Abstract BACKGROUND: Triple-Negative Breast Cancer (TNBC) and HER2-positive (HER2+) breast cancer represent biologically distinct subtypes with differing molecular drivers, clinical behaviours, and responses to therapy. Despite these differences, both subtypes may be associated with significant tumor-stroma interactions and fibrotic remodelling, which may contribute to disease progression and therapeutic response differences. However, comparative spatial characterization of collagen architecture across different tissue compartments in these subtypes remains limited. Second harmonic generation/two-photon excitation (SHG/TPE) microscopy, coupled with AI-based image analysis, enables quantitative assessment of collagen morphometry beyond conventional histopathology. This study aims to compare region-specific collagen features between TNBC and HER2+ breast cancer biopsies and identify distinguishing fibrosis patterns across tumor and non-tumor regions. METHODS: Unstained breast cancer biopsies from TNBC (n=21) and HER2+ (n=158) patients from Dept of Surgery, National Taiwan University Hospital were imaged using SHG/TPE microscopy. Collagen morphometric features were identified using a proprietary AI-based analysis platform (Genesis®200, HistoIndex Pte Ltd, Singapore). Four anatomically distinct tissue regions were assessed: tumor, lobule and duct, stroma-fat, and stroma-fibrosis. Each region comprised of 65 collagen morphometric parameters, and was compared between TNBC and HER2+ groups within each region using normalized Median Relative Differences. Statistical comparisons were performed using the Wilcoxon rank-sum test to evaluate feature-level differences between the two subtypes. RESULTS: Region-specific collagen morphometric profiles demonstrated distinct differences between TNBC and HER2+ breast cancer biopsies. Within the tumor region, collagen parameters were largely comparable between the two groups, with no major differences observed. In the stroma-fibrosis compartment, TNBC samples showed elevations in parameters related to collagen fibres but these trends were not statistically significant. Statistically significant reductions (p 0.05) of specific collagen parameters were observed in lobule and duct regions of HER2+ biopsies versus TNBC biopsies as follows; 25% decrease in number of thick strings per unit tissue area, a 17% reduction in string perimeter, and decrease of 20% and 18% in width of all strings and width of aggregated strings per unit tissue area, respectively. CONCLUSIONS: Quantitative SHG/TPE imaging combined with AI-based collagen morphometric profiling reveals distinct fibrosis patterns between TNBC and HER2+ breast cancer subtypes. The lobule and duct region in HER2+ biopsies demonstrates a unique collagen architecture when compared with rest of TNBC biopsies. These spatially localized differences in extracellular matrix structure may reflect subtype-specific stromal remodelling and could offer utility in subtype classification, prognosis, or therapeutic response prediction. Further validation of these imaging-derived biomarkers in larger cohorts is warranted. Citation Format: K. Akbary, L. Jiaojiao, R. Yayun, D. Tai, C. Hsiao, K. Huang. Region-specific collagen morphometric parameters in HER2+ breast cancer versus TNBC: insights from SHG/TPE and AI-based 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 PS4-05-12.
Akbary et al. (2026) studied this question.