Mengdi Cao , Maomao Cao , Changfa Xia , Fan Yang , Xinxin Yan , Siyi He , Shaoli Zhang , Yi Teng , Qianru Li , Nuopei Tan , Jiachen Wang , Chunfeng Qu , Wanqing Chen
{"title":"基于评分的女性隐性乙肝表面抗原携带者肝细胞癌监测预测模型:中国多中心队列研究","authors":"Mengdi Cao , Maomao Cao , Changfa Xia , Fan Yang , Xinxin Yan , Siyi He , Shaoli Zhang , Yi Teng , Qianru Li , Nuopei Tan , Jiachen Wang , Chunfeng Qu , Wanqing Chen","doi":"10.1016/j.jncc.2025.05.005","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><div>Existing hepatocellular carcinoma (HCC) prediction models lack transferability and generalizability when applied to female populations, resulting in diminished performance and inadequate tools for accurate HCC risk stratification among females. This study aims to develop and validate a score-based prediction model for early detection of HCC in female hepatitis B surface antigen (HBsAg) carriers.</div></div><div><h3>Methods</h3><div>Participants were recruited from a multicenter prospective cohort engaged in liver cancer screening across China including seven high-risk rural areas and one additional high-risk rural area. The study involved 7080 females as the derivation cohort and 2069 as the validation cohort, with all participants aged 35–70 years and HBsAg positive. Laboratory tests and epidemiological surveys were conducted. Key predictor variables were identified through LASSO regression analysis, and score-based prediction models were developed based on Cox proportional hazards model. Model performance including discrimination and calibration was evaluated, and compared to existing prediction models and screening strategies.</div></div><div><h3>Results</h3><div>After a median follow-up of 3.69 and 5.42 years, 147 and 45 HCC cases were identified in the derivation and validation cohorts, respectively. The female HCC (HCCF) model incorporating five independent variables: age, α-fetoprotein (AFP), albumin, alanine aminotransferase, and platelet, showed excellent performance with an area under the receiver operating characteristic curve (AUC) of 0.82 (95 % CI: 0.78–0.86). The HCCF-Enhanced model which included cirrhosis, achieved an AUC of 0.85 (95 % CI: 0.81–0.89). Both models demonstrated superior predictive performance than existing models, with strong predictive accuracy in the validation cohort: AUCs of 0.83 (95 % CI: 0.77–0.89) and 0.88 (95 % CI: 0.83–0.92), respectively. The HCCF model, at a score threshold of 7, achieved the largest Youden’s index and identified 32.80 % of high-risk individuals. When combined with ultrasonography (US), the model detected 37 additional cases, significantly improved screening sensitivity and accuracy compared to the traditional AFP plus US strategy.</div></div><div><h3>Conclusions</h3><div>The developed HCCF models with good performance for HCC prediction in HBsAg-positive females significantly improve screening efficiency and provide an effective tool for surveillance, ultimately helping to optimize prevention and management strategies for HCC.</div></div>","PeriodicalId":73987,"journal":{"name":"Journal of the National Cancer Center","volume":"5 5","pages":"Pages 493-500"},"PeriodicalIF":9.4000,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"Score-based prediction model for female hepatocellular carcinoma surveillance in asymptotic HBsAg carriers: a multicenter cohort study in China\",\"authors\":\"Mengdi Cao , Maomao Cao , Changfa Xia , Fan Yang , Xinxin Yan , Siyi He , Shaoli Zhang , Yi Teng , Qianru Li , Nuopei Tan , Jiachen Wang , Chunfeng Qu , Wanqing Chen\",\"doi\":\"10.1016/j.jncc.2025.05.005\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><h3>Background</h3><div>Existing hepatocellular carcinoma (HCC) prediction models lack transferability and generalizability when applied to female populations, resulting in diminished performance and inadequate tools for accurate HCC risk stratification among females. This study aims to develop and validate a score-based prediction model for early detection of HCC in female hepatitis B surface antigen (HBsAg) carriers.</div></div><div><h3>Methods</h3><div>Participants were recruited from a multicenter prospective cohort engaged in liver cancer screening across China including seven high-risk rural areas and one additional high-risk rural area. The study involved 7080 females as the derivation cohort and 2069 as the validation cohort, with all participants aged 35–70 years and HBsAg positive. Laboratory tests and epidemiological surveys were conducted. Key predictor variables were identified through LASSO regression analysis, and score-based prediction models were developed based on Cox proportional hazards model. Model performance including discrimination and calibration was evaluated, and compared to existing prediction models and screening strategies.</div></div><div><h3>Results</h3><div>After a median follow-up of 3.69 and 5.42 years, 147 and 45 HCC cases were identified in the derivation and validation cohorts, respectively. The female HCC (HCCF) model incorporating five independent variables: age, α-fetoprotein (AFP), albumin, alanine aminotransferase, and platelet, showed excellent performance with an area under the receiver operating characteristic curve (AUC) of 0.82 (95 % CI: 0.78–0.86). The HCCF-Enhanced model which included cirrhosis, achieved an AUC of 0.85 (95 % CI: 0.81–0.89). Both models demonstrated superior predictive performance than existing models, with strong predictive accuracy in the validation cohort: AUCs of 0.83 (95 % CI: 0.77–0.89) and 0.88 (95 % CI: 0.83–0.92), respectively. The HCCF model, at a score threshold of 7, achieved the largest Youden’s index and identified 32.80 % of high-risk individuals. When combined with ultrasonography (US), the model detected 37 additional cases, significantly improved screening sensitivity and accuracy compared to the traditional AFP plus US strategy.</div></div><div><h3>Conclusions</h3><div>The developed HCCF models with good performance for HCC prediction in HBsAg-positive females significantly improve screening efficiency and provide an effective tool for surveillance, ultimately helping to optimize prevention and management strategies for HCC.</div></div>\",\"PeriodicalId\":73987,\"journal\":{\"name\":\"Journal of the National Cancer Center\",\"volume\":\"5 5\",\"pages\":\"Pages 493-500\"},\"PeriodicalIF\":9.4000,\"publicationDate\":\"2025-10-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Journal of the National Cancer Center\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S2667005425000845\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"ONCOLOGY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Journal of the National Cancer Center","FirstCategoryId":"1085","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2667005425000845","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"ONCOLOGY","Score":null,"Total":0}
Score-based prediction model for female hepatocellular carcinoma surveillance in asymptotic HBsAg carriers: a multicenter cohort study in China
Background
Existing hepatocellular carcinoma (HCC) prediction models lack transferability and generalizability when applied to female populations, resulting in diminished performance and inadequate tools for accurate HCC risk stratification among females. This study aims to develop and validate a score-based prediction model for early detection of HCC in female hepatitis B surface antigen (HBsAg) carriers.
Methods
Participants were recruited from a multicenter prospective cohort engaged in liver cancer screening across China including seven high-risk rural areas and one additional high-risk rural area. The study involved 7080 females as the derivation cohort and 2069 as the validation cohort, with all participants aged 35–70 years and HBsAg positive. Laboratory tests and epidemiological surveys were conducted. Key predictor variables were identified through LASSO regression analysis, and score-based prediction models were developed based on Cox proportional hazards model. Model performance including discrimination and calibration was evaluated, and compared to existing prediction models and screening strategies.
Results
After a median follow-up of 3.69 and 5.42 years, 147 and 45 HCC cases were identified in the derivation and validation cohorts, respectively. The female HCC (HCCF) model incorporating five independent variables: age, α-fetoprotein (AFP), albumin, alanine aminotransferase, and platelet, showed excellent performance with an area under the receiver operating characteristic curve (AUC) of 0.82 (95 % CI: 0.78–0.86). The HCCF-Enhanced model which included cirrhosis, achieved an AUC of 0.85 (95 % CI: 0.81–0.89). Both models demonstrated superior predictive performance than existing models, with strong predictive accuracy in the validation cohort: AUCs of 0.83 (95 % CI: 0.77–0.89) and 0.88 (95 % CI: 0.83–0.92), respectively. The HCCF model, at a score threshold of 7, achieved the largest Youden’s index and identified 32.80 % of high-risk individuals. When combined with ultrasonography (US), the model detected 37 additional cases, significantly improved screening sensitivity and accuracy compared to the traditional AFP plus US strategy.
Conclusions
The developed HCCF models with good performance for HCC prediction in HBsAg-positive females significantly improve screening efficiency and provide an effective tool for surveillance, ultimately helping to optimize prevention and management strategies for HCC.