We read with great interest the article by Kang SH et al. examining HCC incidence in chronic hepatitis B patients treated with TAF versus TDF 1. Although we commend the authors' efforts, we raise several methodological and interpretive concerns that limit conclusions about TAF's superiority in reducing HCC risk. HCC is a relatively rare outcome; thus, meaningful comparative inference requires a sufficiently large number of events. In this study, only nine cases occurred (two in the TAF group, seven in the TDF group) among 1364 patients followed for up to 60 months. Given the extremely small number of HCC events in this study, it is essential to consider the basic requirements for adequate statistical power in time-to-event analyses 2. Detecting even a 30% relative reduction in HCC risk—a hazard ratio of 0.7—with 80% power would require roughly 247 total events, far exceeding the nine cases observed. With data this sparse, effect estimates are inherently unstable, confidence intervals widen substantially, and the apparent direction of association could shift with the addition or reclassification of a single case. Moreover, the cumulative incidence rates—0.62% in the TAF group and 0.67% in the TDF group—are effectively indistinguishable, differing by only 0.05%. Therefore, no meaningful statistical conclusion can be drawn regarding a true difference in HCC risk between TAF and TDF based on these event counts, and any claim of superiority is unsupported by fundamental principles of survival analysis and statistical power. Baseline imbalances in key determinants of HCC risk further complicate the interpretation of the study's findings. Patients in the TDF group were more likely to have cirrhosis, lower platelet counts, and higher AFP levels, indicating a population with substantially greater underlying risk for HCC. Although the authors used propensity score matching to address these differences, such methods can improve covariate balance only at the cost of reduced generalizability and the potential introduction of selection bias when substantial numbers of patients must be excluded 3. In this study, achieving balance required removing 398 TDF-treated individuals—approximately 40% of the original cohort—raising serious concerns about representativeness and the validity of the matched comparison. Without detailed characterization of the excluded patients, it is impossible to determine how they differed from those retained or whether their omission materially influenced the observed results. Moreover, the study did not report HCC outcomes among these excluded TDF patients, and any events in this higher-risk group could meaningfully alter the interpretation of the comparative analysis. Propensity score methods can adjust only for known and accurately measured confounders. Critical factors such as duration of infection, long-term viral suppression history, medication adherence, and precise fibrosis staging are often incompletely captured in retrospective datasets and therefore cannot be fully balanced through matching. These unmeasured or imprecisely measured variables may have substantially influenced the observed outcomes. TAF and TDF share the same active antiviral agent, tenofovir; they differ only in intracellular delivery and side effect profiles 4. There is no plausible mechanistic basis by which TAF would confer greater protection against HCC than TDF. Thus, any apparent difference in HCC outcomes should be interpreted with considerable caution and is more likely attributable to confounding, selection bias, chance variation, or other methodological limitations than to a true biological effect. Given the extremely small number of events, chance alone remains a highly plausible explanation for the reported disparity. Given the extremely small number of HCC events, the limitations of the propensity-matching strategy, the persistent potential for residual confounding, and the absence of any mechanistic basis for differential HCC protection between TAF and TDF, the study's findings should be regarded as hypothesis-generating rather than confirmatory. Drawing firm conclusions about drug-level superiority from such sparse and methodologically fragile data risks misleading clinicians—especially given TAF's substantially higher cost—and may inappropriately influence prescribing patterns or guideline development. A sufficiently powered, prospectively designed study with rigorous adjustment for established HCC risk factors is required before any meaningful claim of TAF's superiority can be supported. Until such evidence emerges, these results should not guide clinical decision-making. Data sharing is not applicable to this article as no datasets were generated or analyzed during the current study.
Siraj et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: