Introduction: BCL2 overexpression leading to evasion of apoptosis is common in hematologic malignancies. In DLBCL, 50\% of patients (pts) overexpress BCL2 protein, 30\% overexpress BCL2 and MYC (double-expressor [DE]) and 5-10\% have translocations of BCL2 and MYC (double-hit [DH]), all of which are associated with a poor prognosis. We hypothesized that combination of the BCL2 inhibitor venetoclax (Ven) with chemotherapy would improve DLBCL outcomes based on pre-clinical and early clinical data. The Phase Ib part of the CAVALLI (NCT02055820) study assessed the maximum tolerated dose (MTD) of CHOP + rituximab (R) or obinutuzumab (G) with Ven (200-800mg) in non-Hodgkin lymphoma, recommending 800mg of Ven + R-CHOP for Phase II. The single-arm, multicenter Phase II part of CAVALLI investigated the efficacy of Ven + R-CHOP in all first-line (1L) DLBCL pts, by BCL2 immunohistochemistry (IHC) status within cell-of-origin (COO) subtypes, by BCL2 fluorescence in situ hybridization (FISH) and in DE and DH pts, with the intent to compare against a matched pt population in the R-CHOP arm of the GOYA Phase III study. Here we report the first safety, efficacy and biomarker analyses in all pts from this ongoing Phase II study.Methods: Eligible pts (age ≥18 yrs; Eastern Cooperative Oncology Group performance status ≤2; 1L DLBCL; International Prognostic Index score 2-5; ≥1 measurable lesion \gt;1.5cm) were assigned to receive six 3-weekly cycles of R-CHOP + 800mg Ven daily for 10 days (Days 1-10, except Cycle 1: Days 4-10), followed by two 3-weekly cycles of 800mg Ven on Days 1-10 + R on Day 1. The primary endpoint was PET-CT response 6-8 weeks after the last R dose (EoT), according to a modified version of Lugano 2014 criteria. Secondary endpoints were progression-free survival (PFS) and safety.Biomarker analyses in CAVALLI and GOYA were performed in pre-treatment tumor samples including a BCL2 IHC assay (cutoff: 50\% medium/high expression), MYC IHC assay (cutoff 40\% signal), BCL2 and MYC FISH, and COO assay (Nanostring).Results: 211 pts were enrolled; 208 received any treatment and were included in efficacy and safety analyses. Overall, pt characteristics were similar for CAVALLI and GOYA, except CAVALLI enrolled more Ann Arbor Stage IV (65.4\% vs 47.1\ and BCL2 IHC-positive pts (57.7\% vs 50.0\. The EoT complete response rate in all pts did not differ significantly between CAVALLI and GOYA (69.2\% vs 62.8\ respectively; Table 1), but was improved in BCL2-positive subgroups, specifically in BCL2 FISH-positive (70.0\% vs 47.5\ and DH (71.4\% vs 25.0\ pts. With 20 months' median follow-up in CAVALLI, disease progression and death had occurred in 29 and 6 pts, respectively. When compared with GOYA and adjusted for baseline covariates by Cox methodology, PFS improvement was observed in BCL2 IHC-positive pts (HR, 0.53; 95\% CI: 0.30-0.93), including within both activated B-cell (ABC; HR, 0.43; 95\% CI: 0.19-0.94) and germinal center B-cell (GCB; HR, 0.41; 95\% CI: 0.17-0.95) COO subgroups. No PFS benefit was observed in BCL2 IHC-negative GCB pts; the BCL2 IHC-negative ABC subgroup was too small for evaluation (Figure 1).Grade 3-4 adverse events (AEs) occurred in 85\% (176/208) of pts in CAVALLI versus 66\% (373/574) in GOYA; the majority were cytopenias, infections and febrile neutropenia (Table 2). In CAVALLI, 1\% (3/208) of AEs were fatal versus 5\% (30/564) in GOYA but follow-up was longer in GOYA (57 vs 20 months). The higher rate of AEs in CAVALLI led to dose interruptions/discontinuations of both Ven and R-CHOP; 61\% of pts had \gt;90\% relative dose intensity (RDI) of Ven; 73.2\% of CAVALLI pts had \gt;90\% RDI of each of doxorubicin and cyclophosphamide versus 76.4\% for doxorubicin and 77.6\% for cyclophosphamide in GOYA. There were no major differences in grade 3-4 AEs or the RDI of R-CHOP across intention-to-treat and BCL2-positive subgroups. GCSF prophylaxis was recommended, but not uniformly delivered.Conclusions: The addition of Ven to R-CHOP in 1L DLBCL treatment resulted in improved efficacy in BCL2 IHC-positive pts compared with matched GOYA controls. Higher rates of cytopenia, infection and febrile neutropenia were observed in CAVALLI versus the R-CHOP arm in GOYA. These data further support exploration of Ven + R-CHOP in a high-risk population of BCL2-positive 1L DLBCL, including DH pts.