BOSTON – A new scoring system fared better than the Milan criteria in predicting risk for recurrent hepatocellular carcinoma after liver transplant.
The scoring system combines an assessment of the number and size of tumors on the explanted liver with serum alpha-fetoprotein (AFP) score and the presence of microvascular invasion to assign a numeric risk score to patients who have received liver transplant for hepatocellular carcinoma (HCC).
The risk estimation tool showed good model estimation, with a C statistic of 0.82 (95% confidence interval, 0.77-0.86), according to Neil Mehta, MD, who presented the study validating the risk estimation tool at the annual meeting of the American Association for the Study of Liver Diseases. The C statistic is a measure of goodness of fit, with values closer to 1 indicating better fit.
In a paper published simultaneously in JAMA Oncology, Dr. Mehta and his collaborators said that the tool had “superior recurrence risk classification, compared with explant Milan criteria (net reclassification index, 0.40; P = 0.001) in the study’s validation cohort” (JAMA Oncol. 2016 Nov 13. doi: 10:1001/jamaoncol2016/5116).
Under the Milan criteria, patients with HCC are eligible for liver transplant if there are no extrahepatic manifestations and no vascular invasion. One lesion must be smaller than 5 cm, and up to three lesions smaller than 3 cm are permitted.
Dr. Mehta, a gastroenterologist and transplant hepatologist at the University of California, San Francisco, worked with colleagues to narrow down a long list of factors that in previous studies had been associated with HCC recurrence. The final Risk Estimation of Tumor Recurrence After Transplant (RETREAT) score uses just three factors that multivariable analysis found were most highly predictive of HCC recurrence after liver transplant.
The three factors are microvascular invasion; AFP at the time of liver transplant; and the sum of the largest viable tumor diameter plus the number of viable tumors, assessed at the time of explant.
This last factor uses pathology examination, rather than radiology results, to assess tumor number, size, and viability. If a tumor is found to be completely necrotic, for example, it is not included when counting tumors for this component of the RETREAT score.
To develop the scoring system, Dr. Mehta and his coauthors enrolled adult patients with HCC who had liver transplant at three centers, each with a different wait time for transplant – short, medium, or long. Patients in the development cohort (n=721) had Model for End-Stage Liver Disease (MELD) score exception, and patients always met Milan criteria on imaging.
The validation cohort (n=340) had the same inclusion and exclusion criteria, but differed in several significant ways from the development cohort, with more young people, men, and individuals with hepatitis B or alcohol abuse. Individuals in the validation cohort were less likely to have received locoregional therapy (LRT) before liver transplant, but were more likely to have had a single tumor. They had more microvascular invasion, more tumors that were poorly differentiated, and more tumor staging beyond Milan criteria on explant.
The primary outcome measures for the study were overall survival and 5-year post–liver transplant HCC recurrence.
In developing the RETREAT scoring system, investigators used the final multivariable model coefficients to produce a “simplified point scale reflecting the relative impact of model covariables,” Dr. Mehta and his coauthors wrote.
RETREAT scoring ranges from 0 to 8; in the studied patients, the most common score was 1. Any microvascular invasion is assigned a point value of 2. The sum of the largest viable tumor diameter – in centimeters – plus the number of viable tumors is divided into four categories. If there are no tumors, the score is 0. For a sum of 1.1-4.9, the score is 1; a sum of 5.0-9.9 is assigned a score of 2, and a sum or 10 or greater is assigned a score of 3.
Serum AFP at the time of liver transplant (measured in ng/mL) is given a score of 0 if the value is less than 20. AFP of 20-99 is assigned a score of 1, values of 100-999 a score of 2, and 1,000 or greater a score of 3.
“Predicted risk of 1- and 5-year HCC recurrence rose with each point scored such that a patient with a RETREAT score of 5 or higher … had a predicted 1- and 5-year recurrence risk of 39.3 (95% CI, 25.5%-50.5%) and 75.2% (95% CI, 56.7%-85.8%), respectively,” the researchers wrote.
Dr. Mehta and his coauthors emphasized that the RETREAT scoring system represents an effort to strike a balance between including clinically meaningful variables to predict HCC recurrence and retaining simplicity and ease of use.
The researchers suggested guidelines for post–liver transplant HCC recurrence surveillance based on RETREAT scoring. They recommend no follow-up for a score of 0; surveillance scans and AFP monitoring every 6 months for 2 years for a score of 1-3, and every 6 months for 5 years for a score of 4. RETREAT scores of 5 or higher warrant surveillance every 3-4 months for 2 years, followed by every 6 months through year 5.
This surveillance stratification, they said, should help control costs while retaining a vigilant approach for those most at risk of recurrence. This approach is currently in use at University of California, San Francisco.
Study limitations included some missing information on AFP levels in the development cohort, and a small number of missing histologic tumor grading in both cohorts. In order to address limitations in the retrospective design of the present study, Dr. Mehta and his coauthors are planning another multicenter study to evaluate RETREAT as a surveillance tool and confirm its prognostic value.
The authors reported no conflicts of interest. The study was funded by the Biostatistics Core of the UCSF Liver Center.
On Twitter @karioakes