Deep learning assessment of breast terminal duct lobular unit involution: Towards automated prediction of breast cancer risk

Terminal duct lobular unit (TDLU) involution is the regression of milk-producing structures in the breast. Women with less TDLU involution are more likely to develop breast cancer. A major bottleneck in studying TDLU involution in large cohort studies is the need for labor-intensive manual assessment of TDLUs. We developed a computational pathology solution to automatically capture TDLU involution measures. Whole slide images (WSIs) of benign breast biopsies were obtained from the Nurses’ Health Study. A set of 92 WSIs was annotated for acini, TDLUs and adipose tissue to train deep convolutional neural network (CNN) models for detection of acini, and segmentation of TDLUs and adipose tissue. These networks were integrated into a single computational method to capture TDLU involution measures including number of TDLUs per tissue area, median TDLU span and median number of acini per TDLU. We validated our method on 40 additional WSIs by comparing with manually acquired measures. Our CNN models detected acini with an F1 score of 0.73±0.07, and segmented TDLUs and adipose tissue with Dice scores of 0.84±0.13 and 0.87±0.04, respectively. The inter-observer ICC scores for manual assessments on 40 WSIs of number of TDLUs per tissue area, median TDLU span, and median acini count per TDLU were 0.71, 0.81 and 0.73, respectively. Intra-observer reliability was evaluated on 10/40 WSIs with ICC scores of >0.8. Inter-observer ICC scores between automated results and the mean of the two observers were: 0.80 for number of TDLUs per tissue area, 0.57 for median TDLU span, and 0.80 for median acini count per TDLU. TDLU involution measures evaluated by manual and automated assessment were inversely associated with age and menopausal status. We developed a computational pathology method to measure TDLU involution. This technology eliminates the labor-intensiveness and subjectivity of manual TDLU assessment, and can be applied to future breast cancer risk studies.

[1]  Andrew H. Beck,et al.  Androgen receptor expression in normal breast tissue and subsequent breast cancer risk , 2018, npj Breast Cancer.

[2]  Gretchen L. Gierach,et al.  Age-related terminal duct lobular unit involution in benign tissues from Chinese breast cancer patients with luminal and triple-negative tumors , 2017, Breast Cancer Research.

[3]  R. Tarone,et al.  Involution and the etiology of breast cancer , 1994, Cancer.

[4]  G. Colditz,et al.  Radial scars and subsequent breast cancer risk: results from the Nurses’ Health Studies , 2013, Breast Cancer Research and Treatment.

[5]  T. Bevers,et al.  Breast Cancer Risk by Extent and Type of Atypical Hyperplasia: An Update From the Nurses' Health Studies , 2016 .

[6]  N F Boyd,et al.  Mammographic density, lobular involution, and risk of breast cancer , 2008, British Journal of Cancer.

[7]  G. Colditz,et al.  Expression of IGF1R in normal breast tissue and subsequent risk of breast cancer , 2010, Breast Cancer Research and Treatment.

[8]  Daniel L. Rubin,et al.  Automated Grading of Gliomas using Deep Learning in Digital Pathology Images: A modular approach with ensemble of convolutional neural networks , 2015, AMIA.

[9]  Trevor Darrell,et al.  Fully Convolutional Networks for Semantic Segmentation , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[10]  H M Jensen,et al.  An atlas of subgross pathology of the human breast with special reference to possible precancerous lesions. , 1975, Journal of the National Cancer Institute.

[11]  Neofytos Dimitriou,et al.  Deep Learning for Whole Slide Image Analysis: An Overview , 2019, Front. Med..

[12]  Claes Lundström,et al.  Towards grading gleason score using generically trained deep convolutional neural networks , 2016, 2016 IEEE 13th International Symposium on Biomedical Imaging (ISBI).

[13]  G. Colditz,et al.  Breast cancer risk by extent and type of atypical hyperplasia: An update from the Nurses' Health Studies , 2016, Cancer.

[14]  Romayne A. Thompson,et al.  Age-related lobular involution and risk of breast cancer. , 2006, Journal of the National Cancer Institute.

[15]  P. Caie,et al.  Novel histopathologic feature identified through image analysis augments stage II colorectal cancer clinical reporting , 2016, Oncotarget.

[16]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[17]  H. Jensen.  On the origin and progression of human breast cancer. , 1986, American journal of obstetrics and gynecology.

[18]  Peter D Caie,et al.  Automated Analysis of Lymphocytic Infiltration, Tumor Budding, and Their Spatial Relationship Improves Prognostic Accuracy in Colorectal Cancer , 2019, Cancer Immunology Research.

[19]  M. García-Closas,et al.  Analysis of terminal duct lobular unit involution in luminal A and basal breast cancers , 2012, Breast Cancer Research.

[20]  Nico Karssemeijer,et al.  Using deep convolutional neural networks to identify and classify tumor-associated stroma in diagnostic breast biopsies , 2018, Modern Pathology.

[21]  B. van Ginneken,et al.  Automated deep-learning system for Gleason grading of prostate cancer using biopsies: a diagnostic study. , 2020, The Lancet. Oncology.

[22]  G. Colditz,et al.  The influence of family history on breast cancer risk in women with biopsy‐confirmed benign breast disease , 2006, Cancer.

[23]  Andrew H. Beck,et al.  Systematic Analysis of Breast Cancer Morphology Uncovers Stromal Features Associated with Survival , 2011, Science Translational Medicine.

[24]  G. Colditz,et al.  Benign breast disease, recent alcohol consumption, and risk of breast cancer: a nested case–control study , 2005, Breast Cancer Research.

[25]  Karl Rohr,et al.  Predicting breast tumor proliferation from whole‐slide images: The TUPAC16 challenge , 2018, Medical Image Anal..

[26]  G. Colditz,et al.  Lobule type and subsequent breast cancer risk: Results from the Nurses' Health Studies , 2009, Cancer.

[27]  Yan Xu,et al.  Deep learning of feature representation with multiple instance learning for medical image analysis , 2014, 2014 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP).

[28]  Gretchen L. Gierach,et al.  Standardized measures of lobular involution and subsequent breast cancer risk among women with benign breast disease: a nested case–control study , 2016, Breast Cancer Research and Treatment.

[29]  J. Russo,et al.  Architectural pattern of the normal and cancerous breast under the influence of parity. , 1994, Cancer epidemiology, biomarkers & prevention : a publication of the American Association for Cancer Research, cosponsored by the American Society of Preventive Oncology.

[30]  B. van Ginneken,et al.  Deep learning as a tool for increased accuracy and efficiency of histopathological diagnosis , 2016, Scientific Reports.

[31]  Andrew H. Beck,et al.  Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer , 2017, JAMA.

[32]  Hang Chang,et al.  Unsupervised Transfer Learning via Multi-Scale Convolutional Sparse Coding for Biomedical Applications , 2018, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[33]  Mitko Veta,et al.  Mitosis Counting in Breast Cancer: Object-Level Interobserver Agreement and Comparison to an Automatic Method , 2016, PloS one.

[34]  Mitko Veta,et al.  Detection of acini in histopathology slides: towards automated prediction of breast cancer risk , 2019, Medical Imaging.

[35]  Nathalie Harder,et al.  Automatic discovery of image-based signatures for ipilimumab response prediction in malignant melanoma , 2019, Scientific Reports.

[36]  G. Colditz,et al.  Magnitude and laterality of breast cancer risk according to histologic type of atypical hyperplasia , 2007, Cancer.

[37]  Terry K Koo,et al.  A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. , 2016, Journal Chiropractic Medicine.

[38]  Stephen M. Hewitt,et al.  Quantitative analysis of TDLUs using adaptive morphological shape techniques , 2013, Medical Imaging.

[39]  Ognjen Arandjelovic,et al.  Colorectal Cancer Outcome Prediction from H&E Whole Slide Images using Machine Learning and Automatically Inferred Phenotype Profiles , 2019, ArXiv.

[40]  J. Russo,et al.  Chapter 1: Developmental, Cellular, and Molecular Basis of Human Breast Cancer , 2000 .

[41]  R. Vierkant,et al.  Novel breast tissue feature strongly associated with risk of breast cancer. , 2009, Journal of clinical oncology : official journal of the American Society of Clinical Oncology.

[42]  Francesco Ciompi,et al.  Deep learning assisted mitotic counting for breast cancer , 2019, Laboratory Investigation.

[43]  Dayong Wang,et al.  Deep Learning for Identifying Metastatic Breast Cancer , 2016, ArXiv.

[44]  H. Ashrafian,et al.  Comparative effectiveness of neoadjuvant therapy for HER2-positive breast cancer: a network meta-analysis. , 2014, Journal of the National Cancer Institute.

[45]  Nicolas Brieu,et al.  Automated tumour budding quantification by machine learning augments TNM staging in muscle-invasive bladder cancer prognosis , 2019, Scientific Reports.

[46]  Gretchen L. Gierach,et al.  Terminal duct lobular unit involution of the normal breast: implications for breast cancer etiology. , 2014, Journal of the National Cancer Institute.

[47]  Graham A. Colditz,et al.  The Nurses' Health Study: lifestyle and health among women , 2005, Nature Reviews Cancer.