Identification and molecular characterisation of high-risk pre-malignant breast lesions

© 2019 Tanjina KaderMammographic screening has led to an increased detection of in situ and invasive breast carcinoma. However, lesions with uncertain malignant potential (B3) are also frequently detected with routine mammographic screening. They are recognised as B3 due to their unpredictable but significant association with malignancy either on subsequent excision (i.e. upgrade) or later development of cancer. Atypical ductal hyperplasia (ADH) has long been thought of as a direct precursor of only low-grade (LG) neoplasia pathway, whereas breast papillary lesions (PL) are thought to be a risk factor for breast carcinoma. However, we have a limited understanding of how breast cancer progresses from these early lesions. Additionally, other B3 lesions are often misdiagnosed as ADH due to the subjectivity of pathologists’ criteria, therefore, unnecessary surgical excision is often recommended. Conversely, 10-15% of ADH patients subsequently develop cancer despite being surgically excised. Currently, there is no biomarker for accurate risk prediction of later cancer or upgrade for any of these B3 lesions. The first chapter of this thesis describes a novel method developed to utilise low-input DNA for low-coverage whole genome sequencing (LCWGS) for copy number (CN) profiling, which enabled us to study more ADH cases than previously possible. The second study showed that ADH can be not only a direct precursor of the LG- breast cancer pathway, but also could progress to any grades of ER+ cancer, including high grade (HG), ER- and ERBB2 amplified cancer. To our knowledge, this multipotent nature of ADH with the inclusion of specific initiating CN events for LG and HG pathways was never shown previously. This study also identified a possible progression biomarker based on seven CN loci for individual risk prediction as well as for upgrade. The third study showed in detail that breast PL could be a direct precursor of any grade of breast carcinoma. No previous study has combined the mutational landscape and CN profile to identify the early clonal expansion for these lesions. It was shown in this study that in the absence of PIK3CA mutation, there were three specific CN aberrations (16q loss/1q gain or 11q loss) suggested as cancer-associated, which could be informative in the clinical setting for accurate risk prediction for both upgrades and later development of cancer. Collectively, this thesis aimed at understanding breast cancer progression and has been able to suggest a revised breast cancer progression model including ADH and PL as precursors of both LG and HG breast cancer pathways. This knowledge could be used in the clinical setting in future for personalising management of the patients diagnosed with these B3 lesions in core biopsies. If validated in independent cohorts these potential biomarkers will be developing into a diagnostic assay, which may help to reduce unnecessary surgeries for patients by identifying the group of patients at “low-risk” for developing cancer