Dataset Information

Deep Learning Image Analysis of Benign Breast Disease to Identify Subsequent Risk of Breast Cancer.

ABSTRACT:

Background

New biomarkers of risk may improve breast cancer (BC) risk prediction. We developed a computational pathology method to segment benign breast disease (BBD) whole slide images into epithelium, fibrous stroma, and fat. We applied our method to the BBD BC nested case-control study within the Nurses' Health Studies to assess whether computer-derived tissue composition or a morphometric signature was associated with subsequent risk of BC.

Methods

Tissue segmentation and nuclei detection deep-learning networks were established and applied to 3795 whole slide images from 293 cases who developed BC and 1132 controls who did not. Percentages of each tissue region were calculated, and 615 morphometric features were extracted. Elastic net regression was used to create a BC morphometric signature. Associations between BC risk factors and age-adjusted tissue composition among controls were assessed using analysis of covariance. Unconditional logistic regression, adjusting for the matching factors, BBD histological subtypes, parity, menopausal status, and body mass index evaluated the relationship between tissue composition and BC risk. All statistical tests were 2-sided.

Results

Among controls, direction of associations between BBD subtypes, parity, and number of births with breast composition varied by tissue region; select regions were associated with childhood body size, body mass index, age of menarche, and menopausal status (all P < .05). A higher proportion of epithelial tissue was associated with increased BC risk (odds ratio = 1.39, 95% confidence interval = 0.91 to 2.14, for highest vs lowest quartiles, P _trend = .047). No morphometric signature was associated with BC.

Conclusions

The amount of epithelial tissue may be incorporated into risk assessment models to improve BC risk prediction.

SUBMITTER: Vellal AD

PROVIDER: S-EPMC7898083 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:Lesser degrees of terminal duct-lobular unit (TDLU) involution predict higher breast cancer risk; however, standardized measures to quantitate levels of TDLU involution have only recently been developed. We assessed whether three standardized measures of TDLU involution, with high intra/inter pathologist reproducibility in normal breast tissue, predict subsequent breast cancer risk among women in the Mayo benign breast disease (BBD) cohort. We performed a masked evaluation of biopsies from 99 women with BBD who subsequently developed breast cancer (cases) after a median of 16.9 years and 145 age-matched controls. We assessed three metrics inversely related to TDLU involution: TDLU count/mm(2), median TDLU span (microns, which approximates acini content), and median category of acini counts/TDLU (0-10; 11-20; 21-30; 31-50; >50). Associations with subsequent breast cancer risk for quartiles (or categories of acini counts) of each of these measures were assessed with multivariable conditional logistic regression to estimate odds ratios (ORs) and 95 % confidence intervals (CI). In multivariable models, women in the highest quartile compared to the lowest quartiles of TDLU counts and TDLU span measures were significantly associated with subsequent breast cancer diagnoses; TDLU counts quartile4 versus quartile1, OR = 2.44, 95 %CI 0.96-6.19, p-trend = 0.02; and TDLU spans, quartile4 versus quartile1, OR = 2.83, 95 %CI = 1.13-7.06, p-trend = 0.03. Significant associations with categorical measures of acini counts/TDLU were also observed: compared to women with median category of <10 acini/TDLU, women with >25 acini counts/TDLU were at significantly higher risk, OR = 3.40, 95 %CI 1.03-11.17, p-trend = 0.032. Women with TDLU spans and TDLU count measures above the median were at further increased risk, OR = 3.75 (95 %CI 1.40-10.00, p-trend = 0.008), compared with women below the median for both of these metrics. Similar results were observed for combinatorial metrics of TDLU acini counts/TDLU, and TDLU count. Standardized quantitative measures of TDLU counts and acini counts approximated by TDLU span measures or visually assessed in categories are independently associated with breast cancer risk. Visual assessment of TDLU numbers and acini content, which are highly reproducible between pathologists, could help identify women at high risk for subsequent breast cancer among the million women diagnosed annually with BBD in the US.

Project description:Background: Fibrocystic changes are associated with an increased risk of breast cancer. Genetic alterations have been found in fibrocystic changes with or without epithelial changes, suggesting that critical oncogenic events are occurring at an early stage. Methods: We investigated a unique collective of 17 breast cancer patients who, prior to the diagnosis of invasive breast cancer, underwent open surgical biopsy showing fibrocystic changes of the breast. The time span between biopsy for fibrocystic changes and invasive carcinoma ranged from 1 to 11 years (average 5.3 years). Ten (58.8%) of the patients had an ipsilateral invasive carcinoma, and 7 (41.2%) of the patients developed an invasive carcinoma of the contralateral breast. Massive parallel sequencing targeting genes frequently mutated in breast cancer was performed on the fibrocystic breast tissue as well as the ensuing cancer tissue. Results: In 9 cases, somatic mutations were found in the tumor tissue, the most prevalent being PIK3CA mutations (n = 4), followed by TP53 mutations (n = 2). None of these mutations were present in the previously removed mastopathy tissue. In one of the cases, an ERBB3 E928G mutation was present in the mastopathy as well as in the tumor tissue, with the variant allele frequency in the mastopathy being <0.1%. In two patients, we found two mutations (MAP3K1 L380fs and PIK3CA I391M, respectively) present in the mastopathy as well as in the subsequent breast cancer. These two mutations, however, could also be due to fixation artifacts. Conclusion: Since no significant somatic mutations in the fibrocystic breast tissue, and only doubtful shared mutations between benign and associated cancer tissue were detected, it remains unclear why women with fibrocystic breast disease have a statistically significant increased risk of breast cancer. Further analyses, maybe on the level of gene expression, could help to clarify the role of these benign alterations in the development of breast cancer and help to identify women at greater risk of developing subsequent invasive cancer.