乌鸦传媒

SAN JUAN -- Data from real-world experience with a new gene expression test support earlier findings that the screen can pick out which indeterminate thyroid nodules don't need to be removed, researchers reported here.

Among nearly 340 biopsies, the Afirma test by Veracyte found about half (51%) to be benign, rendering surgery unnecessary, , of Brigham & Women's Hospital, and colleagues reported at the American Thyroid Association meeting here.

About half of patients in this benign group had longer-term follow up, and only one of these samples ultimately proved cancerous, Alexander said.

He said the findings are similar to those reported in an earlier validation study published in the New England Journal of Medicine, which found a 95% negative predictive value. Although the current study was not a validation cohort, he said, the real-world experience suggests that figure is reliable.

At least one clinician in the audience, however, took issue with those data. , of Moffitt Cancer Center in Tampa, Fla., said other datasets have not turned up similar results with regard to negative predictive value.

"You market something you claim has a 95% negative predictive value based on a single publication, when numerous other datasets are not finding same values," McIver said. "We have to repeat the study. I can't think of another example in thyroidology where we've changed our practice based on a single study."

Competing Screens

The mainstay of evaluating thyroid nodules is fine-needle aspiration (FNA), but about 15% to 30% of samples evaluated this way turn out to be indeterminate on cytology.

Most patients with indeterminate cytology are referred for diagnostic thyroid surgery -- although the majority ultimately are found to have benign disease.

To diminish the need for diagnostic surgery, researchers have been turning their attention to molecular analysis of these lesions to determine which are more likely to be malignant and which are benign.

In addition to the , there's also the , and both take different approaches to assessing the genetic profiles of thyroid nodules.

The goal of the Afirma test is to identify benign nodules by measuring expression of 167 genes via messenger RNA (although the exact details are unclear because the design remains proprietary).

On the other hand, Asuragen's miRInform test looks for cancerous nodules by focusing on seven known gene mutations plus microRNA expression changes.

, of the University of Michigan, who was not involved in the study, told 乌鸦传媒 that it's "conceivable these tests could provide complementary information" but it's too early to tell since data on the individual tests are just coming in.

Afirma in the Real World

Cautioning that his presentation was not a validation of the original New England Journal of Medicine data, Alexander presented 'real-world' clinical experience with the Afirma test based on data from five centers between September 2010 and January 2013: Brigham & Women's Hospital, University of Colorado Medical Center, University of Pennsylvania Medical Center, Ohio State University Medical Center, and University of Cincinnati Medical Center.

Of a total of 339 Afirma analyses, 51% were determined to be benign and 44% were deemed suspicious. The findings were similar to a study on clinical data reported at the American Association of Clinical Endocrinologists meeting in May.

Specifically, for atypical or follicular lesion of undetermined significance (FLUS), 55% were labeled benign and 40% were suspicious, and for follicular neoplasm, 49% were benign and 45% were suspicious, Alexander reported.

He said these were very similar to the findings from the validation trial -- although McIver said that the benign call rate when he analyzed data from a Mayo Clinic sample was lower, which could be attributable to differences in patient population and patient selection.

"I think this classifier is overtrained on its test set," McIver said. "We need to approach this with caution."

Alexander reported that of the 121 lesions deemed suspicious and for which patients eventually had surgery, 44% were indeed cancer, similar to the 43% seen in validation trial. And of the 174 determined to be benign, follow-up data were available for 71 patients (41%) over a median of 8.5 months.

They found 84% of these patients had stable, non-malignant disease over that time, and 15% (11 patients) ultimately had surgery: 4 per physician recommendation, 4 for compressive symptoms, and 3 for personal choice, usually cosmetic.

Only one of these nodules (1.4%) was found to actually be cancerous, Alexander said, and there was one additional case of interest that had unusual oncologic characteristics.

He did report that there was significant variation in the proportion of nodules determined to be benign or suspicious between practices -- which McIver also acknowledged could be a reason for the different findings in his analyses.

"While long-term assessment is important, these data lend support to the clinical utility of this diagnostic tool in clinical practice," Alexander concluded.

Clinical Implications

Koenig said it's still unclear how best to use the molecular tests that are available for analyzing indeterminate thyroid nodules. The Afirma test seems to be more widely used "but that may relate more to marketing" than utility, he said.

"At this point, nobody knows what the right thing to do is," Koenig told 乌鸦传媒. "It's an area of uncertainty due to the fact that the data are just coming in. I think over the coming years, the accumulation of experience and data will help answer these questions."

Asuragen will be presenting data on a newer version of its miRInform test later in the meeting.

Comment