Blood biomarker test detects ovarian cancer early

The development of a biomarker strategy to efficiently detect ovarian cancer may be on the horizon for women at high risk for the disease.

Researchers at Yale and other universities analyzed 156 patients recently diagnosed with ovarian cancer and 362 controls to test serum biomarkers for their ability to efficiently differentiate between disease-free and ovarian cancer patients. The researchers evaluated concentrations of leptin, prolactin, osteopontin, insulin-like growth factor II, macrophage inhibitory factor and CA-125 in both training (181 controls and 113 ovarian cancer) and test sets (181 controls, 43 ovarian cancer).

Together, the six biomarkers could differentiate between healthy and cancer cells better than CA-125 alone. Significant improvement over CA-125 (sensitivity 72% at specificity 95%) in the test set was reported in four models (sensitivity 84% to 98% at specificity 95%). These four models had <2% classification error in training sets, according to the researchers. None of the biomarkers alone were strong enough to differentiate between healthy and cancer cells.

The researchers concluded that 221 out of 224 specimens were correctly classified by the chosen model in the test set, with a classification accuracy of 98.7%. – by Stacey L. Adams

Clin Cancer Res. 2008;14:1065-1072.

This is an interesting study conducted by an excellent group of investigators that suggests it may be possible to develop a biomarker strategy for the detection of ovarian cancer. However, it is critically important to state that far greater evaluation of this approach is required before it should be considered a component of routine medical care. In the past, other tests have suggested diagnostic accuracy equivalent to that noted in this report but further research was unable to confirm the findings.

– Maurie Markman, MD

HemOnc Today Editorial Board member

The question in this study is whether you can take many markers that, by themselves, are not satisfactory for screening and put them together to come up with a statistical model which will be adequate for screening for early disease. This is an interesting idea, but the researchers’ evidence is preliminary. At this point in time, I do not think it is ready for prime time and certainly needs validation in a larger number of patients. It is not clear from the paper that you need six markers; four may do the same thing.

– Mark Morgan, MD

Chief, Section of Gynecologic Oncology
Department of Surgery, Fox Chase Cancer Center