Hepatocellular carcinoma (HCC) is a dreaded complication for patients with chronic liver disease (CLD) of different etiologies, including chronic viral hepatitis, metabolic dysfunction associated steatotic liver disease (MASLD) (formerly known as nonalcoholic fatty liver disease [NAFLD] and alcohol related liver disease [ArLD]). HCC remains among the top five causes of cancer deaths across different geographic regions. In 2020, more than 900,000 people were diagnosed with, and 830,000 people died from the disease globally. The figure is predicted to further increase in the coming decade [
1]. The high mortality is contributed by the suboptimal execution of surveillance program and failure to diagnose the cancer at an early, curable stage. Despite the recent advancement in treatment modalities, the survival remains poor in those with advanced, unresectable tumour [
2]. Chronic hepatitis B virus (HBV) and hepatitis C virus (HCV) infection used to account for the majority of HCC cases. The prevalence of viral hepatitis-related HCC is gradually decreasing in response to efforts in augmenting diagnosis and the availability of effective antiviral medication. Non-viral causes, in particularly MASLD, is becoming the most common form of CLD.
Surveillance of HCC with bi-annual liver ultrasound scan in at-risk population is an important secondary prevention strategy. However, recommendation from current guidelines mainly focus on patients with cirrhosis or viral hepatitis and additional risk factors [
3]. Based on cost-effectiveness analyses, HCC surveillance should be performed in individuals with HCC risk exceeding 1.0–1.5% per year [
4]. The annual incidence rates of HCC in MASLD related cirrhosis range between 0.2–2.6%, while the incidence in MASLD individuals without cirrhosis is lower than the threshold to justify routine HCC surveillance [
5]. Moreover, the performance of ultrasound is suboptimal in the MASLD population due to several factors, including obesity impairing visualization of the liver, heterogeneous echogenicity in fatty liver and expertise of the operators in assessing patients with MASLD. MRI had been shown in prospective study to have higher sensitivity and specificity than ultrasound for early tumour detection and is proposed to be a better alternative for HCC surveillance in MASLD [
6]. However, the capacity for MRI is limited for most of the health care system and the cost is likely prohibitive for its implementation as routine surveillance modality. Therefore, additional risk stratification strategy is required to identify suitable patients with CLD for HCC surveillance.
Risk factors associated with HCC are diverse. For MASLD, older age, male sex, cirrhosis, presence of obesity and diabetes, and genetic polymorphism (e.g., PNPLA3, TM6SF2) confer increased risk of HCC. Apart from cirrhosis, none of the listed risk factors is currently considered in the surveillance recommendation for MASLD because of the scarcity of data. The use of risk score, which utilizes a range of clinical parameters and biomarkers, can be used to predict development of disease via risk stratification. Risk scores for HCC prediction were first developed in hepatitis B virus-related HCC e.g., GAG-HCC, CU-HCC and PAGE-B scores. Development of risk scores starts with identification of independent factors associated with HCC in a training or derivation cohort with longitudinal follow-up. Weighting is assigned to individual factors based on multivariable analysis, which are then combined to formulate a prediction score. The risk score is then applied to another validation cohort to test the performance of the score [
7]. There is a lack of HCC risk prediction scores for other etiologies of chronic liver disease.
In this issue of
Clinical and Molecular Hepatology, Su et al. demonstrated the use of Steatosis-Associated Fibrosis Estimator (SAFE) score to predict HCC across various types of liver diseases including MASLD. The SAFE score consists of commonly available parameters, including age, body mass index, presence of diabetes, aspartate aminotransferase (AST), alanine aminotransferase (ALT), globulin and platelet count [
8]. In the original study by Sripongpun et al. [
9], SAFE score was constructed based on logistic regression and machine-learning methods with the aim to distinguish significant fibrosis i.e., ≥F2 from F0/1 in subjects with MASLD in the primary care setting. Notably, the training and testing cohort in the study included biopsy-proven MASLD patients while patients with other liver diseases are excluded. The retrospective cohort in the current paper consists of 12,963 patients with different liver diseases based largely on diagnostic coding. The disease breakdown is HBV (n=5,449), HCV (n=1,819), HBV/HCV (n=433), MASLD (n=2,958), NAFLD (n=1,575), ArLD (n=324), other liver cirrhosis (LC) (n=433), and other CLD (n=2,930) groups. There is likely overlapping between MASLD and NAFLD, and ArLD subgroups. 258 patients (2%) developed HCC, the overall 5- and 10- year cumulative incidence is 2.2% and 3.6%. The 5-year cumulative incidence of HCC by disease group is HBV (2.3%), HCV (5.4%), HBV/HCV (2.9%), MASLD (0.6%), NAFLD (0.3%), ArLD (4.1%), other LC (3.1%) and other CLD (0.06%) groups.
The authors found the SAFE score classifies 1-, 3-, and 5-year HCC risk regardless of CLD etiologies. A high SAFE score (≥100) increased the risks of HCC in subgroups of viral hepatitis, non-viral hepatitis (aSHR 11.10, 95% CI 3.97–31.30) and MASLD (aSHR 4.23, 95% CI 1.43–12.50). Validation in external hospital cohort and community MASLD cohort confirmed the performance of the SAFE score. They concluded that the SAFE score stratifies high risks for HCC in CLD patients regardless of etiologies and helps to select at-risk candidates for HCC surveillance. The large patient cohort with over 30% patients with MASLD or NAFLD is a strength of the study. The association of high SAFE score and risk of HCC is at least partially explained by the presence of significant liver fibrosis as the parameters adopted in SAFE score are included in other liver fibrosis indices such as the APRI (AST, platelet) and FIB-4 (age, AST, ALT, platelet) scores. The authors demonstrated satisfactory time-dependent AUROCs using the SAFE score to classify HCC risk which were similar across disease subgroups. Notably, the performance of SAFE score was better than FIB-4 index in classifying the MASLD or NAFLD related HCC. Among treated and untreated HBV subjects, which makes up the largest proportion of the study cohort, the SAFE score has comparable AUROC as existing HBV-HCC scores (REACH-B, CU-HCC, and PAGE-B).
How can the SAFE score add to better HCC risk stratification in patients with viral and non-viral liver disease etiologies? The SAFE score, with readily available patient parameters, can be built into the clinical computer system and aid clinicians to select patients at risk of significant fibrosis and HCC, especially among MASLD patients, at minimal additional costs. In real life practice using HBV as an example, those determined to be at high or intermediate risk after assessment with risk score should be included in regular HCC surveillance program. Others deemed to be at low risk have minimal risk of HCC and can have lengthened interval for ultrasound assessment. Future prospective study is required to assess the performance of the SAFE score in both primary care and hospital setting, where the case mix of liver diseases and incidence of HCC vary significantly. Furthermore, comparison or incorporation with other risk stratification strategies e.g., GALAD score [
10,
11], GAAD sore, liver stiffness measurement, genetic polymorphism is required before universal adoption of the SAFE score into routine clinical pathway.
FOOTNOTES
-
Authors’ contribution
Michael Kwan-Lung Ko: Original Draft Preparation, Review & Editing. Loey Lung-Yi Mak: Review & Editing.
-
Conflicts of Interest
Michael Kwan-Lung Ko has no conflicts of interest. Mak LY is an associate editor of the journal, and received research funding from Echosens, Gilead Sciences, and Roche Diagnostics Hong Kong Limited.
Abbreviations
alcohol related liver disease
aspartate aminotransferase
metabolic dysfunction associated steatotic liver disease
non-alcoholic fatty liver disease
Steatosis-Associated Fibrosis Estimator
REFERENCES
- 1. Rumgay H, Arnold M, Ferlay J, Lesi O, Cabasag CJ, Vignat J, et al. Global burden of primary liver cancer in 2020 and predictions to 2040. J Hepatol 2022;77:1598-1606.
- 2. Huang DQ, Singal AG, Kanwal F, Lampertico P, Buti M, Sirlin CB, et al. Hepatocellular carcinoma surveillance - utilization, barriers and the impact of changing aetiology. Nat Rev Gastroenterol Hepatol 2023;20:797-809.
- 3. Singal AG, Llovet JM, Yarchoan M, Mehta N, Heimbach JK, Dawson LA, et al. AASLD Practice Guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma. Hepatology 2023;78:1922-1965.
- 4. Parikh ND, Singal AG, Hutton DW, Tapper EB. Cost-effectiveness of hepatocellular carcinoma surveillance: An assessment of benefits and harms. Am J Gastroenterol 2020;115:1642-1649.
- 5. Orci LA, Sanduzzi-Zamparelli M, Caballol B, Sapena V, Colucci N, Torres F, et al. Incidence of hepatocellular carcinoma in patients with nonalcoholic fatty liver disease: A systematic review, meta-analysis, and meta-regression. Clin Gastroenterol Hepatol 2022;20:283-292.e10.
- 6. Kim SY, An J, Lim YS, Han S, Lee JY, Byun JH, et al. MRI with liver-specific contrast for surveillance of patients with cirrhosis at high risk of hepatocellular carcinoma. JAMA Oncol 2017;3:456-463.
- 7. Wong VW, Janssen HL. Can we use HCC risk scores to individualize surveillance in chronic hepatitis B infection? J Hepatol 2015;63:722-732.
- 8. Su TH, Yang SS, Lee MH, Kao WY, Huang SC, Chen FF, et al. High Steatosis-Associated Fibrosis Estimator scores predict hepatocellular carcinoma in viral and non-viral hepatitis and metabolic dysfunction-associated steatotic liver disease. Clin Mol Hepatol 2025;31:796-809.
- 9. Sripongpun P, Kim WR, Mannalithara A, Charu V, Vidovszky A, Asch S, et al. The steatosis-associated fibrosis estimator (SAFE) score: A tool to detect low-risk NAFLD in primary care. Hepatology 2023;77:256-267.
- 10. Marsh TL, Parikh ND, Roberts LR, Schwartz ME, Nguyen MH, Befeler A, et al. A phase 3 biomarker validation of GALAD for the detection of hepatocellular carcinoma in cirrhosis. Gastroenterology 2025;168:316-326.e6.
- 11. Mak LY, Yuen MF. Surveillance for hepatocellular carcinoma in cirrhosis: End of monopoly for serum alpha fetoprotein. Gastroenterology 2025;168:217-219.
Citations
Citations to this article as recorded by
- HCC predictors in routine practice for patients with chronic liver diseases: Correspondence to editorial on “High SAFE scores predict hepatocellular carcinoma in viral and non-viral hepatitis and metabolic dysfunction associated steatotic liver disease”
Tung-Hung Su, Jia-Horng Kao
Clinical and Molecular Hepatology.2026; 32(1): e52. CrossRef
