Ovarian carcinoma as a surrogate tumor for lung adenocarcinomas in evaluating the chemo-stability of a gene expression signature.

Gene expression predictors of response (GEPR) are an exciting and developing area of personalized cancer therapy. Genome-wide expression data can predict drug response and may segregate inter-patient heterogeneity better than single biomarkers alone. Generation of GEPRs frequently employs cell culture systems but validation requires tumor samples from treated patients. We hypothesize that GEPR validation requires tissue acquisition immediately prior to treatment, representing a significant obstacle for validation of GEPRs of second- or third-line therapeutics. To test whether expression data obtained prior to first line therapy can predict response to subsequent lines of therapy, chemo-stability of the signature must first be assessed. Publicly-available microarray data were utilized to identify a surrogate tumor type to assess the in vivo chemo-stability of our published GEPR of erlotinib sensitivity for which pre- and post-therapy samples can be acquired. Expression patterns in multiple epithelial tumors were compared to lung adenocarcinomas (LAC). Correlative analyses identified ovarian carcinoma (OC) and breast carcinoma as most comparable to LAC. We selected OC for further study because of relative ease of tumor acquisition. Chemo-stability of our erlotinib-GEPR was assessed and found to be generally stable using published expression data from OC and LAC in vitro experimentation. These findings support the use of resected OC samples for in vivo analysis of erlotinib-GEPR chemo-stability, and this data may be translated to LAC to determine if tumors acquired prior to frontline therapy are a viable option for anticipating response to a second- or third-line therapeutic.