Response to ‘Statistical considerations in the comparison of immunotherapy and chemotherapy for <scp>POD24</scp> follicular lymphoma’

We thank Yan and Liu for their interest in our study and for their thoughtful comments addressing the methodological challenges inherent to retrospective analyses in patients with progression of disease within 24 months of first-line (POD24) follicular lymphoma.1 We fully agree that baseline imbalances are a key concern in non-randomized studies. For this reason, we applied an inverse probability of treatment weighting (IPTW) approach based on the propensity score to reduce confounding by measured variables. Propensity score-based methods are widely accepted as one of the most appropriate strategies to address treatment selection bias in observational research when randomization is not feasible.2-4 IPTW allows the construction of a weighted pseudo-population in which baseline characteristics are more comparable between treatment groups. Importantly, variables included in the propensity model were predefined a priori based on established clinical relevance in relapsed follicular lymphoma. This included age, stage and follicular lymphoma international prognostic index (FLIPI) score at first relapse, best response to front-line therapy (including refractory disease), prior maintenance therapy, B symptoms at relapse and autologous haematopoietic stem cell transplantation. These factors capture key dimensions of disease burden, aggressiveness and treatment sensitivity, all of which are known to influence post-relapse outcomes. Variable selection was therefore clinically driven rather than data-driven. After weighting, baseline differences between groups were substantially attenuated, and the favourable association of immunotherapy with progression-free survival and duration of response remained consistent. While residual confounding cannot be completely excluded in any retrospective study, the persistence of the effect after adjustment suggests that the observed differences are unlikely to be explained solely by measurable baseline imbalances. We agree that the higher proportion of patients with primary refractory disease and the exclusive occurrence of histological transformation in the chemotherapy cohort likely reflect a more aggressive disease biology. Our study was not designed to establish causality between treatment strategy and transformation risk, and these observations were therefore reported descriptively. Regarding treatment heterogeneity, our intention was not to attribute benefit to a specific agent, but to explore the clinical relevance of chemotherapy-free strategies in this high-risk setting. Importantly, lenalidomide-based approaches, which represented a substantial proportion of the immunotherapy cohort, are currently used as the comparator arm in several ongoing and recent clinical trials evaluating novel combinations in relapsed follicular lymphoma. In this context, grouping contemporary immunotherapy-based approaches reflects current clinical trial design and real-world practice. Finally, we acknowledge the smaller sample size and shorter follow-up of the immunotherapy cohort. Despite these limitations, a durable separation in progression-free survival and duration of response was observed, supporting the clinical relevance of our findings. Our aim was not to propose a new standard of care, but to provide evidence suggesting that POD24 patients progressing after front-line chemoimmunotherapy may benefit from chemotherapy-free strategies. We concur with Yan and Liu that prospective studies are needed to confirm these observations and to refine patient selection. Víctor Navarro: Formal analysis; supervision; writing – review and editing. Ángel Serna: Conceptualization; writing – original draft. Pau Abrisqueta: Writing – review and editing; supervision. The authors acknowledge the GELTAMO group and all participating institutions involved in the original study. The authors have nothing to report. The authors declare no conflicts of interest. No new data were generated or analysed in support of this research.