Unsupervised representation learning of chromatin images identifies changes in cell state and tissue organization in DCIS

• Published in: Nature communications 2024-07, Vol.15 (1), p.6112-16, Article 6112
• Main Authors: Zhang, Xinyi, Venkatachalapathy, Saradha, Paysan, Daniel, Schaerer, Paulina, Tripodo, Claudio, Uhler, Caroline, Shivashankar, G. V.
• Format: Article
• Language: English
• Subjects: 13/1; 14; 14/19; 631/114/1305; 631/67/1347; 631/67/2321; 631/67/327; Biomarkers; Biomarkers, Tumor - genetics; Biomarkers, Tumor - metabolism; Breast cancer; Breast carcinoma; Breast Neoplasms - genetics; Breast Neoplasms - metabolism; Breast Neoplasms - pathology; Carcinoma, Intraductal, Noninfiltrating - genetics; Carcinoma, Intraductal, Noninfiltrating - metabolism; Carcinoma, Intraductal, Noninfiltrating - pathology; Chromatin; Chromatin - metabolism; Female; Humanities and Social Sciences; Humans; Image Processing, Computer-Assisted - methods; Invasiveness; Learning; Medical imaging; Morphology; multidisciplinary; Neoplasm Staging; Next-generation sequencing; Representations; Science; Science (multidisciplinary); Tissue Array Analysis; Tissues; Tumors; Unsupervised Machine Learning
• ISSN: 2041-1723; 2041-1723
• DOI: 10.1038/s41467-024-50285-1

Ductal carcinoma in situ (DCIS) is a pre-invasive tumor that can progress to invasive breast cancer, a leading cause of cancer death. We generate a large-scale tissue microarray dataset of chromatin images, from 560 samples from 122 female patients in 3 disease stages and 11 phenotypic categories. Using representation learning on chromatin images alone, without multiplexed staining or high-throughput sequencing, we identify eight morphological cell states and tissue features marking DCIS. All cell states are observed in all disease stages with different proportions, indicating that cell states enriched in invasive cancer exist in small fractions in normal breast tissue. Tissue-level analysis reveals significant changes in the spatial organization of cell states across disease stages, which is predictive of disease stage and phenotypic category. Taken together, we show that chromatin imaging represents a powerful measure of cell state and disease stage of DCIS, providing a simple and effective tumor biomarker. Monitoring the hallmarks and progression of ductal carcinoma in situ (DCIS) remains challenging. Here, the authors use an unsupervised representation learning approach on chromatin images to discern multiple morphological cell states and tissue features in DCIS.