Medical Device Daily Washington Editor

GAITHERSBURG, Maryland – The 510(k) program is under close scrutiny by forces outside FDA, but reviewers at the Center for Devices and Radiological Health are also tightening down as evidenced by the use of an advisory committee to review an expanded indication for a premarket clearance device for the second time in less than a month (Medical Device Daily, Nov. 24, 2008).

The immunological devices advisory committee met Wednesday to review an attempt to expand the indications for the HE4 EIA assay, made by Fujirebio Diagnostics (Malvern, Pennsylvania) to aid in a decision whether or not to operate on ovarian neoplasms or any adnexal mass, which is any neoplasm located in the uterus, fallopian tubes or the ovaries. The diagnostic is currently cleared in the U.S. as an aid in monitoring recurrence or progressive disease in women with epithelial ovarian cancer (EOC).

One of the complicating factors that surfaced during the hearing is that the requested indication would pertain only to the use of the diagnostic by a gynecological oncologist, who would presumably refer the patient back to the referring physician should the diagnostic indicate a non-cancerous mass. This feature of the application generated some headwind for Fujirebio, which came away empty-handed at day's end.

During the open public hearing, April Donahue, a survivor of ovarian cancer, told the panel "I was one of the people who was fortunate to be diagnosed in stage I" ovarian cancer at age 24. She said her symptoms "were dismissed by four different doctors," but the removal of the ovarian cyst was expected to show the cyst was non-malignant. That proved not to be the case.

"Thank God it was confined to that one ovary," Donahue stated, but her symptoms resurfaced a decade later. A surgeon removed her remaining ovary, which also proved to be cancerous. She characterized the difficulty in obtaining a clear diagnosis of malignancy as a "frustrating struggle that many women experience when they're diagnosed." She also referred to the epidemiological feature of ovarian cancer, noting that "the overall five year survival rate is only 53%," but while survival is 90% at early diagnosis, only one in five are diagnosed in the first year.

Richard Moore, MD, the principal investigator in the study mounted by Fujirebio in support of the expanded indication and a physician at Women and Infants Hospital of Rhode Island (Providence, Rhode Island), said ovarian cancer "remains one of the deadliest of all cancers," resulting in more than 15,000 deaths each year. Moore also noted, "Unfortunately, ovarian cancer rates are either stable ... or increasing slightly."

Screening and early detection are not readily available, Moore pointed out, adding that in his opinion, "this is a tool that should be available to all our patients."

Steven Skates, MD, assistant professor of biostatistics at Harvard Medical School (Boston), said "there is an unmet need in ovarian cancer that statistical modeling can address," adding that CA 125, a biomarker for cancer that is picked up in blood tests, "is suboptimal as a marker" for ovarian cancer.

Skates said that "we believed we needed a combination of markers, not just one" such as CA 125, a cancer biomarker that is picked up from blood samples. "The goal was to find the smallest panel with the highest sensitivity," he said noting that in the pivotal trial (the company conducted two pilot studies to validate the algorithm), the assay checked for both the CA 125 marker and the HE4 marker, developed for the ROMA (risk of ovarian malignancy algorithm).

Moore said that the validation of ROMA was a prospective, double-blinded multi-center study employing "a central pathology review to confirm the site pathology diagnosis." Of the 566 subjects enrolled, 478 completed the study. Two-thirds had benign disease, and about a quarter (24%) had invasive EOC.

Moore said ROMA generated 89 false-positives, but thanks to a sensitivity of 89% at 75% specificity, "we were able to identify 89% of pre-menopausal patients with an EOC." Moore also noted that for early-stage diagnoses, ROMA "also correctly identifies 86% of the stage I and II cancers and 99%" of stage III and IV cancers. "ROMA will provide a risk assessment tool that is easy to interpret," and "will be a valuable addition to the tools to those we currently use to assess the risk of cancer," he said.

According to other company testimony, the effective false negative rate at present is higher than 50%, and ROMA offers an aggregate false negative rate of 6%.

Marina Kondratovich, PhD, a statistician at the CDRH, said "subjects with low risk assessed by a referring physician and having surgery in a community setting were not present in the sponsor's study and therefore are not part of the population cited in the intended use." She also said "I would like to emphasize that the sensitivity of the ROMA test was different for the pre-menopausal group from the post-menopausal group."

Kondratovich presented a statistical table that indicated that the sensitivity of the assay was only 76.5% for pre-menopausal women, whereas the post-menopausal subjects enjoyed a sensitivity rate of 92.4%. Sensitivity in this context is the probability that a positive diagnosis of cancer is correct, whereas specificity is the correct negative finding of cancer.

Skates' response was that FDA's analysis "is not valid" because "when you go into subgroup analyses, it is common to find false positive and false negative results," he said.

Questions about the appropriateness of the test in pre-menopausal women persisted, however, and some panelists expressed reservations about the idea of sending a patient with a negative finding back to the referring gynecological surgeon. However, one panelist pointed out that the CA 125 test is not approved or cleared for detection of ovarian cancer.

During the afternoon open public hearing, David Fishman, MD, director of gynecological oncology at New York University Medical Center (New York), told the panel, "any tool that can improve diagnosis ... saves lives." He also said that the consequences of any ovarian mass that ruptures is that the patient "just went from not needing chemotherapy to needing chemotherapy for six months.

"We need more tools," which "will save lives," said Fishman. "As a clinician who takes care of patients, any tool that helps us take care of patients" is important.

Paul Touhey, Fujirebio's president/CEO, told Medical Device Daily, "We're going to talk to FDA" about labeling language and remarked that "this is an area that needs a [new] product" given that the CA 125 assay was developed in 1981.

He declined to discuss the specifics of the product label language that will be reviewed with the agency, but said the differences between the company and FDA can resolve strictly by revising the language of the label.