REFILE-Time to revisit the PELD score for donor liver allocation, researchers say

Last Updated: 2018-09-19

(Adds author's affiliation in paragraph 3.)

By Megan Brooks

NEW YORK (Reuters Health) - A new analysis suggests that children are at a disadvantage when competing with adults for livers from deceased organ donors in the United States.

The study of more than 4000 children on the United Network for Organ Sharing (UNOS) waiting list found that the Pediatric End-stage Liver Disease (PELD) score significantly underestimates 90-day pretransplant waitlist mortality.

"The PELD score ranks kids appropriately, meaning a child who is sicker has a higher score than a child that is less sick," Dr. Mark Roberts, from the Department of Health Policy and Management at the University of Pittsburgh, Pennsylvania, noted in a phone interview with Reuters Health. "But that comparison fails" when comparing PELD and the adult Model for End-stage Liver disease (MELD).

"This is one of the first actual comparison of what happens when an adult and a child would be vying for the same organ," said Dr. Roberts.

Put it this way, he said: "If you have a child and an adult with the same score, say of 20, the child actually has a much higher risk of death than the adult. Therefore, since organs are distributed on your likelihood of dying in 90 days, it preferentially discriminates against children because the score artificially underestimates their actual risk of death by as much as 17%."

Fair allocation of livers to children and adults is "critically dependent" on the accuracy of mortality estimates provided by the PELD and MELD scores, the authors note in a paper in JAMA Pediatrics September 17.

Yet, widespread reliance on exception points for pediatric recipients suggests that the two systems may not be comparable. "Anecdotally, pediatric transplant physicians have long recognized that the scoring system isn't adequate when comparing children to adults," Dr. Roberts said in a news release.

In their analysis, the researchers looked back at 4,298 children with chronic liver disease (excluding cancer) who were listed for primary liver transplant from early 2002 to the spring of 2014 and followed for at least two years. They also created a reduced cohort of 2,421 that excluded those who received living donor transplantation or PELD exception points.

They found a strong concordance between actual and predicted mortality for the full cohort (C statistic = 0.84) and the reduced cohort (C = 0.81), suggesting very good discrimination. However, the estimated risk of dying using the PELD score significantly underestimated actual 90-day probability of death by as much as 17%, they report.

The PELD system, they conclude in their paper, "may be flawed and may disadvantage children awaiting liver transplant when used to adjudicate organ allocation decisions; a new system that reflects actual 90-day mortality should be developed."

The original PELD score was initially developed using only around 800 children who were listed and received transplants from 1995 to 2000. "Now we have thousands of kids in the UNOS database," said Dr. Roberts, "so I think the answer is simply to revisit and reconstruct the risk prediction scores so they are better and more accurate predictors of how likely a child is to die. As is, it is not very accurate. It does work within children, but when you compare to adults it falls apart."

The authors of a linked editorial note that this analysis is the first validation of the PELD score in nearly two decades. It "definitively shows that the PELD score does not need to be changed; it is highly concordant with waitlist mortality among children," write Dr. Evelyn Hsu from University of Washington School of Medicine and colleagues. "This study gives evidence that children are not given sufficient priority to allow access to transplant."

"There are multiple ways to ensure access for children on the waitlist to the organs they need; one would be to rank children nationally above local adults, and another would be to systematically provide additional points to all children with a calculated PELD score. Any such change is likely to induce minimal overall changes in the adult waitlist and posttransplant outcomes, while potentially reducing or eliminating pediatric liver waitlist mortality," write Dr. Hsu and colleagues.

In a second editorial, Dr. David Goldberg from University of Pennsylvania Perelman School of Medicine, says this is "important work that draws attention to poor PELD score calibration among a broad cohort of pediatric patients."

They say while the findings need to be validated with attention to several caveats they outline in their article, "the subject merits debate within the transplant community to improve allocation equity. Without question, the PELD score must seek to evolve as the MELD score has done previously (with the incorporation of sodium measurement, for example). This, of course, will require identification of clear circumstances in which PELD underestimates 90-day mortality, which may be associated with geographic region, age, causative mechanisms of illness, or factors yet to be determined," Dr. Goldberg and colleagues conclude.

The study had no commercial funding and the authors have declared no relevant conflicts of interest.

SOURCE: https://bit.ly/2NmK3,Wp https://bit.ly/2NmKgc9 and https://bit.ly/2NmK3Wp

JAMA Pediatr 2018.