Skip to main content
Premium Trial:

Request an Annual Quote

Prospective Validation Confirms Sensitivity of Veracyte's Gene Expression Test for Thyroid Nodule Risk


Results from a multi-center prospective validation study have confirmed the sensitivity and high negative predictive value of a gene expression test marketed by Veracyte to identify patients whose thyroid nodules are likely to be benign and who can thus avoid further diagnostic surgery or complete thyroid removal with lifelong hormone replacement therapy.

Veracyte offers its 167-gene Afirma test as a tool to reclassify as either benign or suspicious thyroid nodules that fine needle aspiration analysis has yielded indeterminate results. The company's prospective validation study, published this week in the New England Journal of Medicine, confirmed that the test is able to pick out, with up to a 95 percent negative predictive value, low-risk nodules from the 15 percent to 30 percent that FNA cytology finds inconclusive.

Giulia Kennedy, Veracyte's chief scientific officer, told PGx Reporter this week that the company's goal was to make this validation as "real-world an experience as possible." To end up with a sufficient number of indeterminate nodules, the researchers had to start with cytological assessment of more than 4,900 individual FNA specimens from 49 sites representing both academic and community settings.

She said the study was the "largest conducted in the US for the diagnosis of thyroid nodules."

According to Kennedy, the results — showing over 90 percent sensitivity and negative predictive values ranging from 85 percent to 95 percent depending on the type of indeterminate cytological finding — confirmed what Veracyte expected and has found previously through the development and testing of the classifier.

The test is designed to focus on sensitivity over specificity in picking out malignant nodules, to maximize the certainty that samples deemed benign really are low-risk, with the caveat that many benign samples will still be classified suspicious. The classifier showed only 52 percent specificity in the validation study, marking suspicious about half of the nodules that the reference histopathological review deemed benign.

The Afirma test is run in Veracyte's CLIA-approved laboratory in San Francisco as part of a diagnostic service combining initial FNA cytology assessment through an outside lab with assessment by the expression classifier for any cytologically indeterminate results.

As reported by PGx Reporter sister publication GenomeWeb Daily News, Medicare contractor Palmetto GBA recently agreed this January to cover the Afirma classifier for use in thyroid nodule diagnosis.

Kennedy said the company has already assessed more than 10,000 FNA samples through requests from community and academic endocrinologists, of which a smaller portion have required further molecular classification using Afirma. "We don't run our test on cytologically benign or malignant samples because cytology already does an excellent job on either of those categories," she said. "Our test comes into play in that 15 percent to 30 percent of the time that it is ambiguous."

Erik Alexander, a physician at Brigham and Women's Hospital and the lead author of the NEJM validation study, told PGx Reporter that he views the test as a "clear step" toward personalizing and improving care for thyroid patients. He added that he's enthusiastic about the large validation study as an opportunity to better define how the service could be used in clinical practice.

"Nobody wants surgery they don't need," he said. "This allows us to substantially modify our understanding of cancer risk in many of these patients and that then [opens] a larger opportunity for modifying care recommendations."

In an editorial in the same NEJM issue as the validation report, Larry Jameson of the University of Pennsylvania called the test a "welcome addition" to more personalized medical care and wrote that if results of the gene-expression classifier were used to inform clinical decision making, it could be possible to reduce surgery for indeterminate nodules by "at least one third, or about 25,000 operations per year.

In the validation, Alexander and the other researchers tested the Afirma classifier on 265 indeterminate nodules from an initial group of 3,759 patients from 49 clinical sites.

Of the 265 indeterminate specimens, 85 were malignant according to blinded central histopathological review used as a reference standard. The Afirma classifier identified 78 of these as suspicious, demonstrating 92 percent sensitivity, the group found.

Negative predictive values differed between the three main types of indeterminate nodules. For samples classified "atypia" by FNA cytology, the gene expression algorithm showed 95 percent negative predictive value. For "follicular neoplasm" or suspicions of such, the classifier had a 94 percent negative predictive value. For samples with "suspicious cytologic findings," the predictor dropped to an 84 percent negative predictive value.

According to Alexander, the Afirma test showed itself to be as accurate at picking out benign specimens from FNA cytology indeterminates as FNA cytology is in identifying benign nodules.

In either case, Alexander and Kennedy said, a "benign" finding is not seen as a zero-risk measure. Patients found benign either by FNA cytology initially, or through the Afirma classifier, are followed closely using ultrasound imaging.

"I think the key is to look back and compare with what a benign cytology means. What that really implies is a really low risk of cancer, not no risk of cancer … A negative gene classifier result has a comparable [false negative] rate as would a benign cytology, so people should feel comfortable in recommending a more conservative approach," Alexander said.

"The large majority of indeterminate patients will be headed for surgery anyway, so while it is understandable to be thinking about malignancy, the real question in this situation is who is definitively benign," Kennedy added. "The [Afirma] test avoids surgeries in individuals who have a benign thyroid nodule. It doesn’t mean they can stop seeing a doctor. They still need to be followed just as a cytology benign patient would be followed with repeat ultrasounds."

Larry Jameson suggested in his editorial, however, that those patients who are on the more suggestive side of cytologically indeterminate results might benefit from repeat FNA biopsy or diagnostic surgery even if they receive a "benign" molecular test result. He also stressed the importance of a "low threshold to repeat fine-needle aspiration" for those patients who avoid surgery if ultrasonographic findings indicate rapid growth or other malignant characteristics.

Alexander said that every year in the US up to half a million patients have a new diagnosis of a thyroid nodule, with a "low but real" risk of cancer in each case. Among patients with indeterminate cytology, only 30 percent will have cancer, he said, while 70 percent will not. Without a tool to further classify ambiguous cytology, he said this means too many patients are receiving unnecessary surgery with associated risk.

According to Kennedy, while Afirma, as a lab-developed test, does not currently fall under the regulatory purview of the FDA, Veracyte has conducted its development and validation keeping the possibility in mind that it may have to file for regulatory approval in the future.

Veracyte is also planning the development of a classifier for difficult-to-diagnose lung diseases and lung cancer, according to the company's website.