Letter: Improving the Interpretability and Portability of Tumour Burden Score–Based Prediction of Extrahepatic Progression After Transarterial Chemoembolization ( <scp>TACE</scp> )

Pinto and colleagues report a multicentre evaluation of tumour burden score (TBS) for predicting extrahepatic progression (EHP) after first-line transarterial chemoembolization (TACE) in hepatocellular carcinoma [1]. The reported event pattern—EHP as the first progression in 7.2% and EHP at any time during follow-up in 26.1%—makes a compelling case for pre-TACE risk stratification. A presentation issue may, however, limit clinical interpretation. Because death precludes subsequent EHP, EHP is intrinsically a competing-risk endpoint [2]. The authors used cause-specific hazards and censored death without EHP, which is suitable for association modelling, but many readers will look for absolute risk [3, 4]. A cumulative incidence plot of EHP by TBS category (with death as the competing event), together with a Fine–Gray analysis, would allow clinicians to interpret the stratification as a probability over time rather than as a hazard ratio. Portability of the proposed TBS threshold would also benefit from additional reassurance. The cut-point of 3.66 was selected by minimising the p-value of the hazard ratio in a univariable Cox model, and patients were dichotomised into TBS-low and TBS-high accordingly [5]. Reporting the effect of TBS as a continuous predictor (e.g., per standard deviation increase), alongside bootstrap-based internal validation or a simple quantile-based sensitivity analysis, would help demonstrate that the association is not dependent on a single sample-specific split [6]. Finally, the manuscript offers a clinically useful ‘pre-treatment plus early response’ perspective [7]. The combination of TBS-high and AFP > 1000 yields a striking gradient in 3-year risk of first-episode EHP (approximately 7.8%, 25.6% and 68.7%), and SD/PD on first post-TACE mRECIST remains independently associated with EHP during follow-up [8, 9]. Given the multicentre real-world setting and potential variability in surveillance intensity, it would be helpful to state more explicitly how follow-up imaging was scheduled and how this might influence time-to-EHP estimates, particularly if the authors envision the framework guiding earlier treatment adaptation. Sincerely, Bingfeng Li and Zichen Yu Bingfeng Li: conceptualization, writing – original draft. Zichen Yu: supervision, writing – review and editing. The authors have nothing to report. The authors declare no conflicts of interest. This article is linked to Pinto et al papers. To view these articles, visit https://doi.org/10.1111/apt.70534 and https://doi.org/10.1111/apt.70587. The authors have nothing to report.