NEW YORK (GenomeWeb) – Researchers working within the UK's Early Cancer Detection Consortium are developing a multi-omic blood-based biomarker panel for early cancer detection.
Unlike most cancer biomarker efforts, which typically focus on a single cancer or a specific question within a cancer type, the UK team aims to build a test capable of detecting multiple kinds of cancers, Ian Cree, a University Hospitals Coventry and Warwickshire researcher and one of the leaders of the project, told GenomeWeb.
The focus, he said, is not on developing a test for diagnosing specific types of cancer, but rather for detecting, as early as possible, that a patient likely has a cancer of some sort. Patients flagged by the test could then undergo procedures like imaging or endoscopies to identify the specific nature of their cancer.
Cree and his ECDC colleagues have established a company, PinPoint Cancer, to fund the test's development and to commercialize any product that emerges from their research. He said the company is looking to raise around £10 million ($13 million) to support its activities.
As Cree noted, a test like the one the consortium and PinPoint are pursuing will require very high sensitivity and specificity. "To get to the point where the health economics works, you have really got to be detecting a very high proportion of your cancers," he said. "Otherwise you will miss too many [and miss the benefit of early detection] or have too many false positives, and then you spend a lot of time and money following up those false positives." Cree said he believed that for the test to be effective, it would need to perform with an area under the receiver operating curve (AUC) of .95. AUC is measure of a test's performance, with an AUC of 1 meaning a perfect test with 100 percent sensitivity and 100 percent specificity. An AUC of .95 would indicate sensitivity and specificity in the 90 percent range.
If cancer biomarker work to date is any guide, that will be a very challenging target to hit. By way of comparison, few if any commercially available proteomic tests offer this level of performance, even though they typically have far narrower uses than the pan-cancer early detection Cree and his colleagues are pursuing.
Take, for instance, Vermillion's Overa test, a five protein panel intended to help determine whether an ovarian adnexal mass detected in a patient is likely benign or cancerous. The test is able to identify cancers with a sensitivity of above 90 percent. Its specificity, however, is only in the 70 percent range. This nonetheless marks a significant improvement over Vermillion's first version of the test, OVA1, which had specificity in the 50 percent range.
Likewise, Applied Proteomics' SimpliPro Colon test, which is intended for assessing the risk of advanced adenoma or colorectal cancer in symptomatic patients, detects advanced adenoma with 45 percent sensitivity and 80 percent specificity; stage I and II CRC with 75 percent sensitivity and 78 percent specificity; and stage III and IV CRC with 88 percent sensitivity and 78 percent specificity.
Indeed, while in the early days of genomic and proteomic biomarker research, much was made of the possibility of identifying sets of markers for early detection of cancer in the general population, this has proven an elusive goal. Broadly speaking, these fields have over the years shifted to narrower questions, such as identifying markers for monitoring cancer patients for recurrence, or aiding in the interpretation of features identified during imaging or other clinical procedures.
Cree acknowledged that the ECDC's goal is quite ambitious, but he said that the researchers believed they could achieve better performance than has been seen thus far by combining markers from various omics disciplines. To this end, they undertook a literature review, looking at some 3,990 biomarker studies published between 2010 and 2014. This effort, which they detailed in a paper published last month in EBioMedicine, identified a total of 814 potential markers, ranging from proteins to circulating DNA to metabolites to microRNAs.
Echoing many in the field, Cree noted the relatively small dividends biomarker research has paid to date, despite significant investments.
"We have had twenty years, 20,000 papers, however many dollars that translates into, but we haven't actually gotten very far," he said. One issue is the relatively poor quality of many published biomarker studies, most of which, he noted, look at relatively small cohorts and some of which don't even have control groups.
In their literature search, the researchers weeded out any studies without controls and done in fewer than 50 human subjects, Lesley Uttley, a researcher at the University of Sheffield and first author on the paper, told GenomeWeb.
"The aim was to include [only] biomarkers that are at a stage where they have faced some clinical validation," she said.
The next step, Cree said, will be a roughly 6,000-patient case-control study through which the researchers hope to evaluate the markers further. He said he anticipated that study would take around three years, at the end of which the researchers hope to have a CE-IVD-marked panel that can be used first in a research setting, and then clinically, for early detection across a range of common cancers.
They hope to follow that up with a longitudinal study on the order of 300,000 patients. Cree acknowledged that this was a rather ambitious goal but he said that the ECDC's clinical network, which encompasses 23 UK universities and their associated National Health Service hospitals, make it feasible.
"The thing that kills clinical trials is recruitment," he said. "But we are confident, because we have a population size between the hospitals involved of about 10 million to 15 million people. So we have a large population and large numbers of cancers coming through."
He cited as a similarly scaled effort the ongoing UK Collaborative Trial of Ovarian Cancer Screening (UKCTOCS) trial, which has followed some 200,000 women since 2001. Data from that trial was used in the development of Abcodia's Risk of Ovarian Cancer Algorithm (ROCA) test, which uses longitudinal measurements of the ovarian cancer protein marker CA125, combined with transvaginal ultrasound, to enable early detection of the disease.
Cree said he and his colleagues might likewise explore the potential of longitudinal biomarker measurements for early cancer detection, though he noted that was still "a long way down the line."