Applicability of Risk Scores to an Indian Cohort of Hepatitis B-Related Hepatocellular Carcinoma Patients

Graphical abstract

Sir,

In India, the commonest cause of hepatocellular carcinoma (HCC) is hepatitis B (HBV) and studies have shown it to develop even in long-term treated patients. To assess such risk, various scores have been developed in HBV population from different world regions,¹ e.g. PAGE-B (Europe)², REAL-B (USA and Asia Pacific)³, HCC-RESCUE (Korea)⁴, CAMD (Taiwan)⁵ and most recently the aMAP score⁶ in 11 international patient cohorts (Supplementary Table 1). No Indian studies are available in this respect. In our recent study where 154 HBV patients treated with entecavir were followed up for a mean of 6.85 [±2.6 (2–14)] years,⁷ 1.9% (3/154) developed HCC, two after 6 years and one after 8 years of therapy. The discriminatory and predictive power of the above five HCC risk scores in predicting the occurrence of HCC in this population is presented here by comparing their AUROC and C-index along with the cumulative hazards of HCC in their risk subgroups. Most of these scores include cirrhosis or its surrogate marker in calculation (pretreatment cirrhosis was present in 13% patients in our study) but we excluded scores containing LSM values as it was not done in our study. Kaplan Meier survival and Cox proportional hazards regression analysis were used and P < 0.05 was considered significant.

The results are depicted in Figure 1, Figure 2. The aMAP score performed best in predicting the occurrence of HCC (AUROC 0.9, which was significantly higher than REAL-B, HCC RESCUE, and CAMD scores). PAGE-B score (AUROC 0.79) was the next in predictive efficiency but its AUROC was not significantly different from other scores. aMAP score also had the best discriminatory accuracy (C-index 0.75 and hazards ratio 1.25) followed by PAGE-B score (C-index 0.68 but insignificant hazards ratio). Considering a discrimination or efficiency (AUROC/C-index) value of less than 0.7 as an unsatisfactory prediction model,⁸ no other score had significant predictive or discriminatory value. The high-risk subgroup of aMAP score significantly predicted HCC (hazards ratio 5.25, log rank P = 0.02 compared to intermediate group and a value >58 had a sensitivity/specificity of 100%/82.1% in predicting HCC). The subgroups of other risk scores were not significantly different from each other in their predictability of HCC.

Figure 1.

Comparison of AUROC for various HCC risk scores.

Figure 2.

Cumulative hazards of HCC for various risk scores according to their risk subgroups (L = low risk, M = moderate/intermediate risk, H = high risk).

The aMAP score was developed in worldwide population and its suitability in predicting risk of HCC in treated Indian HBV patients further lends credence to its worldwide applicability. The other scores appear to have less applicability to the Indian HBV population. The INASL Task Force on HCC in 2019⁹ recommended the use of CU HCC and PAGE-B scores for surveillance of HCC in Indian HBV patients. CU HCC score was developed in partially treated patients unlike our study where patients received full treatment. We believe that our result will stimulate future research involving more number of HBV patients on HCC risk scores or to find any new score that may be applicable to the Indian population.

Appendix A. Supplementary data

The following is the Supplementary data to this article: