Related Experiment Video
Updated: Apr 5, 2026
Simultaneous Measurement of HDAC1 and HDAC6 Activity in HeLa Cells Using UHPLC-MS
Published on: August 10, 2017
Predicting Response to Histone Deacetylase Inhibitors Using High-Throughput Genomics
Paul Geeleher1, Andrey Loboda1, Divya Lenkala1
1Department of Medicine (PG, DL, FW, BL, SK, JW, MLM, RSH), Committee on Clinical Pharmacology and Pharmacogenomics (MLM, RSH), and the Comprehensive Cancer Center (MLM, RSH), University of Chicago, Chicago, IL; Oncology Clinical Research, Merck Research Laboratories, North Wales, PA (AL, MN, MC, JH).
Predicting cancer drug response requires evaluating multiple gene expression factors, not single biomarkers. This machine learning approach accurately predicts vorinostat efficacy and toxicity, shifting cancer treatment paradigms.
Area of Science:
- Oncology
- Genomics
- Pharmacology
Background:
- Previous biomarkers for histone deacetylase inhibitors (HDI) failed clinical application.
- Vorinostat response may depend on the combined effects of multiple molecular factors.
- Single biomarkers have not accurately predicted patient response to HDI therapies.
Purpose of the Study:
- To develop a predictive model for vorinostat response using gene expression data.
- To validate a polygenic marker prediction approach in cancer cell lines and clinical trials.
- To shift from single-gene evaluation to a multi-factor approach for predicting drug response.
Main Methods:
- Large-scale gene expression analysis for discovery and validation across cancer cell lines.
- Machine learning to identify aggregated small effects from gene expression data.
- Application of polygenic marker prediction principles in a clinical trial setting.
Main Results:
- Machine learning accurately predicted vorinostat response from gene expression data, outperforming single markers.
- Identified roles for chromatin remodeling, autophagy, and apoptosis in drug response.
- Discovered a novel role for CHD4 in poor clinical outcome and predicted vorinostat-induced thrombocytopenia.
Conclusions:
- A paradigm shift towards evaluating multiple gene expression datasets simultaneously is proposed.
- This multi-factor approach can improve prediction of investigational compound efficacy.
- The method is extendable to other compounds facing clinical adoption challenges.
