according to a in Annals of Oncology.
“Gene expression signature tests for prognosis are available for other cancers, such as breast cancer, and these help with treatment decisions, but no such tests are available for ovarian cancer,” senior investigator, of Lowy Cancer Research Centre, University of NSW Sydney, commented in an interview.
Dr. Ramus and associates developed and validated their 101-gene expression signature using pretreatment tumor tissue from 3,769 women with high-grade serous ovarian cancer treated on 21 studies.
The investigators found this signature, called OTTA-SPOT (Ovarian Tumor Tissue Analysis Consortium–Stratified Prognosis of Ovarian Tumors), performed well at stratifying women according to overall survival. Median overall survival times ranged from about 2 years for patients in the top quintile of scores to more than 9 years for patients in the bottom quintile.
Moreover, OTTA-SPOT significantly improved prognostication when added to age and stage.
“This tumor test works on formalin-fixed, paraffin-embedded tumors, as collected routinely in clinical practice,” Dr. Ramus noted. “Women predicted to have poor survival using current treatments could be included in clinical trials to rapidly get alternative treatment. Many of the genes included in this test are targets of known drugs, so this information could lead to alternative targeted treatments.
“This test is not ready for routine clinical care yet,” she added. “The next step would be to include this signature as part of a clinical trial. If patients predicted to have poor survival are given alternative treatments that improve their survival, then the test could be included in treatment decisions.”
Dr. Ramus and colleagues began this work by measuring tumor expression of 513 genes selected via meta-analysis. The team then developed a gene expression assay and a prognostic signature for overall survival, which they trained on tumors from 2,702 women in 15 studies and validated on an independent set of tumors from 1,067 women in 6 studies.
In analyses adjusted for covariates, expression levels of 276 genes were associated with overall survival. The signature with the best prognostic performance contained 101 genes that were enriched in pathways having treatment implications, such as pathways involved in immune response, mitosis, and homologous recombination repair.
Adding the signature to age and stage alone improved prediction of 2- and 5-year overall survival. The area under the curve increased from 0.61 to 0.69 for 2-year overall survival and from 0.62 to 0.75 for 5-year overall survival (with nonoverlapping 95% confidence intervals for 5-year survival).
Each standard deviation increase in the gene expression score was associated with a more than doubling of the risk of death (hazard ratio, 2.35; P < .001).
The median overall survival by gene expression score quintile was 9.5 years for patients in the first quintile, 5.4 years for patients in the second, 3.8 years for patients in the third, 3.2 years for patients in the fourth, and 2.3 years for patients in the fifth.
This study was funded by the National Institutes of Health/National Cancer Institute, the Canadian Institutes for Health Research, and the Department of Defense Ovarian Cancer Research Program. Some of the authors disclosed financial relationships with a range of companies. Dr. Ramus disclosed no conflicts of interest.
SOURCE: Millstein J et al.