SAN DIEGO—The utility of the Afirma Gene Expression Classifier may help determine if some thyroid cancers are malignant, but researchers here at ENDO 2015, the Annual Meeting of the Endocrine Society, suggested that in most cases other means of diagnosis may be sufficient to identify dangerous lesions.
“These lesions—i.e., atypia of undetermined significance, AUS—are the ones that trouble us,” Jessica Lee Betancourt, MD, Research Fellow in Endocrinology at Cleveland Clinic Florida, said in an interview at her poster presentation at the meeting. The AUS are the lesions that pathologists are not sure about—i.e., maybe they are malignant or maybe they are benign, so further steps are needed to make that determination.
One new method has been the Afirma Gene Expression Classifier, a test that can narrow the chances of determining if a nodule is malignant among those lesions that make a pathologist unsure of the how to call the specimen, she explained. However, the researchers at Cleveland Clinic Florida thought that other methods—such as repeat fine needle aspiration biopsies and non-invasive ultrasound imaging—might prove acceptable as well.
In the study, the team reviewed the records of 119 patients who presented with thyroid nodules larger than one centimeter in size—“If you have a nodule that is greater than one centimeter, we do a fine needle biopsy of that nodule,” she said. “Our goal was to assess the clinical validity and utility of gene expression classifiers in the evaluation of AUS cytology. We also aimed to evaluate the performance of ultrasound for predicting malignancy in the setting of AUS cytology.”
She said the researchers found that when they used a combination of a second fine needle biopsy—as recommended by the Bethesda guidelines for work-up of indeterminate thyroid nodule cytology—and ultrasound, the results in a selected group of patients were equivalent to those of the Gene Expression Classifier.
The key elements of the ultrasound scan are finding reduced echogenicity—or hypoechoic response—and the composition of the lesions as seen on the imaging—that is, whether the nodules are solid or non-solid. “We found that in combination, these features may direct attention to suspicious nodules in need of further investigation,” she said.
“We classified nodules as being benign based on Gene Expression Classifier testing, fine needle aspiration cytology, or histopathology results. Nodules were classified as malignant based on final histopathology. Ultrasound characteristics, fine needle aspiration, Gene Expression Classifier results, final recommendations by our endocrinologist, and subsequent clinical and surgical follow-up were collected.”
About five percent of patients had a diagnosis of AUS after biopsy. In the study, of the 119 patients diagnosed with AUS, 48 of the specimens were sent for gene-expression classification; and of that group, 27 were classified as suspicious, and surgery was performed on 21 of these patients. Once the specimen was excised and studied in the laboratory, it was determined that 14 of the patients had thyroid cancer after histopathology.
The other 71 cases underwent a second fine needle aspiration biopsy, and this time the pathologists said that 19 were benign. The other 52 were still defined as AUS. Gene-expression classification labeled 38 of these AUS cases as suspicious, and 35 of these nodules were excised—and 32 (91%) were confirmed as malignant. The second biopsy was performed three to five months after the first biopsy, she said.
The research team also reviewed the ultrasound scans on 25 lesions from the group of nodules that had been biopsied a second time, specifically noting that these lesions had solid or hypoechoic features—and 23 of those lesions were confirmed as malignant on histopathology. Two of the lesions were found to be benign. So in that cohort, ultrasound was able to pinpoint malignancy with 92 percent accuracy, she said.
On the other hand, nine nodules that were neither solid nor hypoechoic on ultrasound were confirmed as malignant compared with 15 found to be benign.
“In our practice the performance of the Gene Expression Classifier after the first biopsy showing AUS was associated with a low level of diagnostic accuracy—a 66.6 percent likelihood of malignancy,” Betancourt said. “The diagnostic accuracy of Gene Expression Classifier testing was markedly higher, at 91.4 percent, when was done after two consecutive AUS cytologies.
She also pointed out that nodules with two separate AUS cytologies that are both solid and hypoechoic by ultrasound have a high likelihood for malignancy, with a 92 percent positive predictive value: “The positive predictive value of these two ultrasound features is equivalent to the performance of a suspicious gene expression classifier finding in this setting. We think that these patients can be sent to surgery.”
However, malignancy cannot be excluded in those modules with two separate AUS cytologies that are neither solid nor hypoechoic, she cautioned—and for that group of patients, performing the Gene Expression Classifier test could prove beneficial and be very helpful in that setting, she said.
“We recommend sending patients to surgery rather than molecular testing if they have an AUS diagnosis after a second fine needle biopsy and worrisome features on ultrasound. Prospective studies with long-term follow-up are required to evaluate the efficacy of such a protocol for detection and appropriate management of thyroid malignancy.”
Concerns from Dennis Kraus
Asked for his opinion, though, Dennis Kraus, MD, Director of the Center for Head & Neck Oncology and Co-director of the Center for Thyroid & Parathyroid Surgery at Lenox Hill Hospital in New York, said that the study performed at the Cleveland Clinic might not be generalizable to other practices: “I was surprised at the very low rate of just five percent AUS in this study,” he said. “Our rate is probably around 10 to 20 percent.”
He said that gene-expression classifier tests after the first AUS diagnosis are routinely used in his practice, and that any new test takes a while to be accepted, suggesting that continuing to rely on second biopsies and ultrasound evaluation might be considered “anachronistic.” He noted that by performing the test after the first AUS diagnosis, a second biopsy is avoided.
He suggested that the test would prove cost-effective, and added that when a fine needle biopsy is performed at his practice he collects tissue for the biopsy and for a genetic test in case the pathologist calls the tissue AUS.
Also asked for her perspective, Maria S. Guoth, MD, Assistant Clinical Professor of Medicine (Endocrinology), at Yale University School of Medicine, said: “These atypical of undetermined significance nodules are the types of lesions that most doctors would send to surgery. But thyroid surgery is surgery, and these authors have found that if you can use the gene classifier, there is a 92 percent positive predictive value of finding a malignancy. Previously all these AUS nodules were directly going for surgery.
“This is very helpful because many nodules are not cancerous—only about one-third are malignant. For myself, I do the fine needle aspiration biopsy and then put the Afirma test aside, unless the pathology comes back AUS.”