Profiling the suspect
Ovarian cancer is uncommonly deadly. Tumors aren’t usually detected until the disease has spread beyond the ovaries, at which point only a third of patients survive more than five years.
A promising blood test could change this bleak picture, enabling doctors to identify ovarian cancer at its earliest, most treatable stage.
“This has the potential of being an important advance,” says Dr. Philip DiSaia, director of gynecological oncology at the UC Irvine School of Medicine. “Ovarian cancer is extremely difficult to detect early on and is usually diagnosed at advanced stages, which is why the death rates are so high.”
More than 25,000 women are expected to be diagnosed with ovarian cancer this year; 16,000 will die of the disease. Early symptoms, which include indigestion, gas, bloating, a feeling of fullness and pain during intercourse, are often vague. Only 29% of all these cases are caught early, when almost all women can be successfully treated. For women with the most advanced cancers, survival rates plummet.
The only test currently available to detect ovarian cancer measures levels of a cancer protein called CA 125. But the test, which is best at spotting cancer recurrences, is unreliable: It misses half of all early-stage cancers and 20% of late-stage cancers.
A follow-up test, the vaginal ultrasound, is also problematic: It is costly and doesn’t distinguish between benign and malignant tumors.
In contrast, the new diagnostic test, called OvaCheck, analyzes hundreds of thousands of proteins -- not just one -- floating in a drop of blood. Using pattern recognition technology (a computer technique borrowed from artificial intelligence), the test can spot the telltale fingerprint of ovarian cancer proteins.
“We use pattern recognition all the time,” says chief scientist Ben A. Hitt, co-founder of Correlogic Systems Inc. in Bethesda, Md., which makes OvaCheck. “We recognize people not by a single feature on their face, but by the pattern they represent. In the same way, we found features that were characteristic of ovarian cancer, and developed software that would sort through thousands of proteins in the blood to determine if the deviant cells are present.”
In a 2002 study of 116 blood samples, OvaCheck correctly identified 100% of the patients with cancer and 95% of the control samples that were noncancerous (though it said 5% of the women had cancer when they did not).
Since then, the technology has been refined, and subsequent studies indicate OvaCheck picked up 100% of cancers with no false positives, says Peter J. Levine, Correlogic’s chief executive. “But no test is foolproof,” adds Levine, OvaCheck co-inventor with Hitt.
The company hopes to have the test available through licensed laboratories within the next year for the 10 million American women who are at an elevated risk for ovarian cancer. They include women who have a mother, sister or daughter with ovarian or breast cancer; who test positive for the mutated forms of the BRCA 1 or BRCA 2 genes; or who have elevated levels of the CA 125 protein.
Experts say larger studies need to be done before OvaCheck can be used as a screening tool for all women. But for women in high-risk groups who are considering the prophylactic removal of their ovaries, the test could offer peace of mind when used with other diagnostics, such as ultrasound and the CA 125 test. “At present, the test’s greatest value is for the high-risk group,” says DiSaia.
*
(BEGIN TEXT OF INFOBOX)
Tests for other cancers are in the works
Two other companies, LPL Technologies in Cleveland and Ciphergen Biosystems of Fremont, Calif., have devised ovarian cancer tests that use protein-profiling technology, though their tests aren’t as far along in development.
In addition, academic researchers and a few biotech companies, including Correlogic, are developing diagnostic tests based on protein patterns for lymphoma, leukemia and prostate, breast, pancreatic, kidney and lung cancers.
If the Correlogic test gets government clearance, it will be the first cancer test using protein profiling.
