Abstract
Diffuse large B-cell lymphoma (DLBCL) is clinically and biologically heterogeneous. R-CHOP remains the frontline standard, with polatuzumab-R-CHP conferring a subgroup-dependent progression-free survival gain, yet early relapse and primary refractoriness persist. Classical risk indices, especially IPI and revised versions, retain benchmark value but only partially capture adverse biology. This review integrates established and emerging tools: multivariable clinical scores (e.g., NPI, GELTAMO), geriatric models (GPI), host-status metrics (HALP, GNRI), PET-derived tumor burden and dissemination (TMTV, TLG; IMPI), immune-microenvironment signatures, and health-system markers (diagnosis-to-treatment interval). We summarize genetic predictors (e.g., CD79B/PIM1) and dynamic molecular response via ctDNA (early/major molecular response), and appraise interim PET as prognostic but not treatment-directive outside trials. In relapsed/refractory disease, second-line age-adjusted IPI and pre-transplant PET inform autologous transplantation candidacy, while randomized trials establish CD19 CAR-T as superior to salvage chemoimmunotherapy in early relapse or primary refractory settings; bispecific antibodies and antibody-drug conjugates expand options post-CAR-T or for transplant-ineligible patients. On the other side, CNS risk assessment is best approached with CNS-IPI refined by site and genotype; prophylaxis remains individualized given mixed efficacy signals. Overall, risk-adapted, biologically driven care should report NCCN-IPI (alongside IPI) in all patients and incorporate imaging burden, genomics, and ctDNA where feasible.