Who Gets a Heart Transplant? Ranking Process Must Evolve, Again

Because objective, fluctuating physiologic measurements can substantially improve the prediction of who will die while waiting for a heart transplant, scores that better rank patients by medical urgency—and how well they match the specific available organ—will likely be more efficient than the current six-status system for allocating organs, according to new data.

It’s no secret that demand far outweighs supply when it comes to heart transplantation in the United States. The agencies in charge of organ allocation—the Organ Procurement and Transplantation Network (OPTN) and United Network for Organ Sharing (UNOS), both overseen by the US Department of Health and Human Services—last changed the process for deciding who would receive a heart in October 2018, increasing the number of patient “statuses” from three to six based on disease severity as well as other factors and generally making the process fairer.

While this system has been an improvement over what was used before, experts in the heart transplant field argue that a continuous scoring approach will be more equitable.

“Although the six-status policy incorporates some limited physiologic data, it still has few risk categories and relies heavily on treatment as a proxy for medical urgency,” Kenley M. Pelzer, PhD (University of Chicago, IL), and colleagues write in JACC: Heart Failure. “Multivariable prediction models may be significantly more accurate than the six-status system and may better fulfill the federal requirement to rank-order candidates by urgency.”

To TCTMD, senior author William F. Parker, MD, PhD (University of Chicago, IL), said evolution to newer models is imperative “to save the most lives we can—not just in patients who are lucky enough to end up on the list, but the entire advanced heart failure population.”

In a related editorial, Parker as well as Kiran K. Khush, MD, and Alexander T. Sandhu, MD (both from Stanford University School of Medicine, CA), acknowledge that compared with what came before, the six-status system “better stratified candidates by waitlist mortality, shortened waiting times for high priority candidates (without apparent adverse effects on recipient outcomes), and broadened geographic sharing of donor hearts with acuity circles.”

However, they write, “disparities in access to transplantation remain and in some cases may have been exacerbated.”

Two New Models vs the Current

For the study, Pelzer and colleagues compiled data from 32,294 adult heart transplant candidates (mean age 53 years; 73.7% male) listed in the US between 2010 and 2020. Compared with those listed before the policy update (n = 27,200), those listed after October 2018 (n = 5,094) were more likely to be treated with intra-aortic balloon pump (IABP; 4.7% vs 13.2%) and extracorporeal membrane oxygenation (ECMO; 1.6% vs 2.8%).

Following the policy implementation, the six-status system was moderately able to rank-order candidates with regard to the primary outcome of death before receipt of heart transplant (C-index 0.67). However, there was not a difference seen in the primary outcome between those listed as status 4 and 6 (P = 0.8), whereas patients denoted as status 5 actually had lower survival than those marked status 4 (P < 0.001).

Compared with the six-status system, two novel multivariable prediction models derived with prepolicy data—Cox proportional-hazards (CPH) model and random survival forest (RSF) model—ranked candidates correctly more often (C-indexes 0.76 and 0.74, respectively).

Within the RSF model specifically, glomerular filtration rate and ECMO were the variables that had the greatest importance on outcomes, followed by age, pulmonary capillary wedge pressure, “other” ventricular assist device (ie, not LVAD), IABP, mean pulmonary arterial pressure, use of IV inotropes, and cardiac index. This ranking list of variables by importance was similar for the CPH model.

Thus, their findings show that even the six-status allocation system has “limited accuracy,” according to Pelzer and colleagues. “Our work clearly shows that discriminative power increases when a broader set of characteristics, including physiologic measurements, is considered. . . . The inclusion of more-objective physiologic data (including data that would be easily measured by blood tests) could improve the accuracy of the heart allocation system in identifying and prioritizing the most medically urgent candidates.”

Parker said while they expected the newer models to outperform the six-status system, he was surprised to see that the current statuses are “out of order,” partially explained by the fact that those listed as status 5 need multiorgan transplant—usually for heart and kidney—and thus are in renal failure at the time of listing. “That is one of the strongest predictors of death we found in our models,” Parker said.

But even though these newer continuous models performed well, he doesn’t think that either one should be implemented since both were each designed using pre-2018 data. Rather, this study was “just sort of a point of proof of principle that it's relatively easy to outperform these six categories, and we need to develop a new heart allocation score using the data after 2018, which uses all these new variables that are being collected, new laboratory data that is being frequently updated,” Parker said. “Develop[ing] a better heart allocation score for a multivariable prediction model is the next step.”

One “big hole” that will have to be addressed is the use of left ventricular assist devices (LVADs), which work well at keeping patients alive on the waitlist but can often lead to them being pushed down the rankings, Parker said. The problem may only be compounded by the advent of LVAD-only centers. “There is kind of a disincentive built in by both the current system, which gives LVAD patients lower priority and a score would do the same thing,” he said. One solution would be to take this into account and potentially give “LVAD equity points” toward a heart transplant to patients who have been waiting for a certain length of time.

Whatever design the six-status system ultimately evolves into, “it should be the first of multiple iterations,” Khush and colleagues write. “Any allocation system will quickly become outdated and in need of revision; the past cannot perfectly predict the future because clinical practice will change, partly in response to evolving allocation rules.”

For example, they explain, temporary mechanical circulatory support use went up after the 2018 policy was enacted, and this impacted its prognostic significance.

“Instead of ignoring this reality, the allocation system should be intentionally adaptive,” the editorialists continue. “To enable such adaptation, the United Network for Organ Sharing should continue to collect granular patient data and prospectively plan updates to the continuous distribution model. If a specific patient group consistently has higher waitlist mortality, the score should be revised to increase the medical urgency for that cohort. Forward-thinking and flexible adaptations will move us ever closer to an allocation system that finally achieves equitable access to transplantation for the large and diverse US population of patients suffering from end-stage heart failure.”