A predictive model for limited stage diffuse large B-cell lymphoma (DLBCL): A retrospective analysis of 1,252 cases performed by the Intergruppo Italiano Linfomi (IIL)

Introduction: few studies have explored the usefulness of a prognostic index specifically devised for patients with localized DLBCL. The IIL has performed a retrospective analysis of a large group of patients with limited stage DLBCL and developed a new prognostic model. Results: 1,252 patients with localized (Ann Arbor stage I-II) aggressive B-cell lymphoma (IWF: G or H, WHO:DLBCL) diagnosed from 1988 to 2002 and without CNS involvement were identified. Median age was 57 yrs (range 17–91) and M/F ratio was 1.26. Clinical stage was I in 19%, IE in 34%, II in 28% and IIE in 28% patients, respectively. Extranodal involvement was present in 53%, 7% had >1 extranodal site; 26% had Bulky disease (>_10 cm), 28% elevated LDH, 19% B2M, 38% ESR>30, 21% reduced albumin (<3.5 g/dL). Patients were treated with ADM-containing regimens ± IF-RT. The median follow-up was 62 months and 5- year OS was 71%. By univariate analysis 11 variables were found to be predictive of a short survival (P <0.01) (age, stage, number of nodal sites, PS, B symptoms, bulky, ESR, LDH, RT, B2M and albumin). By Cox multivariate analysis, age >_ 65 years (P <0.001), stage II nodal (P <0.001), high LDH (P <0.001) and bulky (P <0.01) were indipendent risk factors (RF). The prognostic model was calculated with the sum of scores assigned to each variable; advanced age, high LDH, and Bulky had a score 2; for Stage a score of 1 was considered for stage Ie-IIe and 2 for stage II nodal. Three groups of patients with a different survival (P <0.000001) were identified. Patients at low (Score 0–1; 387 pts), intermediate (Score 2–3; 484 pts) or high risk (Score 4–8; 381 pts) had a 5 years OS of 87%, 77%, 51% respectively. The predictive performance of the model was internally validated through a non parametric Bootstrap method and through residues’ analysis. Conclusions: This prognostic model will allow us to better select appropriate therapeutic approaches for patients at different risks.