The paper, titled “Baseline Lymphocyte Count Outperforms Dosimetry and Inflammatory Markers for Predicting Severe Radiation-Induced Lymphopenia in Rectal Cancer,” presents advanced predictive modeling of severe radiation-induced lymphopenia in rectal cancer patients undergoing preoperative chemoradiotherapy.

In a cohort of 179 patients, over 600 dosimetric variables automatically extracted using the TotalSegmentator v2 algorithm, 23 peripheral blood count parameters, and 6 inflammatory markers were analyzed. The primary finding of the study is the absolute dominance of baseline lymphocyte count as the strongest factor determining the occurrence of severe immune suppression. A three-variable nomogram based on baseline lymphocyte count, age, and sex achieved an adjusted C-index of 0.746. Incorporating a single lymphocyte measurement at day 14 of treatment enabled the creation of a dynamic prediction system, increasing the cross-validated area under the ROC curve to 0.827. Concurrently, it was demonstrated that in standardized VMAT radiation plans, advanced pelvic dosimetric parameters and inflammatory markers provided no significant added value to the model after correction for multiple testing. From a methodological perspective, the authors also drew attention to the issue of immortal time bias in prior publications, proving that the apparent impact of lymphocyte nadir on overall survival disappears after applying time-dependent Cox models. Based on the developed model, a free risk calculator has been made available at https://kstawiski.github.io/rect-ril/. The analysis code has been deposited in the Zenodo repository under DOI 10.5281/zenodo.19595347.

The study was conducted in collaboration with researchers from Harvard Medical School with support from the National Science Centre (PRELUDIUM) and Medical Research Agency within the Polish Clinical Scholars Research Training Program and the Regional Digital Medicine Centre of the Medical University of Lodz (MULDIMediC). The full text of the article is available at https://doi.org/10.1016/j.ijrobp.2026.08.034