Prospective Validation of an AI Model for Predicting Liver Metastasis in Colorectal Cancer

Tongji Hospital160 enrolled

Overview

This is a prospective, multicenter, observational study designed to validate the predictive accuracy of a pre-developed multimodal deep learning model. The model integrates preoperative contrast-enhanced CT scans, digitized postoperative pathology images, and standard clinical data to estimate the risk of liver metastasis within two years after curative surgery in patients with stage I-III colorectal cancer. The primary objective is to evaluate the model's performance in an independent, prospectively enrolled patient cohort. Participants will receive standard-of-care treatment according to clinical guidelines. The study involves no experimental interventions; it solely involves the collection and analysis of routinely generated clinical data. The goal is to assess the model's potential for clinical translation by providing a reliable tool for stratifying patients' risk of liver metastasis, which could inform personalized surveillance strategies.

Study Type

OBSERVATIONAL

Enrollment

160

Conditions

Colorectal Cancer Liver Metastasis

Interventions

Multimodal Deep Learning Prediction ModelDIAGNOSTIC_TEST

This is a non-therapeutic, prognostic study. The intervention under investigation is the application of a pre-specified multimodal deep learning model that integrates preoperative CT imaging, digital pathology, and clinical data to stratify patients' risk of developing metachronous liver metastasis. This model functions as a prognostic tool and is not used to guide patient management in this study. Its performance is being evaluated prospectively against the actual clinical outcomes.

Eligibility

Sex: ALLMin age: 18 YearsMax age: 75 Years

Medical Language ↔ Plain English

Inclusion Criteria: * Age 18-75 years, any gender. * Clinical diagnosis of primary colon or rectal adenocarcinoma (Stage I-III). Scheduled to undergo curative radical resection for colorectal cancer. * Preoperative contrast-enhanced abdominal/pelvic CT scan performed within 1 month before surgery, with acceptable image quality. * No evidence of distant metastasis (including synchronous liver metastasis) on preoperative examination. * ECOG Performance Status of 0 or 1. * Patient or their legal representative voluntarily participates and provides written informed consent. Exclusion Criteria: * Postoperative pathological confirmation of non-primary colorectal adenocarcinoma or presence of distant metastasis. * Intraoperative determination of non-R0 resection, or performance of palliative surgery/ostomy only. * History of other malignant tumors. * Previous history of liver surgery or liver transplantation. * Death within the perioperative period (within 30 days after surgery). * Refusal to participate in follow-up, withdrawal of informed consent, or loss to follow-up.

Outcomes

Primary Outcomes

Area Under the Receiver Operating Characteristic Curve (AUC)

The discriminatory performance of the pre-specified multimodal deep learning model for predicting the occurrence of metachronous liver metastasis within 2 years after curative resection. The model integrates preoperative contrast-enhanced CT, digital pathology, and clinical data. Performance is evaluated on the entire prospectively enrolled validation cohort.

Time frame: 2 years after surgery

Secondary Outcomes

Liver Metastasis-Free Survival (LMFS) by Risk Group

The difference in liver metastasis-free survival between the high-risk and low-risk groups, as stratified by the model. LMFS is defined as the time from surgery to the first radiological diagnosis of liver metastasis.

Time frame: From the date of surgery until the date of first documented liver metastasis or last follow-up, assessed up to 3 years.