Disparities in Acceptance of Deceased Donor Kidneys Between the United States and France and Estimated Effects of Increased US Acceptance

Key Points Question Would a more aggressive approach to organ acceptance provide a benefit to wait-listed kidney transplant candidates? Findings This cohort study analyzes the use of 156 089 deceased donor kidneys in the United States and 29 984 in France and finds that the US discard rate of these kidneys is nearly twice that of France. It uses computer simulation to model a lower US discard rate similar to that of France, and estimates a US increase of 132 445 allograft life-years. Meaning Greater acceptance of kidneys from older and comorbid deceased donors in the United States could provide major survival benefits to the population of wait-listed patients.


Discrimination
The aim of discrimination is to distinguish between patients who experience an event from those who do not. The AUC (0≤C≤1) is the probability of concordance between predicted and observed events, with AUC=0.5 for random predictions and AUC=1 for a perfectly discriminating model.

Calibration
Calibration refers to the ability to provide unbiased predictions in groups of similar patients. It estimates how close the estimated risk is to the observed risk. A prediction model is considered "well-calibrated" if the difference between predictions and observations in all groups of similar patients is close to 0 (perfect calibration). Any large deviation (P<0.1) indicates a lack of calibration.

Monte-Carlo technique
Acceptance practices were simulated for each kidney separately (not at the donor level). For each kidney k of a given deceased donor d from one of our 2 countries, the following procedures were performed: (i) a uniform 0-to-1 random number Udk was generated; and (ii) its probability Pdk of being discarded was computed from the other-country logistic regression model according to the actual KDRI Kd of the donor, and the kidney k was virtually discarded if and only if Udk ≤ Pdk.

Calculation of the Kidney Donor Risk Index (KDRI) and the Kidney Donor Profile Index (KDPI)
The KDRI and KDPI are scores derived from 10 deceased donor variables and predict risk of kidney allograft failure after transplantation. Lower values for the KDRI and KDPI indicate kidneys with better projected allograft survival.
These indices were developed for the purpose of enabling clinicians to try to make rough assessments of allograft quality and graft failure risk across donors with different attributes.(1) A guidance document from the United Network for Organ Sharing states "The Kidney Donor Profile Index (KDPI) is a numerical measure that combines ten donor factors, including clinical parameters and demographics, to summarize into a single number the quality of deceased donor kidneys relative to other recovered kidneys." (2) The KDRI score for any kidney allograft estimates the risk of failure for that allograft compared to a kidney from a reference donor. This reference donor is defined as 40 years old, non-African-American, 170 cm tall, weighing 80 kg, with a creatinine level of 1 mg/dL, as well as negative history of hypertension, diabetes and hepatitis C virus serostatus.
Notably, race/ethnicity for organ donors is not available according to French national bioethics rules. As a result, as we entered "non-black" as the race for all French donors when calculating KDRI. This approach to calculating the KDRI of the French pool of donated kidneys will have the net effect of slightly over-estimating the quality of these organs. Nonetheless, as shown in the results, French transplant centers are still much more likely to accept kidneys with the highest KDRI scores (i.e. lowest quality kidneys) compared to US centers.
By convention, we mapped the calculated KDRI values onto a cumulative percentage scale from 0-100 to generate the KDPI. Because our analysis focused on kidneys recovered from 2004 to 2014, as recommended by the OPTN, the 2015 scaling factor for converting "KDRI Rao" to the KDRI median was used. Lower values for the KDRI indicate kidneys with better projected allograft survival. The KDRI and KDPI are not formally used in kidney allocation in France, but donor characteristics used to calculate the scores (with the exception of donor race) are presented to transplant centers with organ offers.

Systematic literature review supporting the study novelty
We conducted a comprehensive search of PubMed Plus and EBSCO MegaFILE using the keywords "kidney transplantation" and "discard" for all articles published from October 2008 to March 2018. Search: ((("kidney transplantation") AND discard) AND ("2008/10"[Date -Publication]: "3000"[Date -Publication])). Two investigators independently reviewed the results and eliminated four irrelevant publications, leaving 104 publications. We concluded that similar deceased donor characteristics -such as older age or acute kidney injury -elevate the risk of deceased donor kidney discard in transplant systems across multiple countries. Finally, the systematic review revealed that no studies systematically assessed kidney discard rates between different countries and their impact in term of gain or loss of opportunities for waitlisted patients. No study has used computer simulation algorithms to address the possibility of virtually applying a European Union-based approach of kidney allocation and acceptance to the US transplant system.

eFigure 2: Distribution of deceased donor kidney quality as manifested in the Kidney Donor Risk Profile (KDPI) kidneys in the US and in France.
Panel A shows the distribution of the KDPI score for the transplanted (blue) and discarded kidneys (red) in the US. Panel B shows the distribution of the KDPI score for the transplanted (blue) and discarded kidneys (red) in France. For instance, for a KDPI score of 100%, 78% of the kidneys were discarded in the US, while 29% of the kidneys were discarded in France.

eFigure 3: Change in the mean deceased donor age and Kidney Donor Risk Index (KDRI) in the US and France between 2004 and 2014.
Panel A shows the mean donor age over time in the US for transplanted (black) and discarded (blue) kidneys. Panel B shows the mean donor age over time in France for transplanted (black) and discarded (blue) kidneys. Panel C shows the mean KDRI over time in the US for transplanted (black) and discarded (blue) kidneys. Panel D shows the mean KDRI over time in France for transplanted (black) and discarded (blue) kidneys. The vertical axis is the observed proportion of grafts discarded. The x-axis shows the probability of discard using the model. The dashed black lines correspond to the ideal calibration, the red curve corresponds to the bias-corrected calibration in the US, and the blue curve corresponds to the bias-corrected calibration in France. eFigure 7: Restricted mean survival time at 10 years in the US for the 10 deciles of the KDRI. The pink area under the curve corresponds to the mean survival probability at ten years. The orange area above the curve corresponds to the mean years of allograft lost at ten years. eFigure 9: Kaplan-Meier curves for non-death-censored kidney graft survival, according to the deciles of the Kidney Donor Risk Index in the US after transplantation. This figure shows the probability of graft survival among actual kidney transplant recipients in the US and stratified by the KDRI deciles. The black curve corresponds to KDRI decile 1, the gray curve to KDRI decile 2, the green curve to KDRI decile 3, the light blue curve to KDRI decile 4, the dark blue curve to KDRI decile 5, the yellow curve to KDRI decile 6, the orange curve to KDRI decile 7, the pink curve to KDRI decile 8, the purple curve to KDRI decile 9, and the red curve to KDRI decile 10. The overall difference was significantly different (logrank p<0.0001).

eFigure 10: Kaplan-Meier curves for non-death-censored kidney-graft survival, according to the deciles of the Kidney Donor Risk Index in France after transplantation.
This figure shows the probability of graft survival among actual kidney transplant recipients in the France and stratified by the KDRI deciles. The black curve corresponds to KDRI decile 1, the gray curve to KDRI decile 2, the green curve to KDRI decile 3, the light blue curve to KDRI decile 4, the dark blue curve to KDRI decile 5, the yellow curve to KDRI decile 6, the orange curve to KDRI decile 7, the pink curve to KDRI decile 8, the purple curve to KDRI decile 9, and the red curve to KDRI decile 10. The overall difference was significantly different (logrank p<0.0001). eFigure 11: Estimation of gains in non-death-censored allograft life years from reducing high kidney discard rates in the US, according to KDRI categories. Panel A shows the life years saved by decile of KDRI in the US if the French model of kidney acceptance had been used. The black curve corresponds to KDRI decile 1, the gray curve to KDRI decile 2, the green curve to KDRI decile 3, the light blue curve to KDRI decile 4, the dark blue curve to KDRI decile 5, the yellow curve to KDRI decile 6, the orange curve to KDRI decile 7, the pink curve to KDRI decile 8, the purple curve to KDRI decile 9, and the red curve to KDRI decile 10. The x-axis corresponds to the time post transplantation, and the vertical axis corresponds to the number of allograft life years saved. Panel B shows the life years saved overall if organ utilization patterns in the US had followed the French model. The x-axis corresponds to the time post transplantation, and the vertical axis corresponds to the number of allograft life years saved.