Mutational Profiling Of Premalignant Breast Microbiopsies.

Abstract Ductal carcinoma in situ (DCIS) comprises ~20% of all breast cancer diagnoses in the United States. Nearly all DCIS are treated to prevent progression to potentially lethal invasive cancer, yet it is estimated about half of DCIS progress within 10 years making the systematic treatment controversial. This highlights an urgent need for reliable molecular markers to stratify patients by risk. Small, degraded biopsies have made molecular profiling historically challenging; furthermore, a single biopsy can have several regions varying in nuclear grade, hormone receptor status, and histological architecture requiring laser-capture microdissection. Recent advances in whole-exome sequencing now enable mutational profiling from limited quantities (<10ng) of damaged, formalin-fixed, paraffin-embedded (FFPE) DNA, and present an opportunity to sequence individual regions and relate the phenotypic heterogeneity with genetic alterations. Here, we apply these advances to obtain mutational profiles from pure DCIS microbiopsies, enabling a comprehensive molecular profile of pure DCIS, direct measurement of genetic heterogeneity, and associations between molecular and histopathological features. Whole-exome sequencing was performed on 59 regions across 31 pure DCIS cases varying in nuclear grade (including 19% low-grade), and histological architecture (52% cribriform, 39% solid). Driver analysis revealed breast cancer-like driver alterations in 30/31 cases, and evidence for multiple putative drivers in 26/31, even in low-grade DCIS. The most frequently mutated genes across patients were common breast cancer driver genes: PIK3CA (37%), TP53 (22%), and GATA3 (11%). Phylogenetic trees constructed on multi-region biopsies revealed that regions of the same biopsy shared 12-336 alterations, and all pathogenic driver point mutations. Nevertheless, 1-22 region-specific alterations were found in each biopsy, some of which were likely driver CNA. We identified several histological associations with genetic alterations including high grade (p=2x10-4) and solid DCIS (p=3x10-4) associated with high copy number burden, and a non-significant increase in GATA3 (13% vs 0%) and PIK3CA (50% vs 28%) pathogenic mutations in cribriform vs solid architecture. Lastly, the most genetically diverse DCIS were HER2+/ER- (p=0.07) and those displaying necrosis (p=0.006). Overall, using novel methodology we performed detailed multi-region mutational profiling on difficult to sequence precancer lesions, revealing varying genetic diversity within a biopsy and novel associations between histology and underlying genetic landscape. We illustrate multiple driver genetic alterations and diversity present even in low-grade, pure DCIS lesions, providing key insight into early breast carcinogenesis and representing a critical step towards the development of prognostic markers of progression. Citation Format: Daniela Nachmanson, Mark F. Evans, Joseph Steward, Adam Officer, Huazhen Yao, Thomas J. O'Keefe, Farnaz Hasteh, Gary S. Stein, Kristen Jepsen, Donald L. Weaver, Gillian L. Hirst, Brian L. Sprague, Laura J. Esserman, Alexander Borowsky, Janet L. Stein, Olivier Harismendy. Mutational profiling of premalignant breast microbiopsies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 2176.