2026 Proffered Presentations
S162: SPATIAL IMMUNE CORRELATES FROM THE BARON TRIAL: MULTIPLEX IMMUNOFLUORESCENCE OF OLFACTORY NEUROBLASTOMA TREATED WITH BINTRAFUSP ALFA
Shrey B Shah1; Shivani Ramolia, MD1; Yvette L Robbins, PhD2; Wiem Lassoued, PhD3; Charalampos S Floudas3; James L Gulley3; Gary L Gallia4; Clint T Allen2; Nyall R London Jr., MD1; 1Sinonasal and Skull Base Tumor Program, Surgical Oncology Program, Center for Cancer Research, National Cancer Institute, National Institutes of Health; 2Section on Translational Tumor Immunology, Center for Cancer Research, National Cancer Institute, National Institutes of Health; 3Center for Immuno-Oncology, Center for Cancer Research, National Cancer Institute, National Institutes of Health; 4Department of Otolaryngology-Head and Neck Surgery, Johns Hopkins University School of Medicine
Background: Olfactory neuroblastoma (ONB) is a rare malignancy of the anterior skull base with poor outcomes in the recurrent/metastatic (R/M) setting. Prior studies have suggested that ONB may express PD-L1 and that patients with recurrent ONB possess increased TGF-β pathway activity. This provided a rationale for the treatment of patients with R/M ONB with bintrafusp alfa, a bifunctional anti-PD-L1/TGF-β trap. Our goal was to determine whether differences in the T cell tumor immune microenvironment or TGF-β signaling correlated with R/M ONB response to bintrafusp alfa.
Methods: Pre-treatment tumor tissue from 11 BARON trial participants underwent multiplex immunofluorescence (mIF) with two panels: T-cell (CD4, CD8, FoxP3, Ki67, PD-1), and TGF-β signaling (pSMAD2, pSMAD3, p21, Ki67, vimentin). For each participant, 3–5 spatially distinct samples were segmented into tumor and stroma and quantified with HALO (Indica Labs). Predefined tumor-compartment features included CD8+/CD4+ subsets (including Ki67+PD-1− cytotoxic T cells and FoxP3+ Tregs) and pSMAD H-scores. An exploratory analysis compared baseline, tumor-compartment features, and responders (stable disease (non-PD); n=5) versus progressive disease (PD) (n=5) using two-sided Mann–Whitney U tests at the region level.
Results: Across tumor parenchyma regions, responders displayed a trend towards greater parenchymal cytotoxic T-cell features and less exhaustion/immunosuppression than PD. Relative to PD, responders showed ~1.9-fold higher CD8+ T cells (p=0.34) and ~2.4-fold higher Ki67+PD-1− cytotoxic T cells (p=0.23), with ~10-fold fewer PD-1+ T cells (p=0.19) and ~65% fewer FoxP3+ Tregs (p=0.09) (Figure 1). TGF-β pathway activity differed: pSMAD2 H-score was ~50% lower in non-PD versus PD (p=2×10^−5), with lower Ki67 as well (p=0.03). Furthermore, pSMAD3+ (p=0.16) and pSMAD3 H-Score levels (p=0.005) were more variable but were also greater in PD when compared to non-PD. Inversely, pSMAD3+pSMAD2- cells were found to be significantly higher in responder tumor regions when compared to PD tumor regions (p=0.001) (Figure 2). Multispectral immunofluorescence analysis for MHC and myeloid panels are completed and currently being analyzed.
Conclusions: Patients who responded to bintrafusp alfa were characterized by a non-statistically significant trend towards higher CD8+ and lower FoxP3+ Treg infiltration. Differences in TGF-β signaling were observed, with pSMAD2+ activity significantly higher in non-responders, whereas pSMAD3+pSMAD2- cells were significantly higher in responders. The significance of this difference is unclear and may suggest complex or alternate TGF-β pathway activation in progressive disease. These findings, while in a small cohort, underscore that a robust pre-existing anti-tumor immune milieu may be associated with response to dual TGF-β/PD-L1 blockade in ONB. Validation in larger studies is warranted to confirm these biomarkers as predictors of response to dual TGF-β/PD-L1 blockade.
Figure 1. T Cell Panel with the markers CD4, CD8, FoxP3, Ki67, PD-1.
Figure 2. TGF β Panel with the markers pSMAD2, pSMAD3, p21, Ki67, and Vimentin.