Ki67 Immunohistochemical Expression Level ≥70%, Bulky Presentation ≥7.5 cm, Meningeal Lymphomatosis, and Interim PET ΔSUVmax After 4 Treatment Cycles <71% as Parts of a Practical Scoring System to Predict Progression-Free Survival and Overall Survival in Diffuse Large B-Cell Lymphoma

Abstract

Currently, prognostic models in diffuse large B-cell lymphoma (DLBCL) fail to closely reflect patients' biological, clinical, and survival heterogeneity. We, therefore, assessed the impact of clinical, biological, immunohistochemical (IHC), baseline (0), and interim (after 2 and 4 treatment cycles) PET (PET0, PET2, and PET4) data not yet included in any scoring system on DLBCL outcome. The analysis was conducted on 89 previously untreated adult patients of the Finistere Observatory Cohort (O.Ly.Fin) with documented DLBCL, recruited between January 2010 and December 2017, with progression-free survival (PFS) and overall survival (OS) as primary and secondary endpoints, respectively. Seventy-eight patients were treated with rituximab, cyclophosphamide, hydroxyadriamycin, vincristine, and prednisone (R-CHOP), while 11 received R-dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, and hydroxyadriamycin (EPOCH). Patients were followed up until June 20, 2020. On multivariate analysis, Ki67 ≥ 70% on IHC (K), bulky presentation ≥7.5 cm (B), meningeal lymphomatosis (M), and PET0–PET4 ΔSUVmax &lt;71% (P4) were identified as strong independent predictors of PFS, and all variables but bulky disease also strongly and independently predicted OS. Using these 4 parameters, we designed a scoring model named KBMP4 stratifying patients into low- (0 parameter), intermediate- (1 or 2), and high-risk (≥3) subgroups by the Kaplan–Meier analysis. At a median follow-up of 43 months, PFS and OS were both 100% in the low-risk subgroup, 71.4 and 90.5%, respectively, in the intermediate-risk subgroup, and 0 and 55.5%, respectively, in the high-risk subgroup. Use of the KBMP4 model in clinical practice may improve accuracy in prognostic prediction and treatment decisions in de novo DLBCL patients.