A team comprising researchers from Genomic Health and the Cleveland Clinic have published a study describing the analytical validation of Oncotype DX prostate, the company's RT-PCR-based gene expression assay intended to predict prostate cancer aggressiveness and severity from fine-needle biopsies to help guide treatment decisions for men with early-stage disease.
The study, published in BMC Genomics in October, validated the test by analyzing its precision and accuracy in measuring gene expression over a wide range of sample sizes and other variable conditions, including the small amounts of cancer tissue typically available from prostate needle biopsies.
Oncotype DX prostate, launched earlier this year, analyzes the expression of 17 genes to gauge prostate cancer aggressiveness, reporting a score from 0 to 100, with a lower score indicating that a patient might be able to avoid treatment in favor of continued active surveillance.
Genomic Health has previously presented data on the test's clinical validity, and future publications and presentations are also expected to follow the new report on the test's analytical validity with more data on both clinical validity and clinical utility.
"People say a bad test is as bad as a bad drug, but a good test is also as beneficial as a good drug," Steve Shak, Genomic Health's executive vice president of research and development, told PCR Insider. "This is a great test, and that's what this publication really highlights," he said.
According to Shak, Genomic Health had to make several changes to the sample-handling and PCR methodology underlying the assay to make it work with the extremely small sample sizes that are often seen in prostate needle biopsies.
The company's flagship test, Oncotype DX breast, requires about 300 ng of RNA per sample, whereas in the recently published validation study, the prostate assay was shown to accurately measure gene expression over a wide range of inputs, from .005 ng to 320 ng, Shak said.
In the BMC Genomics study, the Genomic Health and Cleveland Clinic team described how the assay was optimized to work with these biopsies — as small as one millimeter — and presented data from experiments validating the test's performance under a range of different experimental conditions.
According to the study authors, to establish a target for the lower limit of RNA input for the assay, the group extracted RNA from 46 and 167 biopsies from two separate studies to examine the distribution of RNA yields. Among these samples, RNA yields on average were less than 50 ng with the lowest quarter of samples containing between 19 and 34 ng of RNA.
Anticipating even lower levels of RNA in some biopsies due to variability in clinical practice, the team then targeted an RNA input for the assay below this lowest quartile's lower limit.
Changes to the test platform to allow for such small RNA inputs included the addition of a multiplexed preamplification step after reverse transcription to create multiple copies of the starting RNA prior to quantitative assessment of gene expression, the use of robotic liquid handling to reduce the volume of reverse transcription and qPCR reactions, and combining the detection step with the quantitative PCR step, the study authors wrote in their report.
According to the group, this new preamplification was conducted using custom TaqMan PreAmp Master Mix and custom forward and reverse primers for each target gene. On the detection end, the team also switched from Applied Biosystems' ABI 7900 to the Roche LightCycler 480 for final PCR and detection, Shak said.
Instead of using RNA mass as the criteria for whether a sample can be assessed by the assay, the group also designed Oncotype DX prostate to instead use the expression of reference genes to indicate whether a sample is of sufficient quality, which allows for a variety of input sizes, the study authors wrote.
In demonstrating the test's analytical validity, the researchers tested amplification efficiencies, analytical sensitivity, and the test's accuracy over the aforementioned range of sample input sizes.
The researchers found that the median amplification efficiency was almost 94 percent across the 17 genes in the test, with efficiencies for each gene falling within about 6 percent on either side of that value. All the individual gene assays had limits of detection that surpassed pre-specified criteria, the group wrote, and analytical accuracy was also "excellent."
The team also tested the reproducibility and precision of the test by analyzing 10 prostate cancer needle biopsy samples over multiple instruments, reagent lots, operators, days, and RNA input levels, finding that the test's precision deviated only by low amounts under these variable clinical conditions.
"All this analytical performance certainly wouldn't matter if this wasn't yielding a result that is critical for prostate cancer decision making," Shak said. "It was that clinical context of the patient that led us to optimize and develop the performance that the assay has" and that is shown in this paper.
"We focused the prostate test validation on low-risk patients, [who] are most challenged in deciding [whether they] should get aggressive treatment or active surveillance, and those were frequently the smallest of specimens," Shak added. "With breast cancer and colon cancer, we have the ability to work with relatively large paraffin specimens … but those look gigantic compared to the needle biopsies at the time of diagnosis of prostate cancer."
According to Shak, the company has not adapted any of the changes that allow the prostate test to operate with much smaller inputs back into its breast cancer or colon cancer assays — which still require the same input volumes — and said he did not want to speculate on whether that might happen in the future.
He said that Genomic Health has several other studies that have been completed on the prostate test, as well as three studies in progress that deal with additional questions, such as how the test impacts clinical practice.