Post-marketing validation of Rosetta Genomics' miRview Mets 2 has shown that the array-based test exhibited 88 percent concordance with clinical presentation and pathology on a set of cancer-of-unknown-primary samples, exceeding the accuracy of its predecessor PCR-based assay, miRview Mets.
Rosetta researchers presented a range of data on development and validation of miRview Mets 2 at the annual meeting of the American Society of Clinical Oncology earlier this month, demonstrating the test's concordance on CUP specimens, as well as results from earlier validation research showing that the Mets 2 test, which can identify 42 different tumor types, provided an accurate result in 85 percent of 509 blinded samples.
In a statement, Rosetta also disclosed additional results showing that miRview Mets 2 can be successfully performed on calcified bone, a common site for metastatic cancer. The company said that this is an important achievement considering that "most molecular tests are unsuccessful on decalcified, paraffin-embedded bone samples."
Ken Berlin, Rosetta's CEO, told BioArray News this week that the results presented at ASCO, in particular the test's concordance performance, are expected to help differentiate the test as the company steps up its marketing efforts.
Separately, Berlin discussed Rosetta's plans to launch a second array-based test for kidney cancer later this year. Rehovot, Israel-based Rosetta offers its menu of tests through a Clinical Laboratory Improvement Amendments-compliant lab in Philadelphia.
Regarding miRview Mets 2, Berlin said that the 88 percent concordance in true CUP patients is "important because these tests are all validated internally on known primary tumor samples, but really [they] are supposed to be used on those difficult-to-diagnose [metastases] cases and CUP cases.
He added that the test's performance in CUP is also important because “that’s one of the key areas where doctors want to use tests such as these... We certainly plan to use that as a differentiator because we think, looking at other published data out there, that 88 percent is the best that we've seen in the class," Berlin said.
In its ASCO poster, Rosetta said that miRview Mets 2 relies on two classifiers to determine a tumor's tissue of origin — "a binary decision-tree classifier and a k-nearest-neighbors" classifier, both of which assign a result based on the expression of 64 microRNAs. Each classifier predicts one of 42 tumor types and if the two predictions agree, they are incorporated into a single origin diagnosis.
If the predictions disagree, they are ranked by the positive predictive value of each answer.
According to the poster, the test's performance was established on a 509-sample validation set. In 418 of the 489 samples, the reference diagnosis was predicted by at least one of the two classifiers, resulting in an overall sensitivity or positive agreement of 85 percent. Specificity, or negative agreement, was above 99 percent, Rosetta reported.
For 403 "single-prediction cases" — samples that resulted in a single prediction from both classifiers — the sensitivity of the assay was 90 percent.
Ayelet Chajut, Rosetta's executive vice president of R&D and the company's head of molecular biology, noted that the poster included additional data that demonstrated the assay's performance on 55 RNA samples from 52 CUP patients.
"We wanted to see if we could get better results with the new version" of the test, she told BioArray News. "So we took the RNA we had from the same samples [that were used to validate miRview Mets], and studied this cohort on the miRview Mets 2," Chaiut said.
On the CUP samples, miRview Mets 2 showed 88 percent concordance with clinical and pathological data, up from 80 percent for the previous version, she said. According to the poster description, concordance was measured on a four-stage scale, with a clinical match, a pathology match, or pathology unable to rule out the test results counting toward the 88 percent total concordance.
After resolving a dispute with its with one-time partner Prometheus Laboratories last year, Rosetta has taken over its own marketing for the test, and recently began a new commercialization effort. Rosetta had previously relied on Prometheus to market its test in the US (BAN 3/23/2010).
"We just hired four sales reps at the beginning of May," Berlin said. "It's still early but we're starting to see some traction, and hoping to really continue to pick up the pace in the third and fourth quarters of this year."
According to Berlin, Rosetta plans to focus its marketing energy on miRview Mets 2 in the near term, though it has two other tests slated to launch this year.
"As we start out and really relaunch with our own selling effort, we want our primary focus to be on Mets 2. That's what we believe is the biggest near-term opportunity," he said.
"And as we gain traction there, we expect to expand the focus to the lung test, which we expect to launch [soon]."
The "lung test," is Rosetta's PCR-based miRview Lung, which differentiates four main types of primary lung tumors, according to Chaiut, and is slated for launch in mid-July, according to Berlin.
By the end of the year, the company also plans to launch its next array-based test, a kidney assay called miRview Kidney, which is being developed to differentiate the four main types of kidney tumors with what Chaiut says is "very, very high sensitivity and specificity."
According to Berlin, other assays in the works for Rosetta include a prognostic for mesothelioma and a bladder cancer risk stratification test that the company hopes to launch next year.
Berlin said the company plans to develop all future tests except for miRview Kidney on a PCR platform rather than following on miRview Mets 2's array technology. However, he said, the company has not made the final decision on that front.
Berlin said the company plans to use either the PCR platform or the array platform to commercialize future tests, and has not yet decided which platform might be ideal for each assay.
Have topics you'd like to see covered in BioArray News? Contact the editor at mashford [at] genomeweb [.] com.