SAN FRANCISCO (GenomeWeb) – A group of researchers from multiple academic institutions and Veracyte have demonstrated that Veracyte's RNA sequencing-based test for classifying thyroid nodules as benign performed better than a previous version of the firm's test that was based on microarray technology.
The researchers described the test's performance in a validation study published this week in JAMA Surgery. Veracyte funded the study but was not involved in data interpretation.
The researchers analyzed the same set of samples used to validate the earlier microarray version of the test, but ran the RNA sequencing-based version on them retrospectively in a blinded fashion. The new test, which Veracyte markets as Afirma Genomic Sequence Classifier (GSC), had a sensitivity of 91 percent and a specificity of 68 percent, up from 89 percent sensitivity and 50 percent specificity for the Afirma Genomic Expression Classifier (GEC).
Veracyte has already launched Afirma GSC, and earlier this month it noted that around 70 percent of its customers had converted from the GEC to the GSC. It also recently launched its next test in the Afirma product line, which uses RNA sequencing but is geared more toward helping physicians make treatment decisions for patients rather than helping physicians determine whether a patient's nodule is likely benign and does not need to be surgically removed.
Paul Ladenson, a senior author of the study and professor of endocrinology and metabolism at Johns Hopkins University, who is also a consultant for Veracyte, said that the study shows that the Afirma GSC improves on the previous version of the test. He noted that for one particular type of thyroid nodule called Hürthle cell, there was "a dramatic improvement." The GEC test identified around 12 percent of benign cases, while the GSC identified 59 percent of benign cases, he said.
Just around 5 percent to 15 percent of thyroid nodules are malignant, and malignant nodules are treated with surgery, but identifying nodules that should be surgically removed from benign ones can be tricky. Currently, fine needle aspirate biopsies are analyzed via cytology, but in about one-third of cases, the cytology findings are indeterminate. Compounding the problem is that "lumps in the thyroid gland are being recognized with increased frequency," Ladenson said. The Afirma test is designed to analyze samples where cytology is unable to determine if a nodule is benign or malignant.
In the JAMA Surgery study, the researchers built on the Afirma GEC validation study, which analyzed more than 4,800 nodules sampled by fine needle aspiration. Of the 210 that were deemed indeterminate by cytology and used to validate the GEC, 191 had enough leftover material to be analyzed via the GSC test.
The GSC test uses RNA sequencing and machine learning to predict whether a nodule is likely benign or suspicious for being malignant. Similar to the GEC, the new test also analyzes expression data from 1,115 core genes, but it also incorporates information from specific gene mutations, like BRAF, as well as certain gene fusions, loss of heterozygosity, and more.
Out of the 190 samples, the Afirma GSC correctly called 41 out of 45 suspicious samples and 99 out of 145 benign samples for a sensitivity of 91 percent and specificity of 68 percent. The negative predictive value was 96 percent while positive predictive value was 47 percent.
The test "identified even more patients with benign nodules who could avoid surgery," Ladenson said. Avoiding surgery can be important for a number of reasons, he added. Although thyroid surgery is not nearly as invasive as something like heart surgery, it is still a major operation. The thyroid gland is near important structures that can be injured during surgery. In addition, patients can "end up with a hoarse, trouble breathing," or have a range of other problems, he said.
Ladenson said that for patients who are classified as having benign nodules and for whom the nodules are not causing discomfort or other problems, physicians typically recommend follow-up imaging in one to three years to make sure the nodule has not grown in size or does not have any other suspicious features.
Earlier this week, other research groups presented early clinical experience using the GSC at a conference. Researchers from Ohio State University noted that out of 47 indeterminate cases, the GSC was able to classify 72 percent as benign. Separately, researchers from Memorial Center for Integrated Endocrine Surgery reported that the GSC identified 51 percent more nodules as benign compared to the GEC.
Peter Angelos, chief of endocrine surgery and associate director of the MacLean Center for Clinical and Medical Ethics at the University of Chicago, who was not affiliated with the study, said that the study demonstrated a "significant improvement in the test." Compared with the previous version of the test, "it maintained high sensitivity while significantly improving the negative predictive value."
Angelos also wrote an accompanying commentary on the study published in the same issue of JAMA Surgery.
However, he cautioned that such testing should not be seen as a replacement for other clinical measurements. For instance, if a nodule is causing discomfort to a patient, an operation to remove it may be the best option regardless of whether it is benign, so genomic testing may be unnecessary.
In addition, he said, physicians will have to be mindful of their specific patient population and understand that the results found in this study may vary with different populations. As such, he said, both the negative and positive predictive values will vary depending on the prevalence of thyroid cancer within a specific population. "So, it's not the sort of thing where a clinician can order the test and get results that are going to be exactly what these investigators found," he said.
Nonetheless, "I certainly think this is a step forward," he added. "But my concern is that people will reflexively order the test without going through the thought process" of whether the results will change the management. "This is an issue with every screening test. We have to be sure that if we're ordering costly tests that the information we're getting from them are going to be valuable for our patients."