EACR26-0682

Spatial and Single-Cell Profiling Identify Tfh Niches and CD8⁺ Exhaustion as Drivers of Glofitamab Resistance in B-NHL

P. Marcoux1, F. Gava1, M. Tosolini1, P. Gravelle1, C. Schniederjohann2, F. Bouquet3, P. Bruch2, P. Perez-Galan,4, C. Bezombes1, C. Laurent5
1Univ Toulouse, INSERM, CRCT, Toulouse, France
2Department of Hematology, Oncology and Clinical Immunology, Medical Faculty and University Hospital Düsseldorf, Dusseldorf, Germany
3Roche, Bâle, Switzerland
4IDIBAPS, Barcelona, Germany
5CHU Toulouse, IUCT Oncopole,, INSERM, CRCT,, Toulouse, France
Introduction:

CD20×CD3 bispecific antibodies (bsAbs) such as glofitamab have demonstrated substantial efficacy in relapsed/refractory (R/R) B-cell non-Hodgkin lymphoma (B-NHL), however resistance remain frequent. To better understand mechanisms underlying heterogeneous responses, we used a patient-derived 3D ex vivo model called patient-derived lymphoma spheroids (PDLS) that recapitulates tumor and immune microenvironment. We aimed to identify predictive biomarkers of glofitamab response and uncover actionable resistance mechanisms.

Material and method:

PDLS were generated from peripheral blood or lymph node biopsies of 39 R/R B-NHL patients. Glofitamab-induced B-cell depletion was quantified ex vivo and correlated with high-dimensional immune profiling, including multiparameter flow cytometry, single-cell RNA sequencing, and spatial proteomics using CODEX multiplex immunofluorescence imaging. Spatial transcriptomics (Xenium) was performed on an independent cohort of B-NHL. Functional validation included TIGIT blockade and CD4⁺ T follicular helper (Tfh) cell depletion in 3D cultures.

Result and discussion:

PDLS responses segregated patients into two distinct groups and recapitulated available clinical outcomes (8/9 concordant cases). High responders displayed significantly higher baseline CD8⁺ T-cell frequencies, positively correlated with B-cell depletion. Single-cell profiling revealed enrichment of cytotoxic CD8⁺ T-cells with elevated activation signatures in high responders, whereas low responders exhibited increased exhausted CD8+ T-cells. TIGIT blockade enhanced glofitamab-induced depletion selectively in low responders. Low responders were enriched in functionally active Tfh cells. These samples demonstrated higher Tfh signature scores, increased IL21⁺ Tfh activity, and strengthened inferred signaling toward tumor B-cells. Spatial proteomics confirmed enriched B/Tfh niches and reduced intercellular distances between B-cells and Tfh. Xenium spatial transcriptomics are currently underway to better characterize the spatial organization of the immunoprotective niche in primary tumors. Functional Tfh depletion significantly improved glofitamab-mediated B-cell killing in 3D models. Independent clinical transcriptomic data revealed higher Tfh abundance in patients experiencing progression under glofitamab.

Conclusion:

Using a multimodal patient-derived 3D platform, we identify two major predictive biomarkers of glofitamab resistance in R/R B-NHL: exhausted CD8⁺ T-cells and functionally active Tfh cells sustaining tumor-supportive niches. These findings provide a ratioanl for combinatorial strategies integrating bsAbs with immune checkpoint blockade or disruption of Tfh–B-cell crosstalk. Moreover, this study validates PDLS as a relevant preclinical model for investigating immune escape mechanisms and identifying biomarkers predictive of immunotherapy response.