An integrative analysis of cancer gene expression studies using Bayesian latent factor modeling

Abstract

We present an applied study in cancer genomics for integrating data and inferences from laboratory experiments on cancer cell lines with observational data obtained from human breast cancer studies. The biological focus is on improving understanding of transcriptional responses of tumors to changes in the pH level of the cellular microenvironment. The statistical focus is on connecting experimentally defined biomarkers of such responses to clinical outcome in observational studies of breast cancer patients. Our analysis exemplifies a general strategy for accomplishing this kind of integration across contexts. The statistical methodologies employed here draw heavily on Bayesian sparse factor models for identifying, modularizing and correlating with clinical outcome these signatures of aggregate changes in gene expression. By projecting patterns of biological response linked to specific experimental interventions into observational studies where such responses may be evidenced via variation in gene expression across samples, we are able to define biomarkers of clinically relevant physiological states and outcomes that are rooted in the biology of the original experiment. Through this approach we identify microenvironment-related prognostic factors capable of predicting long term survival in two independent breast cancer datasets. These results suggest possible directions for future laboratory studies, as well as indicate the potential for therapeutic advances though targeted disruption of specific pathway components.

Investigating Effects of Acidic pH on Proliferation, Invasion and Drug-Induced Apoptosis in Lymphoblastic Leukemia

“…Cathepsin L also amplifies the proteinase cascade through the activation of urokinase-type plasminogen activator (uPA) which is critical for metastatic cascade in tumor cells [26]. MMP-9 plays a vital role in promoting tumor invasion and tissue remodeling, and participates in the release of cell surface precursorform of many growth factors by proteolysis of several ECM components [2,[29][30][31]. Glunde et al (2003) reported filopodia formation which was observed more frequently in highly metastatic breast cancer cells maintained in acidic pH [32].…”

“…Microenvironmental factors encompass all the physical interactions between cells, extracellular matrix and biochemical factors such as different types of nutrients [2].…”

Latent Factor Analysis to Discover Pathway-Associated Putative Segmental Aneuploidies in Human Cancers

“…The combination of signature-derived latent factors has been shown to improve the prognostic value and predictive power of the original hypoxia and lactic acidosis signatures [26]. A recent analysis of the prognostic factors in breast cancer has highlighted the importance of a glycolytic enzyme PGK1 and tumor glycolysis in the prognostic value of stress signatures [64]. In addition, factor analysis has allowed the molecular dissection of the Ras pathways into individual components [14].…”

Lactic Acidosis Triggers Starvation Response with Paradoxical Induction of TXNIP through MondoA

“…Gene expression signatures associated with 1,4,12, and 24 hour lactic acidosis treatments were derived from RMA expression values using standard Bayesian sparse multivariate regression techniques, full details of which appear in previous publications [18], [115]–[117] For each treatment group (i.e. 1-,4-,12-, and 24-hr lactic acidosis treatment), this estimated the probability of differential expression for each Affymetrix probe as well as the fold change of the differentially expressed transcripts, as compared to the control.…”

“…These values define the in-vitro signature of the treatment. Projection of these signatures into the primary tumor data sets is derived from the weighted inner product of the vector of tumor gene expression values and the vector of regression coefficients associated with the signature, as described in [117]: the score associated with signature k in sample i is defined as , where is the estimated probability of differential expression of probe g in treatment k, is the estimated fold change of the transcript, is the estimated residual variance of probe g, and is the RMA normalized expression value of probe g in sample i. The signature score thus provides a relative measure of the extent to which the pattern of expression described by the signature is present or reversed in a given vector of expression values.…”