Abstract Using drug response data, generated by the OncoPanelTM cell-based profiling service, we investigated the differences between traditional and normalized growth rate inhibition (GR) metrics in drug sensitivity classification and predictive genomic biomarker discovery. Traditional cellular response parameters can be confounded by numerous growth conditions, especially when the mechanism of action of a drug is coupled to cell cycle processes. Metrics such as IC50 or Emax can be influenced by cellular growth rate, adding bias to the apparent sensitivity to or efficacy of the drug. Use of GR metrics aims to eliminate this bias by normalizing the response data to a per-cell division rate. In this study, we analyzed various parameters for a selection of standard of care compounds when tested against a large panel of cancer cell lines. Compounds were tested against between 213 and 297 different human cancer cell lines. Cells were seeded in 384-well plates and allowed to incubate for 24 hours before acoustic energy-based compound addition was made over 10 half-log concentrations, in triplicate. Initial cell density was captured in a parallel plate to enable cell doubling and end-point GR calculations. After 3 or 10 days of incubation, the cells were fixed, stained with DAPI, and imaged using a high-content imaging platform. Cell proliferation dose response curves were fitted using a 4-parameter log-logistic model within a custom curve-fitting software platform. These same data were then re-analyzed using the GR metrics package developed for the statistical programming language R, to determine GR metrics (GR50 and GRmax). Exploratory analysis was conducted on four parameters: IC50, GR50, GRmax, and number of doublings. IC50 and GR50 were used to classify compounds into drug-resistant or sensitive groups, and GRmax was used to classify each as cytostatic, partially cytostatic, or cytotoxic. Each parameter was used in a predictive genomic biomarker search, using baseline mRNA expression data and mutation data from the COSMIC/Sanger and Broad/CCLE databases. Correlation analysis for doublings and IC50, or GR50, suggests that cell lines with faster doubling rates are associated with drug sensitivity, especially with compounds that target cell cycle processes, such as doxorubicin, carboplatin, cisplatin, and oxaliplatin. Genomic analyses revealed different predictive biomarkers for each choice of parameter. The relationships between significant biomarkers were explored across compounds and cancer cell types. In summary, the methods employed in the OncopanelTM cell-based profiling service enable various dose response parameter estimations, including GR metrics. Through careful use of growth-normalized data, we have expanded our capacity to discover predictive genomic biomarkers that are less influenced by cellular growth conditions. Citation Format: Charles R. Wageman, Jesse J. Parry, Lee R. Cavedine, Alyssa M. Croff, Steven M. Garner, Brogan A. Epkins, Natiya E. Robinson, Kristin C. Dempsey, Vanessa L. Norman, Usha Warrior, Alastair J. King. Exploring drug response metrics for biomarker discovery using the OncoPanel™ cell-based profiling service [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 2260.
Read full abstract