Author: Srinivas Cheenu Kappadath, James Long, Armeen Mahvash, Anna Marks, Ajalaya Teyateeti 👨🔬
Affiliation: Mahidol University, The University of Texas MD Anderson Cancer Center, UT MD Anderson Cancer Center 🌍
Purpose:
Alpha-fetoprotein (AFP) has been implicated as prognostic factor of overall survival (OS) for hepatocellular carcinoma (HCC). We present a statistical approach to determine optimal AFP-cutoff and establish them for HCC patients undergoing 90Y-glass trans-arterial radioembolization (TARE).
Methods:
This single-center IRB-approved retrospective study included 180 HCC patients who underwent TARE as part of standard-of-care from November-2016 through October-2018. Fifty-eight patients with less-aggressive disease (no metastases, nodal involvement, or infiltrative tumors) underwent only TARE (TARE-only), while others underwent TARE in combination with sorafenib (TARE+sorafenib). AFP was measured prior to TARE and every subsequent 2-4months until death/disease progression. Optimal AFP-cutoff was determined in two steps. First, the range of potential AFP-cutoff was determined by calculating mean hazard-ratios (HR) using Cox-proportional-hazards regression (with 1000-bootstrapped samples for error estimation). Second, the Kaplan-Meier survivor analysis (KM) was used to determine the median OS (OSmedian) and calculate the z-score statistic across the potential AFP-cutoff range identified (with 500-bootstrapped samples for error estimation). AFP-cutoffs with the highest z-scores were used to establish suitable AFP-cutoff for total, TARE-only, and TARE+sorafenib populations.
Results:
Potential AFP-cutoff based on HR ranged 3.5–32991, 29.5–400, and 3–32888 ng/mL for total, TARE-only, and TARE+sorafenib populations. The corresponding optimal AFP-cutoffs considering KM and z-scores were 1134, 168, and 1134 ng/mL. Within TARE-only with OSmedian 21.6 months, the AFP-cutoff <168 ng/mL (n=44) yielded OSmedian of 33.9±7.5 months (and 8.7±3.1 months for others). Within TARE+sorafenib with OSmedian 12.4 months, the AFP-cutoff <1134 ng/mL (n=90) yielded OSmedian of 13.3±2.7 months (and 7.6±3.0 months for others).
Conclusion:
A statistical approach to determine optimal AFP-cutoff was developed. AFP serves as an important prognostic factor; note the observed four-fold increase in OSmedian (34-versus-9 months) with AFP-cutoff <168 ng/mL for patients with less-aggressive HCC undergoing TARE. The optimal AFP-cutoff depended on the HCC grade (168-versus-1134 ng/mL).