Selection of personalized patient therapy through the use of knowledge-based computational models that identify...

Area of Science:

• Oncology
• Computational Biology
• Genomics

Background:

• Targeted cancer drugs inhibit specific signaling pathways driving tumor growth.
• Current companion diagnostic tests often fail to predict therapy response due to inadequate assessment of pathway activity.
• Estrogen receptor (ER) protein expression is an example of a test that does not reliably predict response in breast cancer treatment.

Purpose of the Study:

• To develop a novel approach for predicting cancer signaling pathway activity.
• To overcome limitations of current diagnostic tests by assessing functional pathway activity.
• To utilize quantitative transcriptome data for pathway activity assessment.

Main Methods:

• Development of knowledge-based Bayesian computational models.
• Interpretation of quantitative transcriptome data, specifically expression levels of transcriptional target genes.
• Calibration of models using limited cell line or patient data.

Main Results:

• The models reliably assess signaling pathway activity in diverse tumor types.
• Proof-of-principle models for the Wnt and Estrogen Receptor (ER) pathways were developed.
• Initial clinical validation on independent cancer datasets demonstrated model efficacy.

Conclusions:

• Bayesian computational models offer a reliable method for assessing functional signaling pathway activity.
• This novel approach enhances the potential for accurate prediction of targeted therapy response.
• The validated models show promise for application across various cancer types.