Prototype Proteomic Test Detects Early Cancer
With the new test, the research team correctly identified all 50 women who provided a serum sample and had ovarian cancer. The researchers also accurately identified 63 of the 66 samples obtained from women without the disease. A report of the study was assigned to fast-track publication in the Feb. 16 issue of The Lancet (PDF).
Perhaps as notable as the tests apparent accuracy is its design, which combines proteomics and artificial-intelligence technology. In essence, a computer program sifted through mass-spectroscopy data from serum samples and selected the protein patterns that differentiated between the specimens from women with ovarian cancer and those without. After a suitable protein pattern was identified, it was used to screen a masked panel of 116 serum specimens from other women, 50 of whom were known to have ovarian cancer. This part of the study, with the masked specimens, was undertaken to validate the protein pattern as a screening tool.
According to the research team, the "positive predictive value" of the testthe probability that a person with a positive test result actually has ovarian cancerwas 94 percent. Eighteen of the serum specimens from women known to have ovarian cancer had been obtained during the earliest stage of the disease, before it spreads to other organs. The positive predictive value of the cancer antigen 125 measurement, a widely used diagnostic test for ovarian cancer, was 35 percent for the same group of 116 masked specimens.
For women whose ovarian cancer is detected in the earliest stage and treated, the five-year survival rate is greater than 90 percent, according to NCI. But 80 percent of women with ovarian cancer do not know that they have the disease until it has spread. At this stage, the five-year survival rate is about 35 percent.
The NCI-led team cautioned that the proteomic test must be studied further and refined to reduce the false-positive rate.