Background: Transarterial chemoembolization (TACE) is the standard first-line therapy for intermediate-stage hepatocellular carcinoma (HCC). However, no latent-classing indices, concerning repeat conventional TACE or switching to another treatment, have been incorporated into the guidelines. Methods: The unsupervised latent class modeling was applied to identify subphenotypes using the clinical and medical imaging data of 1517 HCC patients after the first TACE from four hospitals (derivation cohort: 597 cases; validation cohort: 920 cases); modeling was conducted independently in each cohort. We then explored the relationship of subphenotypes with clinical outcomes in both cohorts and response to treatment strategies after the first TACE in the derivation cohort. Results: Independent latent class models suggested that a three-class model was optimal for both cohorts. In both cohorts, we identified a TACE-refractory subphenotype (Phenotype 1: PS score 1, stage progress, more intrahepatic lesions, and new intrahepatic lesions), TACE-responsive subphenotype (Phenotype 3: PS score 0, No intrahepatic lesions and new intrahepatic lesions), compared to TACE-intermediate subphenotype (Phenotype 2). Compared to Phenotype 1 or 2, patients in Phenotype 3 had significantly lower 3-month or 3-year mortality (all P<0.001). In the derivation cohort, the effects of treatment strategy (surgery/ablation vs. repeat TACE vs. stop TACE) differed significantly in phenotype 2 but not in phenotype 3 (P=0.721 for interaction). Conclusions: Latent class models identified three subphenotypes for HCC after the first TACE treatment. Differences were significant in clinical outcome and response to treatment strategy after the first TACE among three subphenotypes.
Keywords: Hepatocellular carcinoma; Latent class analysis; Subphenotype; Transarterial chemoembolization.
© The author(s).