MicroRNA spatial profiling for assessing drug efficacy in <i>BRCA1</i> -related triple-negative breast tumors

<i>BRCA1/2</i> -mutated breast cancers exhibit homologous recombination deficiency (HRD), making them initially sensitive to poly(ADP-ribose) polymerase (PARP) inhibitors. However, 40-70% of patients develop resistance, necessitating combination strategies and predictive biomarkers. We first investigated approaches to overcome PARP resistance and then explored spatial microRNA (miRNA) profiling as a prognostic tool. Using the <i>K14-Cre Brca1 ^f/f Trp53 ^f/f</i> model with tumors that acquired PARP resistance, we evaluated PARP inhibitor combinations with either PI3K inhibition or Poly(I:C) <i>in vivo</i> . Both combinations improved antitumor activity compared to PARP inhibition alone. Next, to predict resistance we applied a sensitive assay that quantifies and spatially profiles miRNA expression <i>in situ</i> onto FFPE sections from tumors treated for 10 days using nanoliter well arrays with functionalized hydrogel posts. We developed a spatial miRNA analysis framework integrating latent Dirichlet allocation (LDA) and principal component analysis (PCA) to develop "topics" that stratify early tumors as either PARP inhibitor-sensitive or - resistant and distinguish their treatment regimens. We also incorporated immune architecture using Structural Similarity Index Measure (SSIM) maps that revealed co-localization of immune infiltration and miRNA topics. This integrative approach highlights how miRNA-based spatial analysis can predict PARP inhibitor resistance and provide a promising biomarker to inform therapeutic strategies for <i>BRCA1/2-</i> related breast cancers.