By Valerie Ross
Mira Dx this month responded to a team of investigators who raised concerns in a published report about the company's newly launched ovarian cancer risk test PreOvar, stating that their findings are “flawed” and based on “inapposite data.”
Duke University researcher Andrew Berchuck, a co-author of the study, defended the work, and told Gene Silencing News this week that the PreOvar test hasn't been well validated.
PreOvar is designed to identify a variant allele of the oncogene KRAS that Yale University researchers and Mira Dx co-founders discovered interfered with the microRNA let-7. In a paper published last year in Cancer Research, the investigators reported how the KRAS variant was present in more than 60 percent of cases of hereditary breast and ovarian cancer syndrome.
The test was launched in the US last June (GSN 7/22/2010).
However, a recent study by researchers working with the Ovarian Cancer Association Consortium and published in Clinical Cancer Research suggests that the variant allele, rs61764370, is not clinically useful in predicting ovarian cancer risk. The study, which was also presented earlier this month at the Society of Gynecology Oncology conference, compared genetic data from 9,352 women with ovarian cancer and 12,056 control subjects, all of whom were enrolled in other ovarian cancer case-control studies, to determine whether the marker was associated with ovarian cancer risk.
Because the genetic tests run in some of the Ovarian Cancer Association Consortium studies do not screen for rs61764370, the researchers instead used data on the nearby single-nucleotide polymorphism rs17388148, which has been shown to be strongly correlated with the SNP of interest, they wrote.
In three of the smaller genome-wide association studies whose participants were included in the current study, the Illumina genome-wide SNP arrays used to type the samples did not include direct data on rs17388148, either. For these studies, rs17388148 was imputed.
The researchers found no evidence of an association between the marker and familial, serous, or unselected ovarian cancer, or between the marker and ovarian cancer in women with BRCA1 mutations. “These data exclude the possibility of an association between rs61764370 and a clinically significant risk of ovarian cancer or of familial ovarian cancer,” the researchers wrote. “Use of this SNP for ovarian cancer clinical risk prediction therefore appears unwarranted.”
Mira Dx took issue with several features of the OCAC study, and said that the patients in the study were not similar enough to the patients for whom the PreOvar test is intended.
“Specifically, the study tries to compare findings primarily from patients with sporadic ovarian cancer to the population on which the PreOvar test was based — strictly and carefully defined high-risk patients from hereditary breast and ovarian cancer families,” Mira Dx CEO Martin Verhoef said in a statement. He called the study “a classic case of comparing apples and oranges,” and said that its authors “have either ignored or are clearly not aware of the intended use of PreOvar.”
The study further tracks “a substitute marker using imputed data with a known error of at least 10 percent, which likely fundamentally skews the data,” Mira Dx added, and the control groups in the study are of “unknown cancer status.”
Verhoef also said that numerous published and pending studies show that “there is no question of the importance of the KRAS-variant in cancer, and especially in women’s cancer.”
In follow-up comments provided to Gene Silencing News, Verhoef noted that the PreOvar test "is based upon an analytically validated TaqMan assay, with genotyping results independently confirmed in a CLIA-certified and rigorously controlled laboratory setting." In contrast, he said, OCAC's results "come from an industrial-scale effort combining multiple genotyping methods, secondary imputed data, and multiple platforms — all known and inherent sources of errors."
[ pagebreak ]
Berchuck, a gynecologic oncologist at Duke University, said that the study detailed in Clinical Cancer Research was designed as a follow-up for the KRAS-variant study in Cancer Research, which narrowed the target population for PreOvar to patients who have a family history of ovarian cancer but don't carry BRCA mutations.
The sample size, however, was too small to provide enough evidence, he said. “They're marketing the test for BRCA1- and BRCA2-negative familial disease, but that's based on only 31 cases.”
Verhoef told Gene Silencing News that the OCAC study "cannot justifiably be positioned as a follow-up" to the Cancer Research paper because "patient or control inclusion criteria were different, and a different marker was used in almost half of the study participants."
Further, he said, the sample cohorts in the OCAC study have "widely varying inclusion criteria." So while the study's sample numbers are larger, "the criteria are too random for any the conclusions they draw about PreOvar," he said.
Berchuck said that the Cancer Research study was also the reason that the OCAC team investigated whether the marker was predictive of risk for sporadic cancers. “The company wasn't marketing their test for sporadic ovarian cancer, but in the original publication, they did make the claim that the variant predicted an increases risk” of those cancers.
As for Mira Dx's assertion that the imputation technique OCAC used had such a high error rate, Berchuck said that “imputation is a highly accepted technique, and by no means is it 10 percent inaccurate. If you look at the studies in our paper imputed from genome-wide data versus those that were directly genotyped ... the odds ratio is 0.96 [for directly genotyped data], whereas those that were imputed were around 1.09.
“If there was an error, if anything it was in their favor,” he said.
Berchuck also said that he was concerned about the clinical impact of women making medical decisions based on a test that hasn't been borne out by solid evidence.
“A test like this, that hasn't been well-validated, would be used by women to asses their risk,” he said. “A concern would be having prophylactic surgery based on a perception of an increased risk that may not be there. Or on the other hand, maybe false assurance that they're not at risk if they don't carry this variant.”
Verhoef stressed that Mira Dx "markets only to physicians, allowing them to determine how to best use the PreOvar test results in the context of their patients’ family and personal clinical history."
Have topics you'd like to see covered in Gene Silencing News? Contact the editor
at dmacron [at] genomeweb [.] com.