Spatiotemporal evolution of immune evasion: Divergent resistance mechanisms in primary refractory versus new primary cutaneous squamous cell carcinomas during anti-PD1 therapy.

e21553 Background: While anti-PD1 therapy is effective in advanced cutaneous squamous cell carcinoma (cSCC), ~50% of patients exhibit primary resistance. A distinct, paradoxical clinical phenomenon involves emergence of new primary tumors arising during therapy (NPADTs), representing de novo malignancies evolving under systemic immune pressure. We hypothesized that these disparate clinical failures represent fundamentally divergent evolutionary trajectories, requiring distinct therapeutic solutions. Methods: We performed an integrated single-cell RNA sequencing (Chromium Flex) and subcellular-resolution spatial transcriptomics (Nanostring CosMx) analysis on a cohort of 14 patients with cSCC treated with cemiplimab or pembrolizumab [primary responders (n = 4), primary non-responders (n = 5), NPADTs (n = 5), healthy skin (n = 7)]. The dataset comprised > 110,000 single cells. Cellular neighborhoods were mapped via spatial transcriptomics integrated with multiplex immunofluorescence. Resistance archetypes were defined using differential composition analysis (Bayesian framework using sccomp to control for biological variability and compositionality), pathway inference (PROGENy), and spatial neighborhood analysis. Results: Transcriptional profiling revealed two distinct resistance phenotypes. Primary resistant tumors are ‘immune-excluded’, characterized by a paucity of intratumoral CD8+ T cells (differential compositional analysis using sccomp; False Discovery Rate < 0.05) and stromal fibrosis. This exclusion was driven by TGF-ß pathway activation and enrichment of myofibroblastic cancer-associated fibroblasts (myCAFs: ACTA2+COL1A1+ ), forming a barrier to T-cell ingress. In contrast, NPADTs exhibited an ‘immune-dysfunctional’ phenotype. Despite high infiltration of CD8+ T cells comparable to responders, these tumors remained refractory. Mechanistic analysis identified a chronic Type I Interferon signature driven by plasmacytoid dendritic cells (pDC). This chronic inflammatory state was associated with expansion of immunoregulatory mature dendritic cells (mregDCs; LAMP3+BIRC3+IDO1+ ). Spatial analysis revealed that mregDCs function as cellular hubs, recruiting FOXP3+ regulatory T cells (Tregs). We observed a significant spatial correlation between mregDCs and Tregs (Pearson r > 0.5), establishing localized immunosuppressive niches that neutralized effector T cells. Conclusions: This study delineates divergent paths for immunotherapy resistance in cSCC: structural exclusion driven by myCAFs in primary resistance versus functional suppression driven by a pDC-IFN-mregDC axis in NPADTs. These findings offer a precision medicine framework for sequencing therapies in cSCCs.