Molecular Signatures May Help Guide Ovarian Cancer Debulking
NEW YORK (Reuters Health) - Although genetic signatures can't yet predict survival in ovarian cancer, they can help determine the value of tumor debulking, according to two studies published online April 3 and April 4 in the Journal of the National Cancer Institute.
"The first study," Dr. Michael J. Birrer told Reuters Health by email, "examines all of the patient survival signatures and points out that many published gene expression signatures perform poorly and none are clinically useful."
Dr. Birrer, of Massachusetts General Hospital in Boston, and colleagues conducted a systematic review that identified 14 published prognostic models for late-stage ovarian cancer. They had sought to identify successful prognostic gene signatures through systematic validation using public data covering 1,251 patients.
But as they point out, "Most models demonstrated lower accuracy in new datasets than in validation sets presented in their publication."
For the second study, continued Dr. Birrer, "We then generated our own which prognosticates patient survival better than other signatures but is still not clinically useful because it does not alter clinical management."
The survival signature stratified 1,525 patients into high- and low-risk groups (hazard ratio, 2.19) and although superior to The Cancer Genome Atlas (TCGA) consortium signature, previously the best available prognostic model, it did not reach the level required for clinical utility.
However, of seven genes tested, six (POSTN, CXCL14, FAP, NUAK1, PTCH1 and TGFBR2) were statistically significantly associated with surgery outcome and were validated by qRT-PCR.
In addition, POSTN, CXCL14, and phosphorylated Smad2/3 were validated by immunohistochemistry as independent predictors of debulking status.
The sum of immunohistochemistry intensities for these three proteins, say the investigators, "provided a tool that classified 92.8% of samples correctly in high- and low-risk groups for suboptimal debulking."
"This signature is quite robust, biologically based and was validated on two independent sets of tumors," Dr. Birrer said.
He added, "We think this is clinically useful to provide both patient and doctor additional information as to whether to undergo upfront debulking surgery or chemotherapy first with interval debulking after three cycles."
Although a clinically meaningful survival signature was not achieved, "the important point, said Dr. Birrer, "is that larger numbers of gene expression arrays provides stronger signatures and one can anticipate that with more samples clinically useful survival signatures can be generated."
Dr. Brooke L. Fridley of the University of Kansas Medical Center, Kansas City, who co-authored an editorial, told Reuters Health by email that the papers "are excellent examples of the level of transparency that all studies should strive for to ensure reproducibility of research findings and conclusions. These two manuscripts also present an excellent framework for the development and assessment of gene expression signatures to predict clinical outcomes for ovarian cancer patients."
However, Dr. Fridley and colleagues note in their editorial that "big data" are important but "because each woman and her cancer are unique, successful cures and outcomes will only come from informative biomarkers/signatures and treatments that target specific cells within each person's tumor."
"As exemplified by these two articles," they conclude, "we have come a long way toward this goal; however, more research is needed to achieve the ultimate goal of precision medicine for cancer patients."
SOURCES: http://bit.ly/1rxyPrb, http://bit.ly/1mRFvCP, and
J Natl Cancer Inst 2014.