Development and Validation of a Nomogram to Detect Hepatocellular Carcinoma in Patients with Chronic Hepatitis B: A Prospective, Multicenter Study

Abstract

Background More than 50% of hepatocellular carcinoma (HCC) are caused by hepatitis B virus (HBV) infection and almost one third of carriers of hepatitis B virus (HBV) world-wide are in China. Early detection of HCC in HBV-infected patients remains a challenge. Methods Consecutive participants with HBV-related HCC, chronic hepatitis B (CHB), liver cirrhosis with HBV infection (HBV-LC), benign hepatic tumors and healthy controls at 11 hospitals in China from June 2016 to April 2017 were recruited to form a training cohort. Participants from the same hospitals from May 2017 to September 2017 were recruited to form a validation cohort. All HCCs were diagnosed by ultrasound, characteristic radiologic features on CT/MRI, and confirmed by histopathological examination from biopsy or surgically resected specimens. Findings 908 of 2207 (41.1%) participants in the training cohort and 289 of 730 (39.6%) participants in the validation cohort were diagnosed to have HBV-related HCC. The serum levels of PIVKA-II and AFP were significantly higher in patients with HBV-related HCC than in those with CHB, HBV-LC or in controls (P<0.0001). A combination of PIVKA-II and AFP showed improved diagnostic accuracy in differentiating patients with HBV-related HCC from those with CHB and HBV-LC (area under the curve [AUC] 0.922 [95% CI 0.908-0.935], sensitivity 88.3% and specificity 85.1% for the training cohort; the corresponding figures were 0.902 [0.875, 0.929], 87.8% and 81.0%) respectively for the validation cohort. The variables which were included in the nomogram were AFP, PIVKA-II, age and sex. The nomogram performed well in predicting HBV-related HCC in calibration and in discrimination in the training cohort (AUC, 0.941 [0.929, 0.952]); as well as in the validation cohort (0.931 [0.909, 0.953]). The nomogram also performed well in detecting early-stage HBV-related HCC. Interpretation A nomogram and an online tool, which included AFP, PIVKA-II, age and sex in predicting the presence of HBV-related HCC were built and validated with good discrimination and calibration. This nomogram can help clinicians to detect early HCC in patients with HBV infection. Trial Registration Number: The study was registered with ClinicalTrials.gov, number NCT03047603. Funding Abbott Diagnostics (ADD-China-2016), and in part by National Natural Science Foundation of China (No. 81472284 and 81672699). Declaration of Interest: None. Ethical Approval: The study was approved by the Institutional Review Boards at each study center. Informed consent was obtained from each participant.