Description

Data source The study utilized the Chang Gung Research Database (CGRD), an extensive repository of electronic medical records maintained by Chang Gung Memorial Hospital (CGMH) across multiple medical centers and hospitals in Taiwan. Established for administrative and insurance purposes, the CGRD includes comprehensive data on patient demographics, diagnoses, treatments, and outcomes. It covers a significant portion of Taiwan's medical services, with detailed records on outpatient, emergency, and inpatient visits. Notably, the CGRD also includes a cancer registry with thorough documentation of diagnoses, staging, treatment modalities, recurrence, and mortality data. All data in cancer registry database were curated manually and make this information highly reliable1. Diagnoses are coded using the International Classification of Diseases, 9th Revision, Clinical Modification (ICD-9-CM) before 2016 and International Classification of Diseases, 10th Revision, Clinical Modification (ICD-10-CM) thereafter, with all cancer registry entries manually verified for accuracy. This makes the CGRD an invaluable resource for clinical research. The CGRD's extensive coverage of Taiwan's medical landscape, including a significant proportion of both inpatient and outpatient services (20% of all cancer patients' data and about 34% outpatient information in Taiwan), underscores its utility as an excellent data source for robust clinical research. Previous studies have elaborated on the CGRD's structure and content, reinforcing its role as a key resource for healthcare studies.

Study design and population The study initially included all incident HCC patients from January 2002 to December 2018 by using ICD-9-CM codes 155 and ICD-10-CM codes C22. To ensure the population to be analyzed, several exclusion criteria were applied. Patients who did not receive any treatment after diagnosis were excluded to avoid confounding survival outcomes. Age was another critical factor; patients younger than 18 or older than 80 years were excluded to eliminate potential age-related biases in survival analysis. Additionally, patients with missing tumor size were excluded, as tumor size is a crucial variable in assessing prognosis and treatment decisions. Errors in survival or recurrence status or date were also excluded to maintain data integrity. Lastly, patients deemed non-operable were excluded to focus on those who could potentially benefit from surgical intervention. Furthermore, patients who did not fall into the pT1N0M0 or pT2N0M0 staging categories were excluded to ensure only early-stage patients were enrolled. The remaining patients were then divided into two groups based on their stage, in which subjects would be separated by TNM staging pT1N0M0 (solitary HCC) and pT2N0M0 (solitary HCC with mVi or multiple HCCs with none size < 5 cm). Within the pT1N0M0 group, patients who died within 30 days post-surgery were excluded to avoid perioperative mortality bias. Additionally, patients with a tumor size of ≤ 2cm were excluded to focus on more clinically significant tumors. Two subgroups, subjects with tumor size > 2cm and ≤ 6.5cm and > 6.5cm, were further obtained according to our previous report that 6.5 cm would be the cut-off value of pT1 subjects with different prognosis. For the pT2N0M0 group, similar exclusions were made for patients who died within 30 days post-surgery and for errors in the Child score. This group was further categorized into solitary tumor with microvascular invasion (mVI) and multiple tumors with none smaller than 5cm. Also, tumor size 6.5 cm was used to classify subjects with solitary tumor with mVi. The outcomes were compared among different groups of enrolled subjects to realize the associations between operation outcomes and HCC with mVi and tumor size over or below 6.5 cm among solitary HCC patients. This study, approved by the Institutional Review Board of the Chang Gung Medical Foundation, leverages the CGRD's rich data repository to conduct a retrospective cohort analysis (IRB No.: 202300142B0).