Development and validation of a therapy-independent hepatocellular carcinoma survival prediction score

Abstract

Background & Aims

Survival in hepatocellular carcinoma (HCC) is associated with several factors. Our aim was to develop and validate an HCC survival prediction score (HCC-SPS) based on common clinical parameters and excluding the subsequent therapy received, which would be able to prognosticate all patients with HCC at the time of diagnosis.

Methods

The development cohort comprised 1270 patients with HCC seen in our department since January 1988. Univariate analysis was performed for known HCC prognostic parameters. Parameters with P < .1 on univariate analysis were then included in a Cox regression with backward model selection. The HCC-SPS was derived based on the coefficients estimated by Cox regression with selected parameters. The derived HCC-SPS was then validated with 2 independent international cohorts of 220 patients from the United Kingdom and 90 patients from Hong Kong (HK). Points were allocated to the following variables: ALBI grade, AFP level, portal vein invasion, ECOG status and TNM stage.

Results

The total score classified a patient into 3 distinct survival risk categories of low, medium and high risk with median survival (weeks) of 249 (95% CI 195–303), 45 (95% CI 38–52) and 9 (95% CI 8–10) respectively. The scoring system was validated by the cohorts from United Kingdom and HK.

Conclusions

We have formulated an HCC survival prediction score using readily available clinical parameters to risk stratify all HCC patients into distinct survival categories at the time of HCC diagnosis regardless of subsequent treatment received. The score was validated with other independent international cohorts of patients.