Predictive value of metabolic and tumor dissemination parameters in diffuse large B-cell lymphoma: can relapse be predicted despite adequate interim PET response?

OBJECTIVE: To evaluate associations of baseline and interim fluorine-18 fluorodeoxyglucose PET/computed tomography parameters with early clinical outcomes in diffuse large B-cell lymphoma (DLBCL) and identify prognostic/predictive biomarkers, including predictors of discordance between interim PET (iPET) response and clinical outcome. METHODS: We retrospectively analyzed 112 patients with DLBCL. Baseline clinical/biochemical data and metabolic parameters from staging PET (sPET) and iPET were recorded, and delta (Δ) parameters were calculated. iPET response was classified by Deauville criteria, while follow-up outcomes were categorized as remission or refractory/progressive disease. Group comparisons and receiver operating characteristic analyses were performed. RESULTS: iPET predicted outcomes with 79% sensitivity, 71% specificity, 74% accuracy, 82% negative predictive value, and 66% positive predictive value. Baseline dissemination metrics, including maximum distance between the two most distant lesions and maximum distance between the outermost voxels of the two most distant lesions PET (DmaxVoxPET), were higher in patients developing relapse/progression despite adequate iPET response [area under the curve (AUC): 0.81 and 0.77]; iPET-derived metabolic parameters - maximum standardized uptake value (SUVmax), mean standardized uptake value (SUVmean), peak standardized uptake value (SUVpeak), lesion-to-liver ratio (RLL), total lesion glycolysis, and ΔSUVmax, ΔSUVmean, ΔSUVpeak, and ΔRLL significantly predicted remission (AUC: 0.70-0.79), generally outperforming baseline parameters. SUV-based indices and RLL outperformed TLG, while SUV metrics performed comparably. Ferritin was the only predictive clinical/biochemical variable (AUC: 0.79). According to multivariate logistic regression analysis, advanced-stage disease (P = 0.002) and elevated ferritin levels (P = 0.009) were found to be independent risk factors for disease progression. CONCLUSION: iPET provides robust early response assessment and stronger predictive value than sPET in DLBCL, particularly through SUV-based parameters and RLL. While sPET reflects tumor burden and dissemination, baseline dissemination metrics may identify patients at risk of relapse despite favorable iPET findings, supporting risk-adapted management and closer surveillance.