Posted: Thursday, May 19, 2022
Co-expression extrapolation (COXEN) scores specific to treatment of bladder cancer were marginally associated with response and significantly associated with downstaging, according to a study published in European Urology Focus. COXEN is an algorithm that works to identify clinically predictive biomarkers using information from cancer cell line chemosensitivity assays. Garrett M. Dancik, PhD, MS, of Eastern Connecticut State University, Willimantic, Connecticut, and Dan Theodorescu, MD, PhD, of the Samuel Oschin Comprehensive Cancer Institute, Los Angeles, coauthored the article.
“These results validate the ability to identify predictive biomarkers from cell lines without patient response data. The ability to develop such predictive biomarkers without ‘training’ on patient response data was a conceptual breakthrough,” noted the authors.
The COXEN algorithm was validated using a randomized phase II trial that enrolled 237 patients with urothelial carcinoma of the bladder. Patients were randomly assigned to receive either neoadjuvant gemcitabine and cisplatin or dose-dense methotrexate, vinblastine, doxorubicin, and cisplatin. The algorithm used gene-expression data from the NCI-60 panel, chemosensitivity data for individual compounds used in the trial, and gene expression from a bladder cancer cohort to find corresponding genes. Treatment-specific COXEN models were created to predict whether that particular treatment would be favorable for an individual patient.
A total of 167 patient samples were evaluable for COXEN analysis, 82 in the gemcitabine/cisplatin arm and 85 in the dose-dense arm of methotrexate, vinblastine, doxorubicin, and cisplatin. Treatment-specific COXEN scores did not predict the response to treatment; however, COXEN gemcitabine/cisplatin scores were marginally associated with response in the gemcitabine/cisplatin arm (P = .10). Additionally, the COXEN gemcitabine/cisplatin score was associated with downstaging when all the patients were pooled together (odds ratio = 2.33, P = .02). Complete response rates were similar in both treatment arms.
Disclosure: The study authors reported no conflicts of interest.