This study focused on exploring new comprehensive treatment strategies for patients with unresectable combined hepatocellular-cholangiocarcinoma, classifying patients with CHC subtypes based on the combination of artificial intelligence and multi-omics, and exploring the optimal treatment strategies for patients with different subtypes, helping clinicians to screen the most beneficial groups of various treatment schemes, and providing new ideas for safe treatment of high-risk patients.

This will be a multicenter observational study that will be conducted at several leading liver cancer treatment centers. The study will include adult patients with histologically/cytologically confirmed unresectable CHC. Collect patient genomics, proteomics, immune microenvironment, pathology reports, medical images and clinical electronic medical records, etc., to form high-quality and deeply labeled data sets to support the subsequent development and application of AI large models. Based on multi-source heterogeneous data of CHC patients, a large model for comprehensive diagnosis and treatment was constructed. Firstly, multi-modal data of different stages of disease were integrated by using cross-modal multi-course fusion technology to achieve efficient fusion of complex data. Secondly, by fine-tuning the large model, tasks such as CHC classification, prognosis inference and treatment plan recommendation are accurately completed, and potential information in the diagnosis and treatment process is mined.

Inclusion Criteria:

1. age ≥18 years; diagnosis of CHC confirmed by histology or cytology;
2. patients with unresectable or metastatic CHC diagnosed on the basis of unresectable CHC who have received prior local therapy, systemic therapy, or a combination of both and have at least one measurable lesion (RECIST v1.1);
3. survival time ≥ 3 months;
4. ECOG PS 0-2;
5. Child-Pugh A/B.

Exclusion Criteria:

1. pregnant women, lactating women, and men and women of childbearing age who are unwilling or unable to use effective contraception.
2. history of other malignant tumors within the past five years, unless these tumors have been completely treated and have been free of active disease for five years prior to the first dose and are at low risk of recurrence.
3. fully treated carcinoma in situ with no evidence of disease.
4. history of gastrointestinal bleeding or significant bleeding tendency (e.g., with known active ulcers, fecal occult blood, etc.) within the past six months that precludes inclusion in the study; gastroscopy is required if there is persistent fecal occult blood.
5. substantial organ transplantation or bone marrow transplantation within two years prior to the first dose, or active autoimmune disease requiring systemic therapy.
6. other conditions that the investigator deems unsuitable for inclusion in the study. Inadequate information, such as incomplete data from laboratory tests, missing or poor quality imaging data, no prognostic information, etc., that the investigator considers unsuitable for inclusion in the study.

The target population for the planned retrospective analysis of this project is patients with pathologically confirmed mixed hepatocellular cholangiocarcinoma, based on the 2024 NCCN Clinical Practice Guidelines in Oncology: Hepatocellular Carcinoma [Version 2.2024]) and China's Primary Liver Cancer Diagnosis and Treatment Guidelines (2024) are defined as unresectable patients.