograft loss as measured by a KDPI ≥85%. These findings were consistent in various sensitivity analyses. Furthermore, the relative risk of allograft loss in preemptive transplants using high-KDPI kidneys ≥85% is comparable with that of nonpreemptive transplants using kidneys with KDPI 51%–84%. Given only 9% of KDPI ≥85% transplants were preemptive and 50% of recovered deceased donor kidneys with KDPI ≥85% were discarded, strategies to increase the preemptive transplantation of high-KDPI kidneys could safely increase access to kidney transplantation.

To our knowledge, the outcomes of preemptive transplantation using high-KDPI kidneys have not been rigorously evaluated.13,15–18 Although it is known that the use of high-KDPI kidneys for transplantation is best justified when transplantation results in a significant reduction in waiting time on dialysis, the seminal studies evaluating the use of high-KDPI kidneys either excluded preemptive transplants or did not specifically evaluate the outcome of preemptive transplants and the existing studies examining preemptive transplantation of high-KDPI kidneys have been limited to elderly patients.9,10,21,22 The mate kidney analysis significantly reduced the possibility of residual confounding due to unmeasured differences in donor characteristics between preemptive and nonpreemptive transplants and increased confidence in our observational findings.

An important potential explanation for the observed superior outcomes of preemptive compared with nonpreemptive transplants is lead time bias. Preemptive recipients may conceivably have higher levels of residual native kidney function or a lower cumulative burden of comorbidity compared with nonpreemptive recipients.23,24 Our finding that allograft survival among preemptive recipients of kidneys with KDPI ≥85% was longest in those waitlisted with eGFR >10 ml/min suggests a lead time bias. The odds of DGF was similarly related to the level of native kidney function at the time of waitlisting, and DGF was strongly associated with allograft survival. These findings suggest that the benefit of preemptive transplantation in high-KDPI transplants is mediated through a lower risk of DGF that is likely related to the higher level of native kidney function in preemptive recipients. While these observations warrant further investigation, irrespective of the underlying mechanism, the study findings suggest that increased preemptive use of high-KDPI kidneys may safely expand kidney transplantation.

The finding that discard of recovered high-KDPI kidneys for transplantation has not decreased is unsurprising given the lack of polices to encourage the use of these kidneys. In addition, even preemptively listed individuals must opt in to consider offers from donors with KDPI ≥85%. The last major change in deceased donor kidney allocation focused on increasing the utility from the available organ supply by preferentially allocating kidneys with the most favorable donor characteristics as measured by a KDPI of 0%–20% to waitlist candidates in the top quintile of expected post-transplant survival, but did not make any change to the allocation of high-KDPI kidneys.12 The nonutilization of recovered deceased donor kidneys for transplantation has been the subject of numerous studies and is a significant public health concern.25–27 However, few strategies have been proposed to safely expand transplantation from the available supply of donated kidneys. Factors beyond the KDPI likely affect the decision to accept or reject a deceased donor kidney offer. A limitation of registry analyses, including ours, is the inability to capture granular information about organ acceptance behaviors.27 A potentially novel way to gain insight into the acceptance practice of high-KDPI kidneys may be to examine acceptance and discard as a function of the sequence of the organ offer to transplant centers, but that is beyond the scope of this study.28

This study does not compare the outcomes of preemptive transplantation with high-KDPI kidneys with the alternative of remaining on dialysis to wait for a kidney with a lower KDPI, and t