Dear Editor,
We appreciate Dr. Satapathy and colleagues’ interest in our paper with their thoughtful comments [
1,
2]. We agree that the risks of hepatocellular carcinoma (HCC) in patients with Steatosis-Associated Fibrosis Estimator (SAFE) scores ≥100 vary across different etiologies of chronic liver diseases (CLDs). We did not rely on relative hazards alone; rather, we reported absolute, etiology-specific cumulative incidence and annual HCC rates, and demonstrated risk heterogeneity across etiologies in the high SAFE score strata. For example, metabolic dysfunction-associated steatotic liver disease (MASLD) with SAFE ≥100 had an HCC incidence of 0.3%/year, compared with 1.8%/year in hepatitis C virus (HCV) with SAFE ≥100. We also noted that “Even within the same high SAFE score category, the risk of HCC still varied by different CLD etiologies.” We propose using the SAFE score to enrich for higher-risk patients and to guide consideration of HCC surveillance when the etiology-specific absolute incidence exceeds cost-effectiveness thresholds and local practice standards [
3]. Therefore, we suggest that a SAFE score ≥100 be applied only in hepatitis B virus (HBV), HCV, HBV/HCV, cirrhosis, MASLD, and alcohol-related liver disease (ALD). At the same time, we do not recommend its use in NAFLD or other etiologies of CLDs, such as autoimmune hepatitis, where the incidence of HCC falls below the recommended surveillance threshold.
We agree that positive predictive value is limited by the low incidence of HCC in mixed-etiology CLD populations. However, the high negative predictive value makes it a good marker to identify patients with a low risk of HCC who do not need to undergo surveillance. While the low positive predictive value (PPV) indicates some false-positive predictions, further refinement of the SAFE score to include sex and albumin, based on our multivariable analyses, may be pursued in the future. SAFE score may be combined with etiology-specific factors to ensure the absolute risk exceeds the surveillance threshold before recommending routine monitoring in particular scenarios. The threshold for initiating HCC surveillance varies by region and etiology; for instance, surveillance may be recommended in non-cirrhotic HBV patients with an annual risk ≥0.2% [
4]. Cost-effectiveness analyses may re-evaluate the surveillance threshold for MASLD.
In addition, a higher threshold of SAFE score (e.g., 129, as demonstrated by the highest Youden index) may be adopted to improve PPV when resources are constrained. In clinical practice, serial measurement of SAFE scores, adoption of a higher SAFE score threshold (given the 6% increased risk of HCC per 10-point increase in SAFE score), and the use of simple HCC surveillance tools (e.g., blood-based markers) may be helpful.
The SAFE score was originally developed for fibrosis detection, and it has since demonstrated the ability to predict liver-related events [
5], survival [
6], and HCC [
1]. Mechanistically, the SAFE components (age, body mass index, diabetes mellitus, aspartate aminotransferases/alanine aminotransferases, platelets, globulin) relate to liver fibrosis, a key driver for HCC pathogenesis. We reported time-dependent area under the receiver operating characteristic curve (AUROC), area under the precision-recall curve (AUPRC), Brier scores, and overall calibration, as well as per-etiology discrimination. Two independent external validation cohorts from both hospital and community settings, consistently supported the SAFE score in predicting HCC.
Our conclusion indicated that the SAFE score stratifies patients at high risk for HCC across different etiologies of CLD and helps identify at-risk candidates for HCC surveillance. However, we did not propose mandatory surveillance solely based on the SAFE score. Further refinement of risk thresholds according to absolute HCC incidence by etiology, along with integration of other prognostic variables, will be important to guide its implementation in clinical practice.
FOOTNOTES
-
Authors’ contribution
T.H.S. wrote the manuscript. J.H.K. supervised and revised the manuscript.
-
Conflicts of Interest
T.-H. S. received research grant from Gilead Sciences, and was on speaker’s bureaus for Abbvie, Bristol-Myers Squibb, Gilead Sciences, Merck Sharp and Dohme, and Takeda. J.-H. K. has served as a consultant for Abbvie, Gilead Sciences, Merck Sharp and Dohme, and Roche and on speaker’s bureaus for Abbvie, Bristol-Myers Squibb, Gilead Sciences, Merck Sharp and Dohme.
Abbreviations
metabolic dysfunction-associated steatotic liver disease
positive predictive value
Steatosis-Associated Fibrosis Estimator
REFERENCES
- 1. 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.
- 2. Satapathy P, Mehta R, Sah R. Letter to the editor on “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 2026;32:e139-e140.
- 3. Su TH, Wu CH, Liu TH, Ho CM, Liu CJ. Clinical practice guidelines and real-life practice in hepatocellular carcinoma: a Taiwan perspective. Clin Mol Hepatol 2023;29:230-241.
- 4. 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.
- 5. Li G, Lin H, Sripongpun P, Liang LY, Zhang X, Wong VWS, et al. Diagnostic and prognostic performance of the SAFE score in non-alcoholic fatty liver disease. Liver Int 2024;44:15-26.
- 6. 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.
Citations
Citations to this article as recorded by
