Letter: Improving the Interpretability and Portability of Tumour Burden Score–Based Prediction of Extrahepatic Progression After Transarterial Chemoembolisation (TACE)—Author's Reply

We thank Li and Yu for their thoughtful and constructive comments on our study evaluating the tumour burden score (TBS) as a predictor of extrahepatic progression (EHP) after transarterial chemoembolisation (TACE) in patients with hepatocellular carcinoma (HCC) [1, 2]. Their methodological suggestions offer a valuable opportunity to further strengthen the interpretation and clinical applicability of our findings. Regarding competing-risk analysis, we acknowledge that death precludes subsequent EHP, making it an intrinsic competing event. In our cohort, 238 patients (26.7%) died without developing EHP, a substantial competing event that warrants consideration. Although our cause-specific hazard analysis remains appropriate for identifying predictive factors and for understanding the biological association between TBS and EHP risk, we recognise the value of presenting absolute risks for clinical interpretation. We therefore performed a supplementary cumulative incidence function (CIF) analysis treating death without EHP as the competing event. This analysis confirmed a clinically meaningful and statistically significant separation between TBS groups throughout follow-up. At 36 months, the cumulative incidence of EHP was approximately 14% in TBS-low patients and 28% in TBS-high patients, a twofold difference in absolute risk that reinforces the clinical utility of TBS-based stratification. Regarding the portability of the TBS threshold, we agree with Li and Yu that demonstrating robustness beyond a single data-derived cut-point is essential for external validity. To address this concern, we conducted a quartile-based sensitivity analysis of EHP rates across the TBS distribution. This analysis revealed a consistent monotonic increase in EHP rates across TBS quartiles: Q1 (TBS 1.0–2.9) 17.0%, Q2 (TBS 2.9–3.9) 21.8%, Q3 (TBS 3.9–5.1) 28.4% and Q4 (TBS 5.1–42.0) 37.4%. This clear gradient strongly suggests that the TBS-EHP association is continuous rather than dependent on a specific threshold. Furthermore, when analysed as a continuous predictor, mean TBS was significantly higher in patients who developed EHP than in those who did not (4.61 [IQR 3.34–5.61] vs. 3.64 [IQR 2.79–5.00], p < 0.001). These findings support the broader applicability of TBS as a risk stratification tool, regardless of the specific cut-point used. Regarding follow-up imaging protocols, we appreciate the opportunity to clarify this aspect. Participating centres generally followed national and international guidelines recommending contrast-enhanced computed tomography or magnetic resonance imaging at 4–6 weeks after TACE, with subsequent imaging every 3–4 months during the first 2 years and every 6 months thereafter [3-6]. Although our multicentre, real-world design inevitably introduces some variability in surveillance schedules across institutions, we believe the consistency of our findings across these diverse clinical settings enhances the generalisability of our results to routine clinical practice. In conclusion, these supplementary analyses confirm that the association between TBS and EHP risk is robust when assessed using competing-risk methods and across multiple analytical approaches. We believe these findings strengthen confidence in TBS as a clinically useful and easily calculable tool for pre-TACE risk stratification in patients with HCC. Elisa Pinto: writing – original draft, conceptualization, writing – review and editing. Filippo Pelizzaro: writing – review and editing. Laura Bucci: revision. Martina Gambato: writing – review and editing. Fabio Farinati: writing – review and editing, supervision. Edoardo G. Giannini: writing – review and editing, supervision. Francesco Paolo Russo: writing – review and editing, supervision. The authors have nothing to report. 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.70562. Data sharing is not applicable to this article as no datasets were generated or analysed during the current study.