Heterogeneity of interim PET partial metabolic response in DLBCL: Risk stratification using ΔSUVmax.

e19052 Background: Interim 18 F-FDG PET/CT (iPET) is widely used for response assessment in diffuse large B-cell lymphoma (DLBCL). However, the Lugano classification relies on qualitative metabolic response categories, and the prognostic heterogeneity within Lugano-defined partial metabolic response (PMR) remains poorly characterized. Evidence supporting treatment modification based on iPET findings is also limited. Methods: We retrospectively analyzed 125 newly diagnosed DLBCL patients who underwent 18 F-FDG PET/CT at baseline and after four cycles of first-line therapy. Interim response was assessed according to Lugano criteria and categorized as PMR or no metabolic response (NMR). The relative change in tumor SUVmax between baseline and interim PET (ΔSUVmax) was calculated as a quantitative imaging biomarker. Progression-free survival (PFS) was the primary endpoint. Multivariable Cox regression was used to identify independent prognostic factors. A composite risk score incorporating ΔSUVmax and International Prognostic Index (IPI) group was constructed to stratify patients into risk categories. Interaction analyses explored whether baseline risk modified the association between treatment modification after cycle 4 and PFS. Results: Among 125 patients, 122 (97.6%) were classified as PMR and 3 (2.4%) as NMR by Lugano criteria. During follow-up, 31 PFS events occurred. Unfavorable ΔSUVmax (<= 66%) independently predicted inferior PFS (hazard ratio [HR] 4.56, 95% confidence interval [CI] 2.18–9.54; P<0.001). IPI group was also independently associated with PFS. The final model combining ΔSUVmax and IPI achieved a concordance index of 0.752. Risk stratification based on the composite score demonstrated clear separation of PFS among low-, intermediate-, and high-risk groups (log-rank P<0.001). Formal interaction testing using a Cox model including an interaction term between risk group and treatment change revealed a borderline interaction effect (likelihood ratio test P≈0.10), suggesting that the impact of treatment modification may differ according to baseline risk level. Conclusions: Substantial prognostic heterogeneity exists within Lugano-defined PMR patients. Quantitative assessment using ΔSUVmax refines risk stratification beyond qualitative iPET criteria and may help contextualize treatment decisions following interim PET in DLBCL.