Multiplex immunofluorescence based spatial analysis of HER2 positive breast cancer patients treated with bintrafusp alfa: A translational pathology study

Purpose: To characterize spatial remodeling of the tumor immune microenvironment and to define therapy associated cellular and architectural changes in stage II and III HER2 positive breast cancer treated with Bintrafusp Alfa, a bifunctional antibody that combines PD-L1 blockade with a TGFbeta trap. Experimental Design: Formalin fixed paraffin embedded baseline and on treatment biopsies (TP2) from patients enrolled in the MD Anderson clinical trial NCT03620201 were analyzed across five Multiplex Immunofluorescence (mIF) panels that captured: (1) PD axis T cell states, (2) cytotoxic, memory, and regulatory T cell subsets, (3) tumor and myeloid checkpoints and metabolic ligands, (4) T cell co-stimulation and exhaustion markers, and (5) myeloid and neutrophil axes. Spatial analysis was performed using the SPIAT R package to quantify cell-type distributions, identify unsupervised spatial clusters, and define neighborhood compositions. Cases were classified as responsive or refractory based on composite TP2 spatial features such as cytokeratin (CK) fragmentation, intratumoral CD8 positive infiltration, checkpoint-rich interfaces, and mixed immune and tumor clustering. Results: A subset of TP2 specimens showed clear spatial remodeling characterized by(i) fragmentation of CK architecture into smaller epithelial nests separated by broader stromal areas,(ii) increased infiltration of CD8 and CD3 T cells at tumor stroma interfaces and within epithelial nests,(iii) localized PD1, PDL1, OX40 positive, and ICOS positive immune niches with focal TIM3 expression, and(iv) closer proximity of macrophages and neutrophils to fragmented CK borders without widespread activation of myeloid subsets.These features were enriched in responsive tumors, forming mixed tumor and immune clusters and inflamed spatial neighborhoods. In contrast, refractory tumors retained cohesive CK nests, peripheral T cell rims, and limited checkpoint marker expression. Conclusions: Combined PD-L1 and TGFbeta inhibition with Bintrafusp Alfa promotes heterogeneous spatial reprogramming of the breast cancer microenvironment. Beneficial remodeling is characterized by mixed epithelial and immune clusters, intratumoral cytotoxic and memory T cell infiltration, and checkpoint rich interfaces, while refractory cases are defined by persistent epithelial cohesion and suppressive myeloid environments. Overall, spatial compartmentalization, checkpoint expansion, and immune clustering emerged as indicators of response intensity. Our findings stress the need to tailor interventions to the spatial context; immune excluded tumors may benefit from T cell priming or myeloid reprogramming, while inflamed responders might require maintenance of checkpoint inhibition to sustain tumor clearance.