In a prospective study published this month, patients with cancer of unknown primary who were tested using BioTheranostics' CancerType ID assay, and who then received chemotherapy specific to their predicted tumor of origin, survived longer on average than expected for standard CUP treatment.
In the study, published in the Journal of Clinical Oncology, a team from the Sarah Cannon Research institute used CancerType ID on 252 patients, of which 98 percent had a successful prediction of origin. Of these, 194 went on to receive assay-directed chemotherapy.
Overall, the patients in the study who received site-specific treatment survived an average of 12.5 months, compared to the less-than-10-month average survival seen in previous large studies of CUP using standard empiric chemotherapy strategies.
Patients who were predicted to have more treatment-responsive tumor types, like breast and ovarian cancer, had much longer survival than those predicted to have less responsive cancers — an average of 13.4 months and 7.6 months, respectively. Although patient subgroups for any one tumor type were small, the researchers found that for some of the most responsive types, average survival was higher than 20 months.
John Hainsworth, the study's first author, told PGx Reporter that the study is the first prospective trial using a molecular tumor of origin test to guide patient treatment.
Several tests to identify the origin of CUP tumors have been developed. Pathwork Diagnostics offers its Tissue of Origin Test, cleared by the FDA in 2010, through its CLIA-certified lab. Rosetta Genomics also offers a microRNA-based test, miRview mets, to identify the tissue of origin of CUP metastatic tumors.
While Hainsworth and his team have worked with BioTheranostics through the development of CancerType ID, he said that he doesn’t believe the company's assay is better or worse than others. However, having this new prospective evidence offers a further level of validation, he said.
CancerType ID measures the expression of 92 genes to distinguish 28 tumor types and 50 subtypes based on a proprietary algorithm and database. The test has demonstrated sensitivities of 87 percent for the 28 main tumor types and 82 percent for the 50 subtypes, according to BioTheranostics.
Catherine Schnabel, BioTheranostics' senior director of medical affairs, told PGx Reporter this week that the laboratory-developed test is performed in the company's CAP-accredited, CLIA-certified lab.
BioTheranostics' CEO Richard Ding told PGx Reporter that the company sees the Sarah Cannon prospective trial as setting CancerType ID somewhat apart from most other biomarker TOO tests.
"Typically biomarker diagnostic tests don't go through a prospective trial," he said, "and definitely not with overall survival as the clinical endpoint. That is a very high bar, so to speak."
"John Hainsworth and Anthony Greco are internationally recognized researchers. So I think having done the trial with a center of that caliber really does substantiate the clinical utility of the test. And, from the science perspective, [it] positions us well because we have really advanced the field in a very significant way," Schnabel added.
"Both those things should hopefully help with our clinical adoption," she said.
Hainsworth and his colleagues enrolled 289 patients in the study. Only 252 had enough tumor tissue available for the assay to be performed, and, of these, 98 percent, or 247, were successfully given a CancerType ID prediction.
However, only 194 of these patients went on to have site-specific therapy, such as taxane/bevacizumab for breast cancer, gemcitabine/erlotinib for pancreatic cancer, or other standard first-line treatments depending on the predicted origin.
Hainsworth said two factors contributed to this. First, the process of getting consent and performing the assay — about two weeks — combined with additional testing in some cases, was long enough that some patients no longer met the criteria for the study.
"They either had declined a lot, or had a brain metastasis, or something that would throw them out," Hainsworth said.
Another group of patients would still have been good candidates for the study, according to Hainsworth, but did not receive assay-directed therapy based on either the patient's or the doctor's choice. In some cases, patients did not want to receive a more severe treatment based on their predicted tumor type. In others, physicians didn't believe or trust the assay results, he said.
Hainsworth said that the average survival in the remaining 194 patients who did receive assay-directed chemo — 12.5 months — was the "highest result obtained in a large group of CUP patients with any type of treatment."
More importantly, though, he said, were the improvements for those patients predicted to have more responsive tumor types.
"Some CUP patients have cancers that hardly respond to treatment anyway, so even if you know what it is, your chances of having much success are small," he explained.
"So what was really encouraging [was that in] the patients you'd expect to have more responsive cancers — breast, ovarian — those did a lot better than the median: up in the high 20's in months of median survival."
The trial was not randomized, so the survival of patients treated using CancerType ID can only be compared to survival in earlier trials.
Hainsworth said that there are a number of problems with doing a randomized trial in this setting. First, he said, most patients and doctors would not consent to randomization. Secondly, the study would have to be very large to cover enough subjects with different cancer types.
Two particular factors could be confounding, Hainsworth said. First, patients with tumor types that are very aggressive and respond poorly to therapy are likely to do poorly whether treated according to type or not. On the other side of the spectrum, for some highly responsive tumor types, tumor-specific therapy turns out to be almost identical to what would be used based on standard empiric methods in CUP, meaning those patients would do well regardless or what treatment arm they were in.
"It's just small numbers of cancer types right now where you would give a totally different treatment for the cancer versus empiric treatment for unknown primary," Hainsworth said.
However, he highlighted that fact that as more and more molecularly targeted cancer therapies are developed, the utility of tests like CancerType ID will only increase.
"With all these new drugs being added that are specific to one type of cancer, the differences in the way you treat patients are getting bigger and bigger," he said.
"If you have these assays and you trust them to give the tissue of origin they are going to be more valuable in the future."
But even with limited treatment options, Hainsworth said that the evidence in the recent trial is convincing enough to support using CancerType ID or other tests in clinical practice now.
"My guess is that this study will convince people that it's reasonable and probably improves overall group survival, and definitely the subgroup survival for more responsive tumor types, he said."
Currently, molecular testing for tumor of origin is not recommended by professional guidelines, he said.
This may be partly because of a lack of prospective evidence, he added. "You couldn't really say or even address the question of whether patients would respond better or have better survival" using these tests, he said.
BioTheranostics' Ding said that the edition of the National Comprehensive Cancer Network guidelines published before Hainsworth's study suggested that lack of patient outcome evidence is an issue preventing the recommendation of molecular tumor-of-origin testing.
"So we anticipate this [study] and additional data will move the field forward," he said.
Hainsworth said his group is now working on a follow-up study using competing CUP assays to analyze some of the samples from the CancerType ID trial to see how the predicted tumor type results correlate across the different platforms.
He said he is also planning a study combining tumor-of-origin testing with other molecular testing, like HER2 and BRAF tests, to guide even more targeted therapy.
Ding said that a "few thousand" physicians use the CancerType ID test so far, most as early adopters, but some practices use the assay "more regularly."
Last year, Medicare contractor Palmetto GBA authorized reimbursement for CancerTYPE ID as a covered benefit for Medicare Part B patients.
According to Ding, the company is also getting reimbursement from "most" private payors, but not all.
To that end, the company is in the process of completing a health economic study, he said.
Earlier this year Pathwork Diagnostics released a retrospective health economic analysis showing that the use of its TOO test to guide treatment should result in longer survival for CUP patients at a cost per quality-adjusted life year of around $47,000 — less than half the generally accepted threshold of $100,000 for cost-effectiveness in the US, according to the company (PGx Reporter 1/18/2012).