BioCentury
ARTICLE | Distillery Techniques

Biomarkers; other

May 15, 2018 6:29 PM UTC

A gene profiling-based computational model could identify single-gene deletions as markers of response to therapies for high-grade serous ovarian cancer (HGSOC) patients. The model is generated by transfecting a human breast cell line with biology and etiology similar to HGSOC cells with a library of 612 siRNAs for targeting breast cancer, ovarian cancer or DNA repair genes; treating each knockdown variant with 31 therapies; and assigning scores based on cell proliferation to yield 19,406 predictions for gene-drug interactions, including resistance to PARP inhibitors in tumors lacking AT rich interactive domain 1A (ARID1A) and resistance to platinum-based chemotherapy in collagen type IV α 3 binding protein (COL4A3BP; GPBP)-negative tumors. Validation studies in HGSOC patients receiving platinum-based chemotherapy confirmed the model’s predictions that loss of ARID1A and COL4A3BP deletion correlated with poor response, as measured by overall survival (OS) in 315 patients and disease-free survival in 262 patients, respectively. Validation studies in another 154 relapsed, platinum-sensitive HGSOC patients receiving the PARP inhibitor Rubraca rucaparib confirmed the model’s prediction that ARID1A deletion correlated with poor progression-free survival (PFS). Next steps include clinical trials to validate more of the genetic mutations predicted by the model as markers of response in breast cancer patients...