Project 3: Inter-Relationships and Prognostic Significance of Breast Cancer Radiomic Risk Features, Tissue Microenvironment, and Adiposity

University Of Hawaii At Manoa, Honolulu HI

Investigators

Abstract

The risk of breast cancer dramatically differs across US populations. Nonetheless, Asian Americans and Native Hawaiians/Pacific Islanders (AANHPI) have been rarely studied in breast cancer research. Consequently, there are major gaps in understanding the basis for the high incidence and mortality among Native Hawaiians and a steadily rising incidence with comparatively favorable outcomes among Japanese Americans. Adiposity and breast density, established breast cancer risk factors, vary widely across AANHPI females and have direct implications for mammographic screening and primary prevention. Our research to date provides strong evidence that body fat distribution, including visceral adipose tissue (VAT), is an important predictor of breast cancer risk. The influence of adiposity on breast density and other aspects of breast architecture that can be discerned through mammographic screening (e.g. radiomic features) is not well understood. Our long-term goal is to elucidate the population differences in breast cancer risk and outcomes seen in AANHPI subgroups (Native Hawaiian, Micronesian, Japanese, Chinese, Filipina) that can be translated to improved prevention, early detection, and therapeutic strategies. Our central hypothesis is that established radiomic risk features have unique associations with breast cancer incidence in AANHPI subgroups and that they are correlated with tissue biomarkers of risk and prognosis and with adiposity, especially VAT. Study resources include the statewide Hawaiʻi Pacific Islands Mammography Registry linked to the SEER Hawaiʻi Tumor Registry and its Residual Tissue Repository (RTR), and to the Hawaiʻi component of the MEC Cohort Study. Our study is focused on AANHPI, with the following aims: 1) Characterize the relationships of established breast imaging radiomic risk features with tissue protein biomarker profiles reflecting the tissue microenvironment and breast cancer prognosis and with disease-specific survival; 2) Characterize the joint relationships of breast radiomic risk features and different measures of adiposity, including VAT, with post-menopausal breast cancer risk among Native Hawaiian, Japanese American, and White MEC participants. 3) Calibrate commonly used risk prediction models for breast cancer by including established breast radiomic (AI and machine learning) risk features from 2D and 3D mammography in AAPHI and Whites overall and by estrogen/progesterone receptor and HER-2. The expected outcome of the proposed study is to further our understanding of unique relationships between imaging biomarkers derived from advanced machine learning approaches and tissue molecular characteristics and adiposity phenotypes, which will improve risk and prognosis model accuracy and better identify high risk individuals for further assessment or tailored therapy.