Using Machine Learning to Identify Responders to TACE or HAIC for uHCC

Radiomics-based Unsupervised Machine Learning Model to Identify Responders to Transarterial Chemoembolization Versus Hepatic Arterial Infusion Chemotherapy in Unresectable Hepatocellular Carcinoma: a Retrospective Cohort Study

The goal of this observational study is to learn about the efficacy of Transarterial Chemoembolization (TACE) versus Hepatic Arterial Infusion Chemotherapy (HAIC) in patients with unresectable hepatocellular carcinoma (HCC). The main questions it aims to answer are:

Can distinct imaging phenotype subtypes be identified in unresectable HCC patients using radiomics and unsupervised clustering?

Do these different imaging subtypes show significant differences in treatment efficacy (such as objective response rate, progression-free survival, and overall survival) after receiving TACE or HAIC?

Can this method objectively identify which imaging subtype of patients is more suitable for TACE and which may benefit more from HAIC?

Participants in this study are adult patients diagnosed with unresectable HCC (BCLC stage B or C) who have already undergone complete TACE or HAIC treatment as part of their regular medical care between January 2015 and December 2024. Researchers will retrospectively analyze their existing clinical data and pre-treatment medical images to compare outcomes.

Inclusion Criteria:

1. Patients diagnosed with hepatocellular carcinoma (HCC) based on clinical or pathological criteria;
2. Barcelona Clinic Liver Cancer (BCLC) stage B or C, assessed as unresectable by surgical evaluation;
3. Received either TACE or HAIC as the sole interventional treatment modality throughout the treatment course (no crossover between the two modalities allowed);
4. Liver function classified as Child-Pugh Class A or B, or achieved this standard after medical treatment;
5. Performance status (PS) score of 0 or 1.

Exclusion Criteria:

1. Patients with an unclear diagnosis or those with concurrent other malignant tumors;
2. Liver function classified as Child-Pugh Class C or worse, which cannot be improved after hepatoprotective treatment;
3. Patients with severe infections, such as respiratory, biliary tract, or abdominal infections;
4. Patients with severe underlying diseases, particularly immune-related disorders.
5. Patients with severely incomplete or missing baseline or outcome data.