Prediction models that incorporate more than just treatment status could rank order heart transplant candidates by urgency more effectively than the current system, a modeling study suggests.
Since 2018, the U.S. heart transplant allocation system has ranked heart candidates according to six treatment-based “statuses” (up from three used previously), ignoring many objective patient characteristics, the authors write.
Their study showed no significant difference in survival between statuses four and six, and status five had lower survival than status four.
“We expected multivariable prediction models to outperform the six-status system when it comes to rank ordering patients by how likely they are to die on the wait list (medical urgency),” William F. Parker, MD, MS, PhD, of the University of Chicago, told this news organization.
“However, we were surprised to see that the statuses were out of order,” he said. “Status five patients are more urgent than status three or status four patients,” mainly because most are in renal failure and listed for multiorgan transplantation with a kidney.
Objective physiologic measurements, such as glomerular filtration rate (GFR), had high variable importance, offering a minimally invasive measurement with predictive power in assessing medical urgency. Therefore, including GFR and other variables such as extracorporeal membrane oxygenation (ECMO) could improve the accuracy of the allocation system in identifying the most medically urgent candidates, Dr. Parker and colleagues suggest.
The study was published online in JACC: Heart Failure.
‘Moderate ability’ to rank order
The investigators assessed the effectiveness of the standard six-status ranking system and several novel prediction models in identifying the most urgent heart transplant candidates. The primary outcome was death before receipt of a heart transplant.
The final data set contained 32,294 candidates (mean age, 53 years; 74%, men); 27,200 made up the prepolicy training set and 5,094 were included in the postpolicy test set.
The team evaluated the accuracy of the six-status system using Harrell’s C-index and log-rank tests of Kaplan-Meier estimated survival by status for candidates listed after the policy change (November 2018 to March 2020) in the Scientific Registry of Transplant Recipients data set.
They then developed Cox proportional hazards models and random survival forest models using prepolicy data (2010-2017). Predictor variables included age, diagnosis, laboratory measurements, hemodynamics, and supportive treatment at the time of listing.
They found that the six-status ranking at listing has had “moderate ability” to rank order candidates.
As Dr. Parker indicated, statuses four and six had no significant difference in survival, and status five had lower survival than status four.
The investigators’ multivariable prediction models derived with prepolicy data ranked candidates correctly more often than the six-status rankings. Objective physiologic measurements, such as GFR and ECMO, were identified as having significant importance with regard to ranking by urgency.
“The novel prediction models we developed … could be implemented by the Organ Procurement and Transplantation Network (OPTN) as allocation policy and would be better than the status quo,” Dr. Parker said. “However, I think we could do even better using the newer data collected after 2018.”
Modifications underway
The OPTN Heart Transplantation Committee is currently working on developing a new framework for allocating deceased donor hearts called Continuous Distribution.
“The six-tiered system works well, and it better stratifies the most medically urgent candidates than the previous allocation framework,” the leadership of the United Network for Organ Sharing Heart Transplantation Committee, including Chair Richard C. Daly, MD, Mayo Clinic; Vice-Chair Jondavid Menteer, MD, University of Southern California, Los Angeles; and former Chair Shelley Hall, MD, Baylor University Medical Center, told this news organization.
“That said, it is always appropriate to review and adjust variables that affect the medical urgency attribute for heart allocation.”
The new framework will change how patients are prioritized, they said. “Continuous distribution will consider all patient factors, including medical urgency, together to determine the order of an organ offer, and no single factor will decide an organ match.
“The goal is to increase fairness by moving to a points-based allocation framework that allows candidates to be compared using a single score composed of multiple factors.
“Furthermore,” they added, “continuous distribution provides a framework that will allow modifications of the criteria defining medical urgency (and other attributes of allocation) to a finer degree than the current policy. … Once continuous distribution is in place and the OPTN has policy monitoring data, the committee may consider and model different ways of defining medical urgency.”
Kiran K. Khush, MD, of Stanford (Calif.) University School of Medicine, coauthor of a related commentary, elaborated. “The composite allocation score (CAS) will consist of a ‘points-based system,’ in which candidates will be assigned points based on (1) medical urgency, (2) anticipated posttransplant survival, (3) candidate biology (eg., special characteristics that may result in higher prioritization, such as blood type O and allosensitization), (4) access (eg., prior living donor, pediatric patient), and (5) placement efficacy (travel, proximity).”
Candidates will be assigned points based on these categories, and will be rank ordered for each donor offer.
Dr. Khush and colleagues propose that a multivariable model – such as the ones described in the study – would be the best way to assign points for medical urgency.
“This system will be more equitable than the current system,” Dr. Khush said, “because it will better prioritize the sickest candidates while improving access for patients who are currently at a disadvantage [for example, blood O, highly sensitized patients], and will also remove artificial geographic boundaries [for example, the current 500-mile rule for heart allocation].”
Going further
Jesse D. Schold, PhD, of the University of Colorado at Denver, Aurora, raises concerns about other aspects of the heart allocation system in another related commentary.
“One big issue with our data in transplantation … is that, while it is very comprehensive for capturing transplant candidates and recipients, there is no data collection for patients and processes of care for patients prior to wait list placement,” he told this news organization. This phase of care is subject to wide variation in practice, he said, “and is likely as important as any to patients – the ability to be referred, evaluated, and placed on a waiting list.”
Report cards that measure quality of care after wait list placement ignore key phases prior to wait list placement, he said. “This may have the unintended consequences of limiting access to care and to the waiting list for patients perceived to be at higher risk, or the use of higher-risk donors, despite their potential survival advantage.
“In contrast,” he said, “quality report cards that incentivize treatment for all patients who may benefit would likely have a greater beneficial impact on patients with end-organ disease.”
There is also significant risk of underlying differences in patient populations between centers, despite the use of multivariable models, he added. This heterogeneity “may not be reflected accurately in the report cards [which] have significant impact for regulatory review, private payer contracting, and center reputation.”
Some of these concerns may be addressed in the new OPTN Modernization Initiative, according to David Bowman, a public affairs specialist at the Health Resources and Services Administration. One of the goals of the initiative “is to ensure that the OPTN Board of Directors is high functioning, has greater independence, and represents the diversity of communities served by the OPTN,” he told this news organization. “Strengthened governance will lead to effective policy development and implementation, and enhanced transparency and accountability of the process.”
Addressing another concern about the system, Savitri Fedson, MD, of the Michael E. DeBakey VA Medical Center and Baylor College of Medicine, Houston, wonders in a related editorial whether organ donors and recipients should know more about each other, and if so, could that reverse the ongoing downward trend in organ acceptance?
Although some organizations are in favor of sharing more information, Dr. Fedson notes that “less information may have the greater benefit.” She writes, “We might realize that the simplest approach is often the best: a fulsome thank you for the donor’s gift that is willingly given to a stranger without expectation of payment, and on the recipient side, the knowledge that an organ is of good quality.
“The transplant patient can be comforted with the understanding that the risk of disease transmission, while not zero, is low, and that their survival following acceptance of an organ is better than languishing on a waiting list.”
The study received no commercial funding. Dr. Parker, Dr. Khush, Dr. Schold, and Dr. Fedson report no relevant financial relationships.
A version of this article first appeared on Medscape.com.