Abstract PS3-13-22: Epithelial Podoplanin Co-expression in Postpartum Breast Cancer: An Exploratory TME Analysis

Abstract Background: Postpartum breast cancer (PPBC) - diagnosis within 10 years postpartum - shows more aggressive and ∼3-fold higher metastasis risk than nulliparous or ≥10 years postpartum cases. Human data on the PPBC tumor microenvironment (TME) are limited. We evaluated epithelial, macrophage, and endothelial wound-healing phenotypes in the PPBC TME. Objective: Identify TME phenotype(s) associated with PPBC status. Study Population: From the international Breast Cancer Family Registry (BCFR), we included 180 invasive ductal carcinoma cases diagnosed ≤50 years within the New York subset who had a H&E and an unstained slide; 31, 40, and 58 cases diagnosed <5, ≥5 - <10, ≥10 years postpartum, respectively, and 51 nulliparous cases. Methods: Multiplex immunofluorescence panel design, staining, and multispectral imaging (Opal/Vectra Polaris) were conducted blinded to patient clinicopathologic data for Podoplanin (PDPN), Macrophages (CD68), Endothelial Cells (CD31/CD34 antibody mixture), panCK, SEMA7A, PD-L1, and DAPI. Slides were scanned with a 10X objective. Tumor containing regions were annotated by pathologists; cell phenotyping, cell counts, and signal intensities were quantified in HALO (full magnification), with fluorescent intensities normalized to 0-255. Phenotypes were defined by co-/anti-coincidence logic. Within tumor regions, we computed densities (cells/mm2) for markers and 20 colocalized phenotypes, normalized each by DAPI+ nuclear density to yield unitless ratios, and log-transformed these ratios to approximate normality. Phenotypes with ≥25% zero values were dichotomized as present/absent. Continuous predictors were z-scored within cross validation (CV) folds. We defined PPBC cases as <10 years postpartum; nulliparous or ≥10 years postpartum were considered non-PPBC cases. Statistical Analysis: We examined clinicopathologic, epidemiologic, and the tumor phenotype predictor variables stratified by PPBC status (chi2, t-tests). We then fit a LASSO-penalized logistic regression; λ was tuned by repeated 5-fold CV with 5 repeats, optimizing for AUC (λ_min). Continuous predictors were standardized (z-scored) within CV folds to prevent information leakage. Binary indicators for categorical predictors remained 0/1 (not standardized). Predictors with non-zero coefficients at the selected λ comprised the active set. For interpretability, we refit an unpenalized logistic regression model on the active set (‘post-LASSO’) and reported odds ratios (OR) and 95% CIs per 1-SD increase. Sensitivity analyses (i) restricted the sample to parous participants and including parity-related predictors (e.g., breastfeeding), and (ii) redefined PPBC cases to <5 years postpartum in the post-LASSO model. Results: PPBC and non-PPBC cases were similar across predictors, except PPBC cases were younger at diagnosis (p<0.001), and among parous women, PPBC cases were older at first FTP prior to diagnosis (p<0.001). At λ_min, only age at diagnosis and PanCK+PDPN+ were retained. Macrophage- and endothelial/lymphatic-centric phenotypes shrank to zero. In post-LASSO models, age at diagnosis was inversely associated with PPBC (ORper 1-SD 0.39, 95% CI (0.26, 0.58) and PanCK+PDPN+ was positively associated (ORper 1-SD 1.72, 95% CI (1.21, 2.52)). When restricted to the parous subsample, PanCK+PDPN+ was retained by LASSO (post-Lasso ORper 1-SD 2.04, 95% CI (1.11, 4.16)); however, when redefining PPBC cases to <5 years postpartum, PanCK+PDPN+ was null (post-Lasso ORper 1-SD 0.93, 95% CI (0.47, 2.04)). Conclusion: PPBC tumors showed enrichment of PanCK+PDPN+ epithelial cells, consistent with epithelial PDPN upregulation seen in wound-healing-like programs. PanCK+PDPN+ was retained among parous women, but not within a more conservative definition of PPBC. Findings are exploratory and warrant validation. Citation Format: J. A. McDonald, A. J. Kociolek, S. Tun Hein Aung, Y. Liao, B. Vasikj Bjelogrlic, C. Loomis, J. M. Genkinger, M. Terry, K. L. Gardner. Epithelial Podoplanin Co-expression in Postpartum Breast Cancer: An Exploratory TME Analysis [abstract]. In: Proceedings of the San Antonio Breast Cancer Symposium 2025; 2025 Dec 9-12; San Antonio, TX. Philadelphia (PA): AACR; Clin Cancer Res 2026;32(4 Suppl):Abstract nr PS3-13-22.