The race is on to release the first proteomics-based ovarian cancer diagnostic: This week, Ciphergen announced that it hopes to translate a collaborative ovarian cancer study into a commercial assay “by year-end.” Meanwhile, at last week’s IBC Biomarker conference, Lance Liotta of the NCI-FDA clinical proteomics program announced that his group — which he co-directs with Emanuel Petricoin — is “now gathering patients” for a blinded clinical trial that compares the predictability of protein patterns to CA125 for ovarian cancer recurrence. The trial is intended to set the stage for FDA approval of the test as a 510(k) (see PM 2-13-04).
Petricoin previously told ProteoMonitor that he expected the trial would be completed in an 18-month time frame (see PM 7-18-03). Petricoin and Liotta have not yet published identifications for the markers in their patterns, although they say they have made several identifications and will publish these once the identifications have been thoroughly validated.
Quest Diagnostics and LabCorp, which have licensed the technology for the pattern test from the NCI-FDA’s partner, Correlogic Systems, have said that they intend to release homebrew versions of a test early this year.
Ciphergen’s effort, which comes from a collaboration with Johns Hopkins University and several other partners, would use the company’s ProteinChip and SELDI platforms to screen for early-stage cancer in women based on a panel of identified analytes. The test would first launch as an analyte specific reagent in partnership with a “large clinical reference laboratory,” Gail Page, president of Ciphergen’s new diagnostics division, told ProteoMonitor. ASRs, which are homebrew tests, do not require FDA approval. Page emphasized, however, that the company will simultaneously seek FDA approval for the test. “We have had two meetings with the FDA. They have been glad to work with us, and have provided guidance,” she said. “We see a clear path to FDA approval.” She said approval as a 510(k) and a PMA were both possibilities.
Page said the company had not yet decided on a target market for the initial ASR test, but she said that ASRs were the way to go with initial diagnostic test releases. “It gives you an early introduction to the market … and you can look for further validation and clinical utility,” Page said.
The first pre-development study that Ciphergen conducted looked at 500 serum samples — some from healthy controls, some from early-stage cancer patients, and some from late-stage cancer patients. Searching for protein panels that could identify early-stage ovarian cancer among the women, Ciphergen and its partners found several markers and then identified three — using first the SELDI machine and then verifying the identities with the ProteinChip/ABI Q-STAR combination — that appeared to relate to the biological mechanism of ovarian cancer. “If you don’t have that [biological connection], you don’t have the complete biological picture,” Page said.
The company has already patented the three markers, and controls the IP for the use of any or all of them in relation to the ovarian cancer disease state. This is part of Ciphergen’s general strategy of snatching up as much IP as early as possible, Page said. “One of the keys to being successful in diagnostics is making sure you have IP,” she said.
As for software, Ciphergen also recently patented its own in-house pattern recognition algorithm called CART, for classification and regression tree analysis. Page described the software as “a cross-representation of what’s out there.”
Since the conclusion of the initial study at the end of 2003, the process has recently entered a validation phase in which Ciphergen and its collaborators are profiling sera of 1,000 patients from seven different sites in retrospective and prospective analyses of the test’s ability to detect early-stage ovarian cancer. “We follow the path of the woman” from healthy to sick, she said. She said comparisons are made with healthy controls.
Several biomarker researchers at last week’s meeting — including Liotta, as well as one of Ciphergen’s partners at Johns Hopkins, Daniel Chan — said that validation of any test against sick controls would be a necessary validation step to rule out cross-reactivity and establish the test’s specificity as diagnostic tool for ovarian cancer.
Page would not comment on the specificity or sensitivity of Ciphergen’s test, except to say that it is “significantly better” than other currently available tumor markers. She said numbers would be revealed when results from this year’s and last year’s studies are published later this year. Petricoin and Liotta claimed 100 percent accuracy in some of their Q-STAR runs, and have said that they consistently get specificity and sensitivity percentages in the upper 90s. Page expressed skepticism about the likelihood of any 100 percent claim holding up. “When people say 100 percent sensitivity, that raises a red flag for me,” Page said.
Beyond Mass Spec Tests
Ciphergen and the NCI-FDA are not the only groups in the race to put out a proteomics-based ovarian cancer test: Michael Tainsky, a professor at Wayne State University, is also trying to develop a test for ovarian cancer with identified, patentable markers that he hopes will make it to the market at least as a research-grade test “in under a year.” (see pioneer, p. 8). Tainsky, who said he has achieved 93 percent sensitivity and specificity for ovarian cancer using a panel of 480 antigen markers that test for antibody reactions in patients’ blood, said that for about 60 percent of his markers, “it would be impossible for there to be [existing] IP on them.” He plans to partner with a company for commercialization of a simple ELISA test based on the most promising markers — a diagnostic form that he believes would be easier to use and more reliable than a mass spec test would be. “We really devised this so that it can be put into any hospital — any lab that can do an immunoassay — tomorrow,” he said.
Although both Ciphergen and the NCI-FDA are currently testing mass-spec based diagnostics, most speakers at last week’s Biomarker conference favored identification followed by an immunoassay as the ultimate goal, for similar reasons that Tainsky gave. “Immunoassays are clearly the direction things are headed,” Richard Hayes, a senior investigator at the NCI, stated.
But this view was not universal. “Tandem mass spec is robust, stable over time, and can have internal standards,” George Klee, professor at the laboratory of medicine and pathology at the Mayo Clinic, said in his conference presentation. “We’ve heard a tendency ... to move from mass spectrometry to immunoassay, but we’re actually trying to do the opposite.”
Whatever the first test looks like, the commercialization of a successful ovarian cancer proteomic biomarker test will add fuel to an already red-hot biomarker market.