AUTHORS
Julie Haydu, MD, PhD, Catherine Cho2, Anastasia Radko, MSc3, Elina Ohanjanyan, BSc4, Ivan Zubarev, MSc5, Diana Lupova, MSc6, Olga Lukonina, MSc7, Josie Ford8, Sarah Liu9, Jeffrey Barnes, MD, PhD10, Ephraim Hochberg, MD11, Patrick Connor Johnson, MD12, Ronald Takvorian, MD13, Elliot C Woods, MD, PhD14, Jeremy S. Abramson, MD15, Jacob D. Soumerai, MD16, Michael Goldberg, PhD17, Abner Louissaint Jr, MD, PhD18
AFFILIATIONS
1 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
2 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
3 BostonGene Corporation, Waltham, Massachusetts, United States
4 BostonGene Corporation, Waltham, Massachusetts, United States
5 BostonGene Corporation, Waltham, Massachusetts, United States
6 BostonGene Corporation, Waltham, Massachusetts, United States
7 BostonGene Corporation, Waltham, Massachusetts, United States
8 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
9 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
10 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
11 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
12 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
13 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
14 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
15 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
16 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
17 BostonGene Corporation, Waltham, Massachusetts, United States
18 Mass General Brigham Cancer Institute, Boston, Massachusetts, United States
ABSTRACT
Introduction: The immunologic mechanisms driving CD20/CD3 bispecific antibody (BsAb) response in relapsed/refractory (r/r) large B-cell lymphoma (LBCL) are poorly understood. Using high-dimensional flow cytometry and bulk RNA-seq on peripheral blood, we aimed to define immune profiles and predictive signatures in patients (pts) with r/r LBCL receiving BsAb monotherapy.
Methods: Peripheral blood samples (n=27 pts) were prospectively collected at baseline, on treatment (cycle 3 day 1 +/- 1 cycle), and at end of treatment/progression and analyzed by multiparameter flow cytometry and bulk RNA-seq (BostonGene). Immune cell populations, expressed as percentages of parent populations, were compared between subgroups using differential analysis (Mann-Whitney U/Wilcoxon test; median fold change ≥1.2, p<0.05). Differentially expressed genes between subgroups were identified using DESeq2.
Results: Twenty-seven pts were included (median age 75y). Among these, 18 (67%) had de novo DLBCL and 7 (26%) had LBCL arising from indolent B-cell lymphoma. Median number of prior therapies was 2 (range 1-7) and 21 pts (78%) received prior CAR T-cells (median=114 days, range 36-820 days). BsAb was glofitamab (20) and epcoritamab (7). The overall (OR) and complete response (CR) rates were 56% and 33%. CRS and ICANS occurred in 26% (7/27, all grade 1-2) and 7% of pts (2/27, all grade 1), respectively. BsAb treatment shifted the peripheral immune composition from CD4+ and CD8+ transitional and effector memory Tcells to terminally differentiated CD8+ T cell subsets (CD57+, p=0.04; CD45RA+, p=0.02). While antecedent CAR T was not associated with BsAb response (50% no prior CAR vs 64% prior CAR, p=0.68), prior CAR was associated with a higher percentage of effector memory CD4+ (42.6% vs 15.9%, p=0.03) and PD-1+ CD8+ (7.4% vs 2.9%, p=0.05) T cells and increased CD56+ CD16- NK cells (22.4% vs 5.4%, p=0.003). Clinical responders (CR or partial response) showed baseline enrichment of CXCR5+ (p=0.02) and CX3CR1+ (p=0.04) CD8+ T cell subsets compared to the pre-specified on-treatment time point. We also identified baseline innate-like populations as predictors of BsAb response: dendritic cells (AUC 0.750, p=0.04), CD57+ NK cells (AUC 0.759, p=0.04), and γδ T cells (AUC 0.830, p=0.002). On-treatment analysis showed increases in activated PD-1+ (p=0.02) and CD39+ CD45RA+ (p=0.008) CD8+ memory T cell subsets, and CD27- memory B cells (p=0.02), in clinical responders. Bulk RNA-seq performed in baseline samples revealed increased expression of B-cell-associated genes (CD180, p=0.02, FCRL1, p=0.05, VPREB3, p=0.02, IGHM, p=0.02) in responders, whereas a baseline signature consistent with myeloid-derived suppressor cell (MDSC) activation and trafficking (CCL2, p=0.009; CSF1, p=0.008; SERPINB2, p=0.009; SIGLEC1, p=0.004; IRAK2, p=0.009; IFI44L, p=0.02; OASL, p=0.01; IL10, p=0.009) was increased in non-responders at baseline.
Conclusions: Our multimodal approach defined distinct pre-treatment and dynamic immune profiles associated with BsAb efficacy and resistance in r/r LBCL. Clinical responders demonstrated a larger preexisting pool of stem-like, activation-ready T cells and innate-like cytotoxic effectors along with B cell associated transcriptional signatures, indicating a less depleted immune reserve, whereas non-responders showed hallmarks of monocyte-skewed immunosuppression. We also observed a pharmacodynamic signature associated with clinical response, defined by on-target T cell engagement and progressive differentiation toward CD57+ and CD45RA+ subsets following BsAb administration. Prior CAR T therapy induced a chronically activated baseline but did not reduce BsAb response rate, confirming BsAb treatment viability post-CAR. These findings underscore peripheral immune system profiling as a minimally invasive tool for BsAb response prediction and identify MDSC-mediated immunosuppression as key to overcoming BsAb resistance, which we will validate in prospective ongoing BsAb studies.
DISCLOSURES
Julie Haydu, MD, PhD: Genmab - Research Funding; Pfizer - Consultancy (Includes expert testimony); Astrazeneca - Consultancy (Includes expert testimony); Integra - Consultancy (Includes expert testimony); ADCT - Consultancy (Includes expert testimony)