Abstract Next-generation sequencing and gene expression signature analysis of primary tumors from The Cancer Genome Atlas (TCGA) revealed striking similarities between triple-negative breast cancer (TNBC) and ovarian cancer (OVCA). We hypothesized that these similarities may reveal transcriptionally-identifiable subgroups within a mix of TNBC and OVCA that are uniquely sensitive to certain drugs. To test this hypothesis, gene expression profiles for TNBC and OVCA cell lines in the Cancer Cell Line Encyclopedia were analyzed by unsupervised hierarchical clustering, which revealed two major unique subgroups containing both TNBC and OVCA cell lines that clustered according to the previously defined Mesenchymal and Basal subclasses of TNBC. Differential gene expression between “Mesenchymal-like” and “Basal-like” TNBC/OVCA cell lines was used to generate a gene signature that was subsequently validated using gene expression data in the Genomics of Drug Sensitivity in Cancer (GDSC) database. Drug sensitivity data from GDSC was then utilized to profile differential sensitivity of “Mesenchymal-like” and “Basal-like” cell lines to the 99 anti-cancer drugs with coverage of >50 breast and ovarian cell lines. Mesenchymal-like TNBC/OVCA cells were uniquely sensitive to HSP90 inhibition compared to Basal-like TNBC/OVCA cells for both HSP90 inhibitors in GDSC: CCT018159 (p<0.001) and 17-AAG (p=0.012). Strikingly, Mesenchymal-like TNBC/OVCA cells were most sensitive to HSP90 inhibition among all 33 solid tumor cancer lineages in GDSC, as well as other subgroups of breast and ovarian cancers. Differential sensitivity of Mesenchymal-like and Basal-like TNBC/OVCA cells to HSP90 inhibition with 4 agents (CCT018159, 17-AAG, AT13387, PU-H71) was validated using growth assays for 12 cell lines (6 of each subgroup) previously characterized in GDSC. To further validate the predictive value of our gene signature, gene expression data for 6 TNBC/OVCA cell lines with uncharacterized drug sensitivity was used to classify cell lines as “Mesenchymal-like” (n=4) or “Basal-like” (n=2). Our gene signature successfully predicted differential sensitivity to HSP90 inhibition, with the 4 Mesenchymal-like cell lines displaying greater sensitivity to HSP90 inhibition. The gene signature has been used to select PDX models with predicted differential sensitivity to HSP90 inhibition, and treatment studies are ongoing. In summary, we developed a novel approach to generate a gene expression signature that robustly predicts sensitivity to HSP90 inhibitors in preclinical models of TNBC and OVCA. HSP90 thus represents a therapeutic opportunity in an overlapping subpopulation of TNBC and OVCA. Our approach to identify novel combinations of drugs and histology/lineage-independent cancer subgroups may be used to discover new therapeutic opportunities in other cancer types. Citation Format: Shee K, Ung MH, Cheng C, Miller TW. Unique overlapping subtypes of triple-negative breast and ovarian cancers and sensitivity of “mesenchymal-like” cancers to HSP90 inhibition is revealed by integrated gene expression and drug sensitivity profiling [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-07-04.
Read full abstract