Is Dynamic AFP-R Key to Transplant Selection in Liver Cancer?

A hepatocellular carcinoma (HCC) transplant selection tool incorporating dynamic alpha-fetoprotein response (AFP-R) accurately predicted patient outcomes and could potentially increase access to the curative procedure, a retrospective analysis suggested.

In adults who underwent liver transplant for HCC, using the New York/California (NYCA) score — which incorporates dynamic AFP-R, defined as the difference between maximum pre-liver transplant AFR levels — accurately predicted both recurrence-free and overall survival, according to researchers led by Karim Halazun, MD, Weill Cornell Medicine in New York City.

Five-year cumulative incidence of recurrence risk was 9.5% for low-risk patients, 20.5% for acceptable-risk patients, and 40.5% for high-risk patients (P<0.001 for all), the group reported in JAMA Surgery.

The NYCA score also showed superiority to other available tools. Using receiver operating characteristic analysis, Halazun’s team showed that the validated score was significantly better at predicting recurrence, with a Harrell C statistic of 0.66, as compared to 0.60 for Metroticket 2.0, 0.57 for French-AFP, and 0.58 for the Milan Criteria (MC).

“Our data suggest that AFP-R is a more robust biological metric than simply using AFP at listing or transplant,” the authors wrote.

According to Halazun and colleagues, while the adoption of the MC for selecting HCC patients eligible for liver transplant has improved survival for thousands of patients, it has its detractors.

“Criticisms levied against MC’s dichotomous nature, restrictiveness, and lack of tumor biological indexes resulted in the emergence of several scores to improve selection,” the group wrote.

They pointed out that while several European centers have adopted tools that include the AFP level, these tools are reliant on static AFP levels.

The current analysis included 2,236 adults (81% men, average age 58 years) who underwent liver transplant for HCC at eight centers in North American and Europe from 2001 to 2013. Hepatitis C diagnosis was most common in North American patients (60.1%) while alcohol-related liver disease was most common among the European cohort (37.2%).

The original NYCA score included points for tumor size and number, and AFP-R. Those scores were assigned to patients in the current study and a validated NYCA score was generated.

The researchers suggested that the superiority of the AFP-R approach extends beyond improved survival outcomes, as it also allows for the inclusion of patients who would have been excluded by both the MC and French-AFP.

Overall, 83.1% of patients who did not meet MC, 69.2% of patients who did not meet French-AFP, and 76.1% of patients who did not meet Metroticket 2.0 would be re-categorized into either low-risk or acceptable-risk NYCA groups, with 5-year recurrence-free survival greater than 75% and a 5-year overall survival greater than 70%.

This “underlies the importance of moving away from dichotomous single AFP non-kinetic scores,” they wrote.

They recommended that incorporating such a response-based biological assessment of HCC patients into selection tools should become standard “as our understanding of their importance evolves.”

However, in a commentary accompanying the study, Chirag S. Desai, MD, and A. Sidney Barritt IV, MD, MSCR, both of the University of North Carolina School of Medicine in Chapel Hill, wrote that before implementing such a metric as AFP-R, “a few critical issues from these data need to be considered.”

For example, everyone in the study underwent transplant and were deemed to have appropriate tumor biologic factors. Thus, Desai and Barritt noted there was substantial selection bias in the study population. In addition, they pointed out that the study took place at a time when liver graft allocation changed to include the Model for End-Stage Liver Disease (MELD) system, and that there were many differences in pretransplant treatment strategies, listing criteria, waiting time, and other transplant-related practices in this study cohort.

“In summary, this study advances our knowledge regarding the predictive ability of AFP-R,” they wrote, but added that “universal adaptation for deceased donor graft allocation still needs prospective validation.”