Researchers at the University of Pittsburgh Medical Center have developed a new ovarian cancer screening test that they claim has a specificity of 98 percent and a sensitivity of about 93 percent, giving it a higher diagnostic power than any other published ovarian cancer assay to date.
UPMC researchers led by Anna Lokshin, an assistant professor of medicine and pathology, used a biomarker panel comprising 20 proteins to test 150 ovarian cancer patients and 150 controls.
The researchers first used 60 cases and 60 controls as a training set to develop the biomarker panel, then did a validation study of 40 cases and 40 controls. Following that, they tested another 50 cases and 50 controls that had never been seen before by the computer. The specificity and sensitivity numbers reported are from the second, blinded validation set.
Lokshin said that the new test gives the highest numbers, in terms of specificity and sensitivity, for any case-control ovarian cancer study to date. Results of the study were presented at the annual meeting of the American Association for Cancer Research, held this week in Washington, DC.
Lokshin and her research team are currently working with UPMC to patent the test. Lokshin said there are "two companies standing in line" to negotiate a deal to try to commercialize the test, but so far no deals have been made with them.
"I count [the Yale/LabCorp test] as our closest contender. But we are ahead of them in that we had larger [patient sample] numbers in our study, and all our cases were stage I and II ovarian cancer, while theirs included a lot of stage III and IV."
UPMC's test has strong rivals. In May 2005, researchers at Yale University and the Nevada Cancer Institute published a paper in the Proceedings of the National Academy of Sciences of the United States of America describing an ovarian cancer test based on a four-protein biomarker panel that had specificity and sensitivity of 95 percent (see ProteoMonitor 5/13/2005). Yale exclusively licensed the test to LabCorp in February.
"I count [the Yale/LabCorp test] as our closest contender," said Lokshin. "But we are ahead of them in that we had larger [patient sample] numbers in our study, and all our cases were stage I and II ovarian cancer, while theirs included a lot of stage III and IV."
Lokshin added that while the Yale test had a slightly higher sensitivity than the UPMC test at a specificity of 95 percent, if the specificity of the Yale test were raised to 98 percent, the sensitivity would probably go down considerably, to below 90 percent.
"You have to compare sensitivity at a set specificity," said Lokshin.
Gil Mor, an associate professor of obstetrics and gynecology who led the development of the Yale ovarian cancer test, said that he had not seen the presentation of the UPMC group at the AACR meeting, but that "98 percent is quite impressive."
However, Noah Kauff, the director of Ovarian Cancer Screening and Prevention at Memorial Sloan-Kettering Cancer Center, cautioned that even if the test had 100 percent sensitivity, a test with 98 percent specificity would mean that if 2,500 women in the general population underwent the test, one woman would have ovarian cancer detected, but 49 women would have unnecessary surgery, and with it the risk of associated complications.
"What has been the goal up to this point is to develop a screening test for ovarian cancer that leads to one cancer found in every 10 surgeries done as the result of an abnormal screen," said Kauff. "Because the incidence of ovarian cancer in post-menopausal women is only approximately one per 2,500 women per year, that would require a test to have a 99.6 percent specificity to achieve this result."
Ciphergen is another company developing an ovarian cancer test. The company, in collaboration with researchers at the Johns Hopkins School of Medicine, the Danish Cancer Society, and the University of Copenhagen's Rigshospitalet hospital, recently conducted a prospective study that evaluated 202 patients who had a persistent pelvic mass or pelvic pain.
Ciphergen's test, which is based on a panel of seven blood biomarkers, in addition to the commonly used CA-125 biomarker, aims to discriminate ovarian cancer from benign pelvic masses, including endometriosis and benign ovarian cysts. When used to test a blinded set of 76 women, the test was found to have a positive predictive value of 80 percent, according to Ciphergen.
Asked to compare the new UPMC test to Ciphergen's ovarian cancer test, Lokshin said that she could not "because it's apples and oranges."
While Ciphergen's prospective study appeared to target women who are at high risk for ovarian cancer, the new UPMC ovarian cancer test is targeted for the general female population, said Lokshin.
"If you're testing women with symptoms [of ovarian cancer], then [the cancer] is likely to be already in the late stage," said Lokshin. "Our test is a general screening test, not just for women who have high risk, or women with symptoms."
Ciphergen, which also presented at AACR this week, did not return calls for comment.
In addition to LabCorp and Ciphergen, Correlogic, in collaboration with LabCorp and Quest, is also developing an ovarian cancer test called OvaCheck which is based on a study led by Lance Liotta and Emanual Petricoin that was published in the February 2002 issue of Lancet. The study showed that a proteomic pattern was able to discriminate ovarian cancer from unaffected women (see ProteoMonitor 6/25/2004).
The panel of 20 protein biomarkers that are the basis of the new UPMC ovarian cancer test was selected out of an original panel of more than 80 proteins, Lokshin said. The original panel was selected based on published reports of proteins associated with ovarian cancer, and cancer in general.
"When we were putting together this [original] panel, I tried to involve proteins with different functions representing the areas of tumor growth, angiogenesis, immune response, and proteases that participate in metastasis," explained Lokshin. "Then we gave the computer all the data, and an algorithm decided which combination [of biomarkers] was the most powerful."
The researchers used a Luminex bead-based, multiplexed assay technology called LapMAP to analyze their proteins. The technology is much better suited than ELISA for the analysis of a large panel of proteins, said Lokshin.
"If you start doing ELISA on 80 proteins, you need 200 microliters of blood, and nobody will give you so much blood," she said. "The LapMAP takes only 50 microliters of blood for the whole thing."
UPMC researchers are still in the process of testing an additional 150 cases and controls, Lokshin said. All of the case and control samples tested so far were obtained from the Gyncologic Oncology Group • a multi-center, non-profit organization • for a nominal fee of $25 per sample, she added.
After the additional 150 samples have been tested, the researchers plan to validate their results with a retrospective study involving close to 1,000 samples, Lokshin said. In parallel to that study, the researchers plan to analyze 120 samples from the National Cancer Institute's Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial, a large-scale clinical trial in which patients had a physical exam and gave a blood sample every year for 10 years prior to the development of one of the four types of cancer.
"We will run our panel using those samples," said Lokshin. "We can see how the markers are changing at certain stages of cancer development."
In addition, the researchers plan to analyze 200 samples from the United Kingdom Collaborative Trial of Ovarian Cancer Screening, another large-scale clinical trial similar to the NCI's PLCO trial, except that it involved only ovarian cancer.
Lokshin said she hopes to receive the PLCO samples by September, and to finish the large retrospective study within a year.
"This test is very, very promising, but it is only in the first phase of the assay validation," said Lokshin.
Lokshin noted that in the UKTOCS study, 27 out of 12,000 women were eventually diagnosed with ovarian cancer. Of those 27, 24 were diagnosed in the late stages of ovarian cancer, despite the fact that they had been receiving yearly exams and blood tests for at least 10 years prior to their cancer diagnosis.
"It's not surprising that so many were in the late stages, because a transvaginal ultrasound sees a tumor, and if you can see a tumor, then, by definition, it is already big in size," said Lokshin. "And CA-125 levels are proportional to tumor size."
• Tien-Shun Lee ([email protected])