This story originally ran on Dec. 10.
By Tony Fong
When Vermillion filed its 510(k) application with the US Food and Drug Administration for its ovarian cancer diagnostic in June 2008, the company was facing two daunting obstacles.
Its cash reserves were fast evaporating and with no revenue and a tanking stock, the company was approaching extinction. A thumbs-up from the FDA on the test, called OVA1, could change all that, but the agency’s history with proteomics-based tests made the picture for the test at best doubtful.
According to one estimate, the agency had previously cleared only 109 proteins for use in biomarker-based tests, none of which had been discovered by proteomics methods or technologies.
Then in late March of this year, Vermillion slid further into the abyss. Faced with mounting debt but no means of raising cash, the company was forced to file for Chapter 11 bankruptcy protection [See PM 04/02/09].
At that point, there were no other potential sources of revenue for the Fremont, Calif.-based company, and if OVA1 did not meet approval with the FDA, the firm would in all probability be forced to close its doors.
This past Sept. 11, however, in defiance of the odds and expectations, regulators cleared OVA1, making it the first proteomic in vitro diagnostic multivariate index assay to receive FDA clearance [See PM 09/17/09]. As a result, a once-moribund firm appears to be on the road to revival and is now on the cusp of emerging from Chapter 11. Its stock, which had nosedived to a low of $0.01 per share as recently as July 29, rebounded to a high of $23.70 on Nov. 12. Most recently it has traded at about $20 per share.
Next month, the US Bankruptcy Court in Delaware will hold a hearing on Vermillion's recently filed Chapter 11 reorganization plan [See PM 12/04/09]. If the court approves the plan, OVA1, which the company said is the first FDA-cleared laboratory test that can indicate the likelihood of ovarian cancer with high sensitivity before a biopsy or exploratory surgery, could also be known for something else: the little test that saved it.
Step 1: Find Markers, Any Markers
The serum-based OVA1 assay is composed of five biomarkers — transthyretin; apoliprotein A-1; beta2-microglobulin; transferrin; and cancer antigen 125. It combines the results of five immunoassays into a single numerical score. According to Vermillion, it is indicated for women who are older than age 18, who have an ovarian adnexal mass present for which surgery is planned, and who have not yet been referred to an oncologist.
Its story begins a decade ago when Vermillion, then called Ciphergen, formed a collaboration with Daniel Chan, director for the Center for Biomarker Discovery at Johns Hopkins University.
In 2000, the company began establishing Biomarker Discovery Centers in California, Pennsylvania, Japan, and Denmark in an effort to discover and develop biomarkers using its surface enhanced laser desorption/ionization, or SELDI, platform.
Three different types of centers were established to carry out that goal. One was based on a fee-for-service model to provide proteomics and protein biomarker services to the pharmaceutical and biotechnology sectors. Another BDC model saw Vermillion partner with clinical investigators who would supply the company with clinical samples for analysis in-house.
The third model was one in which Vermillion provided funding to academic partners, who would then carry out the research in their own labs while Vermillion provided technical support. The BDC at Johns Hopkins was the only one that followed this model.
According to Eric Fung, the chief scientific officer at Vermillion, the collaboration with Chan was originally targeted generally at cancer.
[ pagebreak ]
"We hadn't really honed in on ovarian cancer," he told ProteoMonitor. "We were doing work with Dr. Chan in prostate cancer, colon cancer, pancreatic cancer, and breast cancer … [and] had a number of discovery programs in all of these different types of cancers."
By then Chan already had a reputation as an expert in tumor biomarkers because of his association with the Johns Hopkins University School of Medicine, where he had access to a large collection of clinical samples. Just as importantly, Fung said, Chan had a "longstanding history of working with diagnostic companies and understanding the process of diagnostic [development]."
At that point, while Vermillion had already developed protocols for plasma and serum fractionation, as well as bioinformatics work, on the SELDI, it was still new to the diagnostic development universe and unschooled in how to bring discoveries through the pipeline to the clinic.
"We thought that [Chan's] mix of understanding the science and understanding the commercialization aspect of it would be a very important set of skills to utilize," Fung said.
When the choice was made to specifically target ovarian cancer biomarkers, Fung said, both clinical utility as well market potential drove the decision.
Because widely used and accepted testing methods for breast cancer and prostate cancer — mammography and the PSA test, respectively — existed, they were ruled out, he said. Discovering and developing biomarkers that could supplant or even complement those tests posed too much of a challenge.
"In contrast, ovarian cancer doesn't really have a test out there per se," Fung said. While CA125 is cleared by the FDA for monitoring recurrence of ovarian cancer in women who have already had the disease, "in point of fact, it's widely used for pre-operative diagnosis," he said, "and we felt that … there was a market opportunity for a novel test."
According to Chan, the first important study stemming from the collaboration was one done in 2001-2002 that examined 81 archived specimens from Johns Hopkins Hospital. "This was the first study that we did using SELDI for ovarian cancer," he told ProteoMonitor.
He and his co-researchers identified three potential candidate biomarkers: transferrin, haptoglobin, and immunoglobulin heavy chain. Individually, the three biomarkers performed no better than CA125, but in combination with CA125, transferrin and haptoglobin achieved sensitivity of 94 percent, compared to 81 percent with CA125 alone.
That study was followed by a larger one that investigated samples from 503 patients for profiling. This time the research was expanded to include multiple sites and collaborators from MD Anderson; Duke University Medical School; the Royal Hospital for Women in Sydney, Australia; University Hospital Groningen, The Netherlands; and Queen Mary's School of Medicine in the UK.
Three more candidate protein biomarkers were identified: apolipoprotein A1 and a truncated form of transthyretin, both of which were down-regulated in cancer, and a cleavage fragment of inter-alpha-trypsin inhibitor heavy chain H4, which is up-regulated in cancer.
A study describing the work was published in 2004 in Cancer Research. In it, the researchers reported the three markers improved correct detection of stage 1 and stage 2 of ovarian cancer to 74 percent from 65 percent.
"So it was a modest increase over CA125 in terms of detecting early-stage ovarian cancer," Fung said. At that point, although Vermillion had the set of biomarkers, it had yet to translate them into a test. Questions still remained regarding the proper clinical scenario for use of the markers, the appropriate population for them, and how a physician should act on the results, among other factors.
[ pagebreak ]
Honing in on Intended Use
It was not until the following year that Vermillion began considering using the biomarkers to develop an FDA-approved test that could help distinguish between benign and malignant tumors. Such a test would help physicians better decide when to refer a patient to a gynecologic oncologist.
According to Fung, because of the nature of ovarian cancer, translating the biomarkers it had discovered into a generalized screen was not feasible. Ovarian cancer is a relatively "low prevalence" disease, which means that for a biomarker to have effect as a screening tool, it would need be extremely specific for the disease.
According to the National Cancer Institute, 21,550 new cases of ovarian cancer are projected to be diagnosed in the US in 2009, and about 14,600 patients will die of the disease. By comparison, 194,280 new cases of breast cancer will be diagnosed this year, with 40,610 deaths.
"We're talking about over 99-percent specificity" for a biomarker to be useful as a screening tool, Fung said. "And certainly … we had not identified any marker with that type of specificity. I don't think anybody has."
Even if they had identified such a highly specific marker, validating it would have been prohibitively expensive and time-consuming.
"Because this was a relatively low -prevalence disease, it really takes tremendously large studies to demonstrate both the utility and performance of the marker," Fung said. "And we knew we could not afford that kind of trial both in terms of costs and time."
Having decided on an ovarian tumor triage test, the company then embarked on a multi-institutional study running more than 600 samples to find markers that would be able to differentiate malignant tumors from benign ones. Initially a list of between 20 and 35 biomarkers were identified as possible candidates. That number was whittled to seven biomarkers that appeared consistently across different sample sets. Some also had been previously described in the scientific literature, "so we knew that they were robust across different populations," Fung said.
In one presentation made in the spring of 2006 of a prospective clinical study comprised of more than 400 patients divided between patients with ovarian cancer and healthy controls, the seven markers combined with CA125 had a positive predictive value in a blinded test set of 80 percent, compared to 59 percent for CA125 alone.
In a follow-up study later that year comparing 183 women — 42 with ovarian cancer, 65 with benign tumors, and 76 with gastrointestinal disorders — the panel demonstrated a specificity of 94.3 percent and a sensitivity of 78.6 percent, demonstrating its ability to differentiate cancer from benign ovarian disease, Vermillion said at the time.
Two of the three biomarkers described in the Cancer Research article, transthyretin and apolipoprotein A1, were included among the seven biomarkers. The third, the ITIH4 fragment, was excluded because of its instability, though Vermillion continues to evaluate it.
In July 2005, Vermillion and Quest entered into an agreement to commercialize three assays that were under development at Vermillion. At the time, though, both companies refused to identify what those assays were, it was widely assumed that the ovarian cancer test would be one of them [See PM 07/29/05].
By the end of 2005, Vermillion had started planning for the submission of its ovarian cancer test to the FDA for 510(k) clearance, a class of tests the agency considers to be of minimal risk to patients and "equivalent to a predicate," or a device that is already on the market.
While no proteomics-based IVDMIA had ever been cleared by the FDA, Vermillion felt "it was the cleanest way to get this test out to the marketplace," Fung said, and in spite of its relationship with Quest Diagnostics, Vermillion did not seriously consider offering OVA1 as a CLIA-based test.
[ pagebreak ]
In anticipation of the commercialization of its test, Vermillion also began taking preparatory steps. In 2006, it formed a scientific advisory board, which included Chan; Robert Bast, who developed the CA125 antibody that led to the CA125 radioimmunoassay; and Ian Jacobs, who heads the UK Collaborative Trial of Ovarian Cancer Screening aimed at the evaluation of screening techniques for the disease.
The company ended 2006 by hiring the Emergo Group to help OVA1 win regulatory approval in Europe, and hiring Mosaic Health Care Consultants to develop reimbursement strategies and programs for it.
In early 2007, Vermillion began the final step before filing with the FDA — initiating a prospective clinical trial for OVA1 in order to demonstrate its improved performance over current standards of care, which were physical and radiological exams. The trial, begun earlier that year, eventually enrolled 550 patients across 27 clinical trial sites.
Fred Ueland, an associate professor of gynecological oncology at the University of Kentucky, was the principal investigator on the trial. Vermillion felt it was important to have a specialist "who sees patients with ovarian cancer on a daily basis running the show," Fung said.
In addition, while some clinical trials are conducted by the original researchers who discovered the biomarkers, Chan said that it was important that he exclude himself from the OVA1 clinical trials in order to avoid any conflicts of interest.
By then, Vermillion had already been in contact with the FDA about its plans to submit an application. Fung declined to say much about the nature of the talks Vermillion had with the FDA.
He said his company and the agency had multiple discussions centered on "clinical need in the ovarian cancer state," as well as "how to frame the intended-use statement."
A month after the prospective trial began, Vermillion presented data at the Society of Gynecologic Oncologists' annual meeting showing the test correctly identified 84 percent of true ovarian cancer cases, compared to 33 percent using standard diagnosis methods without the test.
Then in March 2008, the company announced preliminary results from the clinical trial that demonstrated it could stratify women with pelvic masses into high- and low-risk categories, helping doctors to determine which patients should be referred to a specialist prior to surgery.
According to Vermilion, the test was able to rule out malignancy with 95-percent certainty, and had 90-percent sensitivity for detecting malignant ovarian tumors.
During the prospective trial, the company eliminated two of the seven biomarkers to come up with the final five that now comprise OVA1. Fung declined to identify the two excluded markers, citing confidentiality, but said that there were issues around their analytical and/or clinical stability.
On June 25, 2008, Vermillion announced it had submitted OVA1 for 510(k) clearance from the FDA.
For the next 15 months, the company waited as the agency made its decision. In the meantime, its finances continued on a steep decline. In 2006, it sold its SELDI business to Bio-Rad Laboratories in order to concentrate on its diagnostic development programs. That left Vermillion essentially with no revenue source, and all the while that it was developing OVA1, it issued warnings in documents filed with the US Securities and Exchange Commission that if it could not find a source of income soon, it faced the possibility of having to shut down.
In April 2007, the company's independent auditor PriceWaterhouseCoopers gave a "going concern" qualification to Vermillion's annual report, meaning PwC had questions about the company's ability to stay in business.
Before it submitted its 510(k) application, the company's stock had run afoul of Nasdaq minimal listing requirements. In July of 2008, the firm enlisted the help of investment bank ThinkPanmure to explore ways of raising shareholder value, but on Sept. 25 after seven months of meetings with Nasdaq officials and maneuvers to pump up its share value, the exchange delisted Vermillion's stock [See PM 09/25/2008].
[ pagebreak ]
Then, on March 30, Vermillion filed for Chapter 11 protection. In its filing with bankruptcy court, the company said that it and the FDA had had several discussion regarding its OVA1 application, and that it "continues to work diligently to provide information and data in response to FDA's inquiries related" to the submission.
As a result of the Chapter 11 filing, the company halted the development of its other tests, which are directed toward peripheral artery disease, thrombotic thrombocytopenic purpura, and early-stage ovarian cancer. It was a gamble: If the agency turned down the application, it likely spelled the end of Vermillion.
While the company understood the bleakness of the situation, Fung said that its correspondence with the FDA also suggested to company officials that some good news was brewing.
"I certainly didn't know the exact date that it would be cleared, but based on some of the questions we needed to answer, I had a pretty good idea we were going to be cleared," he said.
On Sept. 11, 2009, Vermillion got what it had been waiting for: The FDA sent the company an e-mail saying it could begin selling OVA1 in the US.
Being Demanding in Discovery
The company had achieved what no other proteomics firm had accomplished. While other protein-based tests and genomics-based multivariate tests had been cleared by the FDA, OVA1 was the first true proteomics-based multivariate test to gain such status, Chan said.
From the beginning, he added, he and Vermillion had been told by others in the proteomics community that the FDA would never approve such a test because a multivariate panel would be too complicated and the agency wouldn't know how to evaluate OVA1 properly.
How did Vermillion do it, then?
First, Vermillion and its collaborators performed much of the heavy lifting during the discovery stage, performing multiple studies at multiple sites and with multiple population groups to arrive at their candidate biomarkers. By adopting such stringent criteria early on in the process, Chan said that he and his collaborators improved the chances that biomarkers identified in the discovery stage would survive the validation stage.
"I think most markers and most studies fail because they go through the validation step and it just doesn't pass the clinical performance," he said. "I think people should take a different viewpoint now, saying, 'Hey, in discovery we really need well characterized, well-designed studies.' Don't try to wait until the validation step.
“A lot of people try to spend not as much money in the beginning because they think, 'If I cannot validate this, I won't have any money, so let's try not to spend so much money'" on the discovery stage, he added.
In the validation step, he also recommended that samples from one population be used for training and samples from a different population be used for testing.
Finally, the decision to go after a narrow market, benign versus malignant tumors, "was a key for this particular test … because it's clinically useful and it [was] doable in terms of proving [OVA1's] intended use."
Finally, now that one proteomic IVDMIA has successfully scaled the FDA mountain, Chan said he believes it would provide both inspiration and a roadmap for other diagnostic firms developing similar tests.
One result of OVA1's clearance is that he has been asked by scientific journals to write about his part developing the test. "They would like to publish a success story," he said. "Obviously not every proteomic test is going to follow the same path, but the fact that Vermillion was successful in getting this approved and used clinically will help other companies and other investigators [to achieve] that," Chan said.
As for Vermillion, as it awaits a decision from bankruptcy court on its Chapter 11 reorganization, it is working with Quest on commercializing OVA1 in the US. The test is expected to hit the market in the first quarter of 2010. Recently, Gail Page, the former CEO of Vermillion and current executive chairperson of the board, said that it once OVA1 has become available in the US, the company hopes to bring it to market in Europe and Japan.