Ovarian Cancer Blood Test Accuracy Rates Improved

SAN DIEGO — Refinements to an experimental blood test for ovarian cancer have made it 97.5% sensitive and 99.7% specific for this rare, hard-to-detect disease, Dr. Aliza L. Leiser reported at the annual meeting of the Society of Gynecologic Oncologists.

“This test can discriminate between disease-free women and early cancer patients,” Dr. Leiser said in a presentation of validation studies conducted with serum samples from 160 newly diagnosed ovarian cancer patients and 365 healthy controls.

The investigators have started a prospective longitudinal trial to evaluate the screening tool in high-risk patients. Dr. Leiser, of Yale University, New Haven, Conn., said the group hopes to enroll 250–300 women. Participants are to submit blood samples every 3 months and undergo frequent transvaginal ultrasound and CA-125 screening.

Two years ago, Dr. Gil Mor of Yale's department of obstetrics, gynecology, and reproductive medicine announced that he and his colleagues had achieved 95% specificity, 95% sensitivity, 95% positive predictive value, and 94% negative predictive value with a four-protein blood test (Proc. Natl. Acad. Sci. U.S.A. 2005 May [Epub doi:10.1073/pnas.0502178102]). The test was considered highly promising, but the investigators said it had to be made at least 99.6% specific to avoid a high false-positive rate in the general population.

The revised test adds two more proteins—macrophage inhibitory factor and CA-125—to the original four-protein set of leptin, prolactin, osteopontin, and insulinlike growth factor II. Using bead-based multiplex technology, it determines levels of all six biomarkers simultaneously from a single blood sample. The new test also had a positive predictive value of 99.3% and a negative predictive value of 99.2% in the validation studies. It misclassified four stage I-II cancers; all stage III-IV cancers were identified correctly.

Applying these results to a 1 in 2,500 incidence of ovarian cancer in the general population, Dr. Leiser said its positive predictive value would be 12.6% and its negative predictive value 99.2%. She said the test is 91% sensitive for stage I-II disease and 100% sensitive for stage III-IV disease.

In a discussion of the study, Dr. Laura J. Havrilesky, of Duke University, Durham, N.C., noted, however, that only 24% of ovarian cancers in the validation studies were stage I-II cancers. “The test is useful only if it results in reduction of disease-related mortality, and this would be best achieved if it results in detection of the early-stage cancers.”

Dr. Leiser said that finding a sufficient number of early-stage patients is difficult because the disease is rarely detected when it is most treatable. She said her group is soliciting samples from other centers to expand the validation studies to a larger pool of patients with early-stage disease. The group also is trying the test in 1,200 healthy controls and 500 patients with other cancers and inflammatory conditions that might affect expression of the same biomarkers. The results so far have been negative in all patients with endometrial cancer, she said.

For information on submitting blood samples for testing, visit www.yaleobgyn.org/oncology/ovarian_cancer.html

SAN DIEGO — Refinements to an experimental blood test for ovarian cancer have made it 97.5% sensitive and 99.7% specific for this rare, hard-to-detect disease, Dr. Aliza L. Leiser reported at the annual meeting of the Society of Gynecologic Oncologists.

“This test can discriminate between disease-free women and early cancer patients,” Dr. Leiser said in a presentation of validation studies conducted with serum samples from 160 newly diagnosed ovarian cancer patients and 365 healthy controls.

The investigators have started a prospective longitudinal trial to evaluate the screening tool in high-risk patients. Dr. Leiser, of Yale University, New Haven, Conn., said the group hopes to enroll 250–300 women. Participants are to submit blood samples every 3 months and undergo frequent transvaginal ultrasound and CA-125 screening.

Two years ago, Dr. Gil Mor of Yale's department of obstetrics, gynecology, and reproductive medicine announced that he and his colleagues had achieved 95% specificity, 95% sensitivity, 95% positive predictive value, and 94% negative predictive value with a four-protein blood test (Proc. Natl. Acad. Sci. U.S.A. 2005 May [Epub doi:10.1073/pnas.0502178102]). The test was considered highly promising, but the investigators said it had to be made at least 99.6% specific to avoid a high false-positive rate in the general population.

The revised test adds two more proteins—macrophage inhibitory factor and CA-125—to the original four-protein set of leptin, prolactin, osteopontin, and insulinlike growth factor II. Using bead-based multiplex technology, it determines levels of all six biomarkers simultaneously from a single blood sample. The new test also had a positive predictive value of 99.3% and a negative predictive value of 99.2% in the validation studies. It misclassified four stage I-II cancers; all stage III-IV cancers were identified correctly.

Applying these results to a 1 in 2,500 incidence of ovarian cancer in the general population, Dr. Leiser said its positive predictive value would be 12.6% and its negative predictive value 99.2%. She said the test is 91% sensitive for stage I-II disease and 100% sensitive for stage III-IV disease.

In a discussion of the study, Dr. Laura J. Havrilesky, of Duke University, Durham, N.C., noted, however, that only 24% of ovarian cancers in the validation studies were stage I-II cancers. “The test is useful only if it results in reduction of disease-related mortality, and this would be best achieved if it results in detection of the early-stage cancers.”

Dr. Leiser said that finding a sufficient number of early-stage patients is difficult because the disease is rarely detected when it is most treatable. She said her group is soliciting samples from other centers to expand the validation studies to a larger pool of patients with early-stage disease. The group also is trying the test in 1,200 healthy controls and 500 patients with other cancers and inflammatory conditions that might affect expression of the same biomarkers. The results so far have been negative in all patients with endometrial cancer, she said.

For information on submitting blood samples for testing, visit www.yaleobgyn.org/oncology/ovarian_cancer.html

SAN DIEGO — Refinements to an experimental blood test for ovarian cancer have made it 97.5% sensitive and 99.7% specific for this rare, hard-to-detect disease, Dr. Aliza L. Leiser reported at the annual meeting of the Society of Gynecologic Oncologists.

“This test can discriminate between disease-free women and early cancer patients,” Dr. Leiser said in a presentation of validation studies conducted with serum samples from 160 newly diagnosed ovarian cancer patients and 365 healthy controls.

The investigators have started a prospective longitudinal trial to evaluate the screening tool in high-risk patients. Dr. Leiser, of Yale University, New Haven, Conn., said the group hopes to enroll 250–300 women. Participants are to submit blood samples every 3 months and undergo frequent transvaginal ultrasound and CA-125 screening.

Two years ago, Dr. Gil Mor of Yale's department of obstetrics, gynecology, and reproductive medicine announced that he and his colleagues had achieved 95% specificity, 95% sensitivity, 95% positive predictive value, and 94% negative predictive value with a four-protein blood test (Proc. Natl. Acad. Sci. U.S.A. 2005 May [Epub doi:10.1073/pnas.0502178102]). The test was considered highly promising, but the investigators said it had to be made at least 99.6% specific to avoid a high false-positive rate in the general population.

The revised test adds two more proteins—macrophage inhibitory factor and CA-125—to the original four-protein set of leptin, prolactin, osteopontin, and insulinlike growth factor II. Using bead-based multiplex technology, it determines levels of all six biomarkers simultaneously from a single blood sample. The new test also had a positive predictive value of 99.3% and a negative predictive value of 99.2% in the validation studies. It misclassified four stage I-II cancers; all stage III-IV cancers were identified correctly.

Applying these results to a 1 in 2,500 incidence of ovarian cancer in the general population, Dr. Leiser said its positive predictive value would be 12.6% and its negative predictive value 99.2%. She said the test is 91% sensitive for stage I-II disease and 100% sensitive for stage III-IV disease.

In a discussion of the study, Dr. Laura J. Havrilesky, of Duke University, Durham, N.C., noted, however, that only 24% of ovarian cancers in the validation studies were stage I-II cancers. “The test is useful only if it results in reduction of disease-related mortality, and this would be best achieved if it results in detection of the early-stage cancers.”

Dr. Leiser said that finding a sufficient number of early-stage patients is difficult because the disease is rarely detected when it is most treatable. She said her group is soliciting samples from other centers to expand the validation studies to a larger pool of patients with early-stage disease. The group also is trying the test in 1,200 healthy controls and 500 patients with other cancers and inflammatory conditions that might affect expression of the same biomarkers. The results so far have been negative in all patients with endometrial cancer, she said.

For information on submitting blood samples for testing, visit www.yaleobgyn.org/oncology/ovarian_cancer.html