Conference Coverage

Gene Expression Classifier Cuts Some Thyroid Surgery

View on the News

Test Will Change Management Landscape

We’re now in an exciting era of molecular diagnostic testing to address the vexing problem of the cytologically indeterminate thyroid nodule, a finding in up to one-third of thyroid nodule patients that in the majority necessitates an operation, which in the majority of patients is unnecessary, in retrospect. The application of both mutational testing and this novel approach of gene expression profiling promises to change the landscape for patients with cytologically indeterminate thyroid nodules.

The name of the game in thyroid nodule assessment is having a sufficiently high negative predictive value from your tests and evaluation to tell the patients they don’t need an operation. A cytologically benign thyroid nodule meets that criterion with a 97% predictive value. In contrast, the negative predictive value of an indeterminate biopsy is only 65% or so. Most patients and most doctors in the United States are not sufficiently reassured by that finding to recommend against surgery.

The gene expression classifier test bridges that gap in cytologically indeterminate thyroid nodules, providing a negative predictive value of 95%, akin to that which currently satisfies us with cytologically benign thyroid nodules.

This test clearly is appropriate only for cytologically indeterminate thyroid nodules, not those already known to be definitively benign or malignant. Within the cytologically indeterminate group, there are two categories for which this test appears to have a sufficiently high negative predictive value: the atypical category and the follicular neoplasm categories. But it’s not good enough for the third category: cytologically suspicious lesions. A cytologically indeterminate thyroid nodule already gives us maybe a 70% negative predictive value. To go to an 85% negative predictive value for suspicious lesions isn’t sufficient.

I see patients with thyroid nodules every single week. The gene expression profile test is something that we currently actively employ in our management of patients with cytologically indeterminate nodules.

We have undertaken formal cost-utility analyses investigating the potential impact of a novel diagnostic such as the one described today. Even with negative predictive values in a range lower than those achieved by this test in the subsets of indeterminate nodules, this is a remarkably cost-saving test. Novel diagnostic innovations that either improve or maintain a comparable quality of life and save money simultaneously are as rare as hens’ teeth in contemporary medicine, so it’s very exciting to see a test that, from a health economics perspective, will be practical to implement.

Dr. Paul W. Ladenson is director of endocrinology and metabolism at Johns Hopkins University, Baltimore. He has been a consultant to Veracyte.


 

FROM THE ANNUAL MEETING OF THE ENDOCRINE SOCIETY

HOUSTON – A novel gene expression test could eliminate the need for a third of operations done for cytologically indeterminate thyroid nodules by winnowing out low-risk nodules, a large, prospective, double-blind, multicenter study suggests.

The study included fine-needle aspirates of thyroid nodules 1 cm or larger that were classified as indeterminate on cytology and that had corresponding histopathological specimens from excised lesions. A test that measures the expression of 167 genes was applied to 265 cytologically indeterminate nodules, with the patients, physicians, and pathology laboratories blinded to the results of the Afirma gene expression classifier test during management.

In general, cytologically indeterminate thyroid aspirates get classified one of three ways. The negative predictive value of results from the Afirma gene expression classifier test was 95% for the subset cytologically classified as "atypia (or follicular lesion) of undetermined clinical significance," 94% for "follicular neoplasm or lesion suspicious for follicular neoplasm," and 85% for "suspicious cytologic findings," Dr. Erik K. Alexander and his associates reported in a press briefing at the annual meeting of the Endocrine Society.

The Afirma test correctly identified as "suspicious" 78 of the 85 nodules that ultimately were found to be malignant, for a sensitivity of 92% and a specificity of 52%, he said.

A closer look at the seven aspirates that had false negative results from Afirma testing found that six samples lacked a sufficient quantity of thyroid follicular cells to be considered a sufficient sampling of the nodule, said Dr. Alexander of Brigham and Women’s Hospital and Harvard University, Boston.

Dr. Erik K. Alexander

The results are good enough that clinicians could take a conservative approach to managing most patients whose thyroid nodules are cytologically indeterminate on fine-needle aspiration but benign on the gene expression classifier test, he said.

The New England Journal of Medicine published the study online on June 25 (doi: 10.1056/NEJMoa1203208).

The new gene expression test could eliminate the need for 25,000 of the 75,000 surgeries each year in patients with cytologically indeterminate thyroid aspirates, but at the risk of likely malignancy in 5%-10% of nodules classified as benign, particularly nodules that are cytologically indeterminate but suggestive of cancer, Dr. J. Larry Jameson said in an editorial published simultaneously online by the journal (doi: 10.1056/NEJMe1205893).

In this high-risk group, repeating the fine-needle aspiration biopsy or performing a diagnostic hemithyroidectomy might be reasonable even when the gene expression classifier suggests a benign nodule, said Dr. Jameson of the University of Pennsylvania, Philadelphia.

For patients with indeterminate aspirates and negative Afirma test results who are monitored rather than sent to surgery, clinicians should have a low threshold for repeating fine-needle aspiration if ultrasound tests show rapid nodule growth or characteristics suggestive of cancer, he suggested.

Dr. Jameson called the new gene expression classifier test a "welcome addition" to tools for managing thyroid nodules that could substantially reduce costs by avoiding operations, even when considering the added cost of the test.

One of the strengths of the 49-site study was that 80% of participants came from community-based sites and 20% from academic medical sites, representing a real-world sample, coinvestigator Dr. Bryan R. Haugen said at the press briefing.

He described recent use of the Afirma test at his institution on aspirates from 126 patients with indeterminate cytology results from January 2011 to April 2012. The gene expression classifier results said 56% were benign, 41% were suspicious, and 3% could not produce a result. Seventy patients avoided surgery. Among patients who underwent surgery, 52% of nodes were malignant, said Dr. Haugen, head of endocrinology, metabolism, and diabetes at the University of Colorado, Denver.

Dr. Haugen and Dr. Alexander emphasized that the Afirma test should only be used for aspirates with indeterminate cytology, not for biopsies classified as cytologically benign, because of the 70% specificity when used on benign lesions.

Thyroid nodules are common and can be found in 25%-50% of adults. When fine-needle aspiration biopsy produces indeterminate cytological results, most patients have undergone hemithyroidectomy or total thyroidectomy for diagnosis and treatment.

Veracyte, which makes the gene expression classifier, funded the study and tested the samples in its laboratory. Veracyte employees helped design and supervise the study, conducted the statistical analysis, and made up 6 of the 18 investigators. Dr. Alexander disclosed financial associations with Veracyte, Asuragen, Genzyme, and the Boston Clinical Research Institute. His coinvestigators disclosed financial associations with Veracyte and multiple other companies. Dr. Jameson disclosed financial associations with Quest Diagnostics, Novartis, and Ferring, and said that he knows some of the investigators professionally.

Next Article:

   Comments ()