Ignite Creation Date:
2026-03-26 @ 3:20 PM
Ignite Modification Date:
2026-03-31 @ 5:17 AM
Study NCT ID:
NCT07444905
Status:
ACTIVE_NOT_RECRUITING
Last Update Posted:
2026-03-03
First Post:
2026-02-24
Is NOT Gene Therapy:
True
Has Adverse Events:
False
Brief Title:
Evaluation of a Machine Learning-Based Prediction Strategy for Extrahepatic Metastasis in Hepatocellular Carcinoma
Sponsor:
Tongji Hospital
Organization:
Study Overview
Official Title:
Evaluating a Machine Learning-Based Strategy for Predicting Extrahepatic Metastasis and Guiding Risk-Adapted Surveillance in Hepatocellular Carcinoma
Status:
ACTIVE_NOT_RECRUITING
Status Verified Date:
2026-03
Last Known Status:
None
Delayed Posting:
No
If Stopped, Why?:
Not Stopped
Has Expanded Access:
False
If Expanded Access, NCT#:
N/A
Has Expanded Access, NCT# Status:
N/A
Acronym:
None
Brief Summary:
This is a multicenter prospective observational cohort study in patients with hepatocellular carcinoma (HCC) after curative-intent treatment. The study aims to prospectively validate previously developed machine learning-based risk stratification models for extrahepatic metastasis (with a focus on lung and bone metastasis) and to evaluate their potential clinical utility in real-world postoperative surveillance and management pathways.
The study does not assign treatments or surveillance strategies to participants. Clinical care is determined by treating physicians according to local practice. The study will assess model performance (including discrimination and calibration), risk stratification ability, and implementation-related outcomes such as model adoption, decision impact, and changes in monitoring intensity or referral pathways. The study will also explore clinical and resource-related outcomes associated with model-informed risk stratification in routine practice.
The study does not assign treatments or surveillance strategies to participants. Clinical care is determined by treating physicians according to local practice. The study will assess model performance (including discrimination and calibration), risk stratification ability, and implementation-related outcomes such as model adoption, decision impact, and changes in monitoring intensity or referral pathways. The study will also explore clinical and resource-related outcomes associated with model-informed risk stratification in routine practice.
Detailed Description:
Background and Rationale
Hepatocellular carcinoma (HCC) is associated with a substantial risk of postoperative recurrence and extrahepatic metastasis, which can significantly affect prognosis and treatment opportunities. Early identification of patients at high risk of extrahepatic metastasis may support more appropriate surveillance intensity, earlier multidisciplinary evaluation, and more timely treatment planning. However, risk-stratified postoperative surveillance strategies for extrahepatic metastasis are not well established in routine clinical practice.
The investigators previously developed machine learning-based risk stratification models for extrahepatic metastasis in retrospective multicenter cohorts, with particular focus on lung and bone metastasis. The present study is designed to prospectively validate these models and evaluate their potential value in real-world clinical workflow and decision-making.
Study Objectives Primary Objective
To prospectively validate the performance of pre-specified machine learning-based risk stratification models for postoperative extrahepatic metastasis risk in HCC (especially lung and bone metastasis), including discrimination and calibration in a multicenter real-world setting.
Secondary Objectives
To evaluate risk stratification performance across clinically relevant subgroups and participating centers.To assess implementation-related outcomes in routine practice, including model adoption and clinical decision impact.To evaluate whether model-informed risk stratification is associated with changes in postoperative surveillance patterns, referral pathways, and timing of multidisciplinary assessment.To explore clinical outcome signals (for example, time-to-event outcomes and clinically actionable detection window-related outcomes) associated with model-informed risk stratification.To explore resource utilization and health economic implications of model-informed postoperative management in real-world settings.
Study Design
This is a multicenter, prospective, observational cohort study. Participants are enrolled and followed according to routine clinical care at participating hospitals. This study does not involve randomization, mandated intervention assignment, or protocol-driven treatment allocation. The study is intended to validate and evaluate a risk stratification tool and its real-world implementation, rather than to test an interventional treatment.
The risk models, variables, and prespecified risk stratification framework were developed prior to initiation of prospective enrollment. Prospective data collection is used to evaluate model transportability, calibration, clinical utility, and implementation characteristics in independent real-world cohorts.
Study Population
Eligible participants are adults with hepatocellular carcinoma who have undergone curative-intent treatment and are entering postoperative follow-up. Detailed inclusion and exclusion criteria are provided in the protocol and include availability of required baseline clinical variables and follow-up information for outcome assessment.
Study Procedures and Data Collection
Data are collected prospectively during routine care and may include:
Baseline demographic and clinical characteristics；Tumor-related and treatment-related variables；Follow-up imaging and laboratory surveillance information；Occurrence and timing of extrahepatic metastasis (especially lung and bone metastasis)
Recurrence and survival-related outcomes, where available
Clinical management decisions (for example, surveillance intensity, referrals, multidisciplinary team discussion, and treatment planning)
Implementation metrics (for example, model use/adoption and decision impact documentation, where available)
Resource utilization variables for exploratory economic analyses, where available
No study-mandated treatment is assigned. Clinical decisions remain under the responsibility of treating physicians.
Outcomes and Analytic Framework
The study will evaluate model performance using prespecified statistical methods, which may include measures of discrimination, calibration, and clinical utility for time-to-event outcomes. Analyses may also assess center-level heterogeneity, temporal performance, and model updating/recalibration strategies where appropriate.
Implementation and clinical utility analyses may include evaluation of:
Adoption and use of the risk stratification tool
Changes in surveillance intensity or follow-up pathways associated with risk strata
Decision impact on referral and management planning
Exploratory associations with clinically meaningful outcomes and resource use
Any analyses intended to emulate hypothetical management strategies (for example, target trial emulation-based analyses) will be clearly reported as observational and estimated/hypothetical, and will not be interpreted as randomized treatment effects.
Study Significance
This study is expected to provide prospective multicenter evidence on the validity and real-world clinical utility of machine learning-based risk stratification for extrahepatic metastasis in HCC. The results may inform postoperative surveillance optimization, risk-adapted management pathways, and future implementation research.
Hepatocellular carcinoma (HCC) is associated with a substantial risk of postoperative recurrence and extrahepatic metastasis, which can significantly affect prognosis and treatment opportunities. Early identification of patients at high risk of extrahepatic metastasis may support more appropriate surveillance intensity, earlier multidisciplinary evaluation, and more timely treatment planning. However, risk-stratified postoperative surveillance strategies for extrahepatic metastasis are not well established in routine clinical practice.
The investigators previously developed machine learning-based risk stratification models for extrahepatic metastasis in retrospective multicenter cohorts, with particular focus on lung and bone metastasis. The present study is designed to prospectively validate these models and evaluate their potential value in real-world clinical workflow and decision-making.
Study Objectives Primary Objective
To prospectively validate the performance of pre-specified machine learning-based risk stratification models for postoperative extrahepatic metastasis risk in HCC (especially lung and bone metastasis), including discrimination and calibration in a multicenter real-world setting.
Secondary Objectives
To evaluate risk stratification performance across clinically relevant subgroups and participating centers.To assess implementation-related outcomes in routine practice, including model adoption and clinical decision impact.To evaluate whether model-informed risk stratification is associated with changes in postoperative surveillance patterns, referral pathways, and timing of multidisciplinary assessment.To explore clinical outcome signals (for example, time-to-event outcomes and clinically actionable detection window-related outcomes) associated with model-informed risk stratification.To explore resource utilization and health economic implications of model-informed postoperative management in real-world settings.
Study Design
This is a multicenter, prospective, observational cohort study. Participants are enrolled and followed according to routine clinical care at participating hospitals. This study does not involve randomization, mandated intervention assignment, or protocol-driven treatment allocation. The study is intended to validate and evaluate a risk stratification tool and its real-world implementation, rather than to test an interventional treatment.
The risk models, variables, and prespecified risk stratification framework were developed prior to initiation of prospective enrollment. Prospective data collection is used to evaluate model transportability, calibration, clinical utility, and implementation characteristics in independent real-world cohorts.
Study Population
Eligible participants are adults with hepatocellular carcinoma who have undergone curative-intent treatment and are entering postoperative follow-up. Detailed inclusion and exclusion criteria are provided in the protocol and include availability of required baseline clinical variables and follow-up information for outcome assessment.
Study Procedures and Data Collection
Data are collected prospectively during routine care and may include:
Baseline demographic and clinical characteristics；Tumor-related and treatment-related variables；Follow-up imaging and laboratory surveillance information；Occurrence and timing of extrahepatic metastasis (especially lung and bone metastasis)
Recurrence and survival-related outcomes, where available
Clinical management decisions (for example, surveillance intensity, referrals, multidisciplinary team discussion, and treatment planning)
Implementation metrics (for example, model use/adoption and decision impact documentation, where available)
Resource utilization variables for exploratory economic analyses, where available
No study-mandated treatment is assigned. Clinical decisions remain under the responsibility of treating physicians.
Outcomes and Analytic Framework
The study will evaluate model performance using prespecified statistical methods, which may include measures of discrimination, calibration, and clinical utility for time-to-event outcomes. Analyses may also assess center-level heterogeneity, temporal performance, and model updating/recalibration strategies where appropriate.
Implementation and clinical utility analyses may include evaluation of:
Adoption and use of the risk stratification tool
Changes in surveillance intensity or follow-up pathways associated with risk strata
Decision impact on referral and management planning
Exploratory associations with clinically meaningful outcomes and resource use
Any analyses intended to emulate hypothetical management strategies (for example, target trial emulation-based analyses) will be clearly reported as observational and estimated/hypothetical, and will not be interpreted as randomized treatment effects.
Study Significance
This study is expected to provide prospective multicenter evidence on the validity and real-world clinical utility of machine learning-based risk stratification for extrahepatic metastasis in HCC. The results may inform postoperative surveillance optimization, risk-adapted management pathways, and future implementation research.
Study Oversight
Has Oversight DMC:
True
Is a FDA Regulated Drug?:
False
Is a FDA Regulated Device?:
False
Is an Unapproved Device?:
None
Is a PPSD?:
None
Is a US Export?:
None
Is an FDA AA801 Violation?: