The biology and clinical impact of bone marrow (BM) infiltration in patients with diffuse large B-cell lymphoma (DLBCL) remains unclear in the rituximab era. We retrospectively analyzed 232 patients diagnosed with DLBCL at our center between 1999 and 2014. Concordant-presence of large cells similar to those of the lymph node biopsy- and discordant-infiltration by small cells forming lymphoid aggregates, lacking cytological atypia-BM infiltration was defined by histological criteria and further characterized by flow cytometry (FCM). Cell of origin (COO) was determined using Hans’ algorithm. For the clonal relationship between tumor and discordant BM, the VDJH rearrangement was analyzed. Survival analyses were restricted to 189 patients treated with rituximab and chemotherapy. Thirty-six (16%) had concordant, and 37 (16%) discordant BM infiltration. FCM described different indolent lymphomas among discordant cases, clonally related with DLBCL in 10/13 available samples. Median follow-up was 58 months. 5-year-progression-free survival (PFS) for non-infiltrated, discordant and concordant groups was 68%, 65% and 30%, respectively (p < 0.001). Combining COO and BM infiltration, patients with discordant BM and non-germinal center B-cell COO also had decreased 5-year-PFS (41.9%). In multivariate analysis, concordant BM had an independent effect on PFS (HR 2.5, p = 0.01). Five-year cumulative incidence of central nervous system (CNS) relapse was 21%, 4% and 1% in concordant, discordant and non-infiltrated groups, respectively (p < 0.001). In conclusion, concordant BM infiltration represents a subset with poor prognosis, whereas the prognostic impact of discordant BM infiltration could be limited to non-CGB cases.

No abstract