Clinicopathologic features and outcomes of de novo transformed indolent lymphoma

8061 Background: Untreated transformed indolent lymphoma (unTIL) can present as a composite lymphoma (COM), 2 or more separate sites of disease with one site transformed (discordant; DIS) or following indolent lymphoma sequentially (SEQ). Current practices are guided by small retrospective studies or extrapolated from trials involving non-transformed lymphomas. Methods: 353 patients (pts) with biopsy-proven unTIL treated with curative chemoimmunotherapy between 01/2000 to 01/2019 were included (intention to treat). All indolent B-cell lymphomas (iNHL) were included except CLL/SLL. Patients with MCL and non-DLBCL transformation were excluded. Prior therapy for iNHL was not allowed except one line of non-chemotherapy-based therapy. Kaplan-Meier method was used for time-to-event analysis including progression-free (PFS) and overall survival (OS). Results: 252 (71%) pts presented with COM, 50 (14%) DIS and 51 (14%) SEQ lymphoma. The underlying iNHL was: 308 (87%) follicular lymphoma, 37 (10%) nodal MZL, 7 (2%) MALT lymphoma, and 1 (0.3%) WM. Frontline therapy (FLT) included: RCHOP for 271 (77%), DAEPOCHR for 60 (17%), clinical trial for 7 (2%), BR for 4 (1%), RHCVAD for 2 (1%), RCEOP for 2 (1%), radiation therapy for 2 (1%), RFND for (1%) 2, RCVP for 2 (1%), and rituximab only for 1 (0.3%). 9 (3%) pts had ASCT in first remission. 50 (15%) pts received maintenance rituximab (MR) with fewer cases of HGBL-DH compared to the non-MR cohort (0% v 10%). With a median follow-up of 3.4 years (range 0.1-19.1), 4-year PFS and OS rates were 59% and 88%. By univariate analysis (UVA) the underlying type of iNHL, cell-of-origin and choice of induction therapy (DAEPOCHR v RCHOP) were not associated with inferior outcomes. SEQ transformations were associated with inferior OS on UVA (P = 0.02) which was not significant on multivariable analysis (MVA) (P = 0.3). MVA identified ECOG PS > 1, B symptoms and HGBL-DH as independent prognostic factors for inferior PFS and OS. In patients who achieved PR or greater following FLT, MR was associated with improved PFS on MVA (HR 0.6, 95% CI 0.3-0.9, P = 0.04) without an OS benefit (P = 0.2). 39 (31%) pts relapsed with iNHL only (mPFS 2.4 yrs, 4-yr OS 94%) and 88 (69%) relapsed with transformed lymphoma (mPFS 1.1 yrs, 4-yr OS 69%) with no significant difference in pattern of relapse with MR (P = 0.2). Conclusions: The clinicopathologic features of unTIL are similar to those of de novo DLBCL. Escalation of therapy beyond R-CHOP may not be required in the absence of HGBL-DH. unTIL should be included in future clinical trials involving de novo DLBCL given the similar clinicopathologic features.