Correlated metabolomic, genomic, and histologic phenotypes in histologically normal breast tissue.
Xuezheng SunDelisha A StewartRupninder SandhuErin L KirkWimal W PathmasiriSusan L McRitchieRobert F ClarkMelissa A TroesterSusan J SumnerPublished in: PloS one (2018)
Breast carcinogenesis is a multistep process accompanied by widespread molecular and genomic alterations, both in tumor and in surrounding microenvironment. It is known that tumors have altered metabolism, but the metabolic changes in normal or cancer-adjacent, nonmalignant normal tissues and how these changes relate to alterations in gene expression and histological composition are not well understood. Normal or cancer-adjacent normal breast tissues from 99 women of the Normal Breast Study (NBS) were evaluated. Data of metabolomics, gene expression and histological composition was collected by mass spectrometry, whole genome microarray, and digital image, respectively. Unsupervised clustering analysis determined metabolomics-derived subtypes. Their association with genomic and histological features, as well as other breast cancer risk factors, genomic and histological features were evaluated using logistic regression. Unsupervised clustering of metabolites resulted in two main clusters. The metabolite differences between the two clusters suggested enrichment of pathways involved in lipid metabolism, cell growth and proliferation, and migration. Compared with Cluster 1, subjects in Cluster 2 were more likely to be obese (body mass index ≥30 kg/m2, p<0.05), have increased adipose proportion (p<0.01) and associated with a previously defined Active genomic subtype (p<0.01). By the integrated analyses of histological, metabolomics and transcriptional data, we characterized two distinct subtypes of non-malignant breast tissue. Further research is needed to validate our findings, and understand the potential role of these alternations in breast cancer initiation, progression and recurrence.
Keyphrases
- gene expression
- mass spectrometry
- copy number
- body mass index
- risk factors
- dna methylation
- adipose tissue
- machine learning
- squamous cell carcinoma
- electronic health record
- single cell
- physical activity
- deep learning
- risk assessment
- transcription factor
- bariatric surgery
- weight loss
- gas chromatography
- obese patients
- genome wide
- heat shock
- breast cancer risk