Abstract
Background/Objectives: Accurate prediction of severe adverse events (SAEs) is crucial for optimizing supportive care while maintaining treatment intensity in diffuse large B-cell lymphoma (DLBCL). We evaluated the predictive value of the Geriatric Nutritional Risk Index (GNRI) for SAEs in de novo DLBCL and examined potential interactions with treatment regimen and age. Methods: This multicenter retrospective study included 555 adults treated with standard immunochemotherapies. SAEs, defined as grade ≥ 3 non-hematological adverse events or febrile neutropenia, were independently assessed by board-certified hematologists. Results: Multivariable logistic regression identified GNRI as an independent predictor of SAEs (odds ratio 0.982, 95% confidence interval 0.967–0.997). Restricted cubic spline modeling revealed a significant non-linear association between GNRI and SAE risk (p = 0.045). No significant interaction was observed between GNRI and regimen or age (p = 0.894 and 0.217, respectively), a finding consistent across subgroups in forest plot analyses. Conclusions: This study showed that lower diagnostic GNRI was independently associated with higher SAE risk regardless of treatment regimen or age. These findings highlight the potential utility of GNRI as a simple clinical indicator for identifying patients at higher risk of treatment-related toxicity, although they are derived from a retrospective, tertiary-care cohort and require confirmation in external prospective studies.