An official website of the United States government
Abstract Breast carcinoma is the most common cancer among women worldwide that consists of a heterogeneous group of subtype diseases. The whole-slide images (WSIs) can capture the cell-level heterogeneity, and are routinely used for cancer diagnosis by pathologists. However, key driver genetic mutations related to targeted therapies are identified by genomic analysis like high-throughput molecular profiling. In this study, we develop a deep-learning model to predict the genetic mutations and biological pathway activities directly from WSIs. Our study offers unique insights into WSI visual interactions between mutation and its related pathway, enabling a head-to-head comparison to reinforce our major findings. Using the histopathology images from the Genomic Data Commons Database, our model can predict the point mutations of six important genes (AUC 0.68–0.85) and copy number alteration of another six genes (AUC 0.69–0.79). Additionally, the trained models can predict the activities of three out of ten canonical pathways (AUC 0.65–0.79). Next, we visualized the weight maps of tumor tiles in WSI to understand the decision-making process of deep-learning models via a self-attention mechanism. We further validated our models on liver and lung cancers that are related to metastatic breast cancer. Our results provide insights into the association between pathological image features, molecular outcomes, and targeted therapies for breast cancer patients.
more »
« less
This content will become publicly available on July 1, 2025
Visual Intratumor Heterogeneity and Breast Tumor Progression
High intratumoral heterogeneity is thought to be a poor prognostic indicator. However, the source of heterogeneity may also be important, as genomic heterogeneity is not always reflected in histologic or ‘visual’ heterogeneity. We aimed to develop a predictor of histologic heterogeneity and evaluate its association with outcomes and molecular heterogeneity. We used VGG16 to train an image classifier to identify unique, patient-specific visual features in 1655 breast tumors (5907 core images) from the Carolina Breast Cancer Study (CBCS). Extracted features for images, as well as the epithelial and stromal image components, were hierarchically clustered, and visual heterogeneity was defined as a greater distance between images from the same patient. We assessed the association between visual heterogeneity, clinical features, and DNA-based molecular heterogeneity using generalized linear models, and we used Cox models to estimate the association between visual heterogeneity and tumor recurrence. Basal-like and ER-negative tumors were more likely to have low visual heterogeneity, as were the tumors from younger and Black women. Less heterogeneous tumors had a higher risk of recurrence (hazard ratio = 1.62, 95% confidence interval = 1.22–2.16), and were more likely to come from patients whose tumors were comprised of only one subclone or had a TP53 mutation. Associations were similar regardless of whether the image was based on stroma, epithelium, or both. Histologic heterogeneity adds complementary information to commonly used molecular indicators, with low heterogeneity predicting worse outcomes. Future work integrating multiple sources of heterogeneity may provide a more comprehensive understanding of tumor progression.
more »
« less
- PAR ID:
- 10516501
- Publisher / Repository:
- MDPI
- Date Published:
- Journal Name:
- Cancers
- Volume:
- 16
- Issue:
- 13
- ISSN:
- 2072-6694
- Page Range / eLocation ID:
- 2294
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)Central nervous system (CNS) tumors come with vastly heterogeneous histologic, molecular, and radiographic landscapes, rendering their precise characterization challenging. The rapidly growing fields of biophysical modeling and radiomics have shown promise in better characterizing the molecular, spatial, and temporal heterogeneity of tumors. Integrative analysis of CNS tumors, including clinically acquired multi-parametric magnetic resonance imaging (mpMRI) and the inverse problem of calibrating biophysical models to mpMRI data, assists in identifying macroscopic quantifiable tumor patterns of invasion and proliferation, potentially leading to improved ( a) detection/segmentation of tumor subregions and ( b) computer-aided diagnostic/prognostic/predictive modeling. This article presents a summary of ( a) biophysical growth modeling and simulation,( b) inverse problems for model calibration, ( c) these models' integration with imaging workflows, and ( d) their application to clinically relevant studies. We anticipate that such quantitative integrative analysis may even be beneficial in a future revision of the World Health Organization (WHO) classification for CNS tumors, ultimately improving patient survival prospects.more » « less
-
Colorectal cancer (CRC) is the third-most leading cause of cancer-related deaths in the United States. To advance the understanding of CRC tumor progression, models which mimic the tumor microenvironment (TME) and have translatable study outcomes are urgently needed. CRC patient-derived xenografts (PDXs) are promising tools for their ability to recapitulate tumor heterogeneity and key patient tumor characteristics, such as molecular characteristics. However, as in vivo models, CRC PDXs are costly and low-throughput, which leads to a need for equivalent in vitro models. To address this need, we previously established an in vitro model using a tissue engineering toolset with CRC PDX cells. However, it is unclear whether tissue engineering has the capacity to maintain patient- and/or cancer stage-specific tumor heterogeneity. To address this gap, we employed three PDX tumor lines, originated from stage II, III-B, and IV CRC tumors, in the formation of 3D engineered CRC PDX (3D-eCRC-PDX) tissues and performed an in-depth comparison between the 3D-eCRC-PDX tissues and the original CRC-PDX tumors. To form the tissues, CRC-PDX tumors were expanded in vivo and dissociated. The isolated cells were encapsulated within poly(ethylene glycol)-fibrinogen hydrogels and remained viable and proliferative post encapsulation over the course of 29 days in culture. To gain molecular insight into the maintenance of PDX line stage heterogeneity, we performed a transcriptomic analysis using RNA seq to determine the extent to which there were similarities and differences between the CRC-PDX tumors and the 3D-eCRC-PDX tissues. We observed the greatest correspondence in overlapping differentially expressed human genes, gene ontology, and Hallmark gene set enrichment between the 3D-eCRC-PDX tissues and CRC-PDX tumors in the stage II PDX line, while the least correspondence was observed in the stage IV PDX line. The Hallmark gene set enrichment from murine mapped RNA seq transcripts was PDX line-specific which suggested that the stromal component of the 3D-eCRC-PDX tissues was maintained in a PDX line-dependent manner. Consistent with our transcriptomic analysis, we observed that tumor cell subpopulations, including human proliferative (B2M+Ki67+) and CK20+ cells, remained constant for up to 15 days in culture even though the number of cells in the 3D-eCRC-PDX tissues from all three CRC stages increased over time. Yet, tumor cell subpopulation differences in the stage IV 3D-eCRC-PDX tissues were observed starting at 22 days in culture. Overall, our results demonstrate a strong correlation between our in vitro 3D-eCRC-PDX models and the originating in vivo CRC-PDX tumors, providing evidence that these engineered tissues may be capable of mimicking patient- and/or cancer stage-specific heterogeneity.more » « less
-
Background: We investigated the association between reproductive risk factors and breast cancer subtype in Black women. On the basis of the previous literature, we hypothesized that the relative prevalence of specific breast cancer subtypes might differ according to reproductive factors. Methods: We conducted a pooled analysis of 2,188 (591 premenopausal, 1,597 postmenopausal) Black women with a primary diagnosis of breast cancer from four studies in the southeastern United States. Breast cancers were classified by clinical subtype. Case-only polytomous logistic regression models were used to estimate ORs and 95% confidence intervals (CI) for HER2+ and triple-negative breast cancer (TNBC) status in relation to estrogen receptor–positive (ER+)/HER2− status (referent) for reproductive risk factors. Results: Relative to women who had ER+/HER2− tumors, women who were age 19–24 years at first birth (OR, 1.78; 95% CI, 1.22–2.59) were more likely to have TNBC. Parous women were less likely to be diagnosed with HER2+ breast cancer and more likely to be diagnosed with TNBC relative to ER+/HER2− breast cancer. Postmenopausal parous women who breastfed were less likely to have TNBC [OR, 0.65 (95% CI, 0.43–0.99)]. Conclusions: This large pooled study of Black women with breast cancer revealed etiologic heterogeneity among breast cancer subtypes. Impact: Black parous women who do not breastfeed are more likely to be diagnosed with TNBC, which has a worse prognosis, than with ER+/HER2− breast cancer.more » « less
-
Breast cancer treatment can be improved with biomarkers for early detection and individualized therapy. A set of 86 microRNAs (miRNAs) were identified to separate breast cancer tumors from normal breast tissues (n = 52) with an overall accuracy of 90.4%. Six miRNAs had concordant expression in both tumors and breast cancer patient blood samples compared with the normal control samples. Twelve miRNAs showed concordant expression in tumors vs. normal breast tissues and patient survival (n = 1093), with seven as potential tumor suppressors and five as potential oncomiRs. From experimentally validated target genes of these 86 miRNAs, pan-sensitive and pan-resistant genes with concordant mRNA and protein expression associated with in-vitro drug response to 19 NCCN-recommended breast cancer drugs were selected. Combined with in-vitro proliferation assays using CRISPR-Cas9/RNAi and patient survival analysis, MEK inhibitors PD19830 and BRD-K12244279, pilocarpine, and tremorine were discovered as potential new drug options for treating breast cancer. Multi-omics biomarkers of response to the discovered drugs were identified using human breast cancer cell lines. This study presented an artificial intelligence pipeline of miRNA-based discovery of biomarkers, therapeutic targets, and repositioning drugs that can be applied to many cancer types.more » « less
