Clinical Validation of the Afirma Genomic Sequencing Classifier for Medullary Thyroid Cancer


Interview with Gregory Randolph, M.D., Harvard Medical School, Boston

Dr. Gregory Randolph is Professor of Otolaryngology Head and Neck Surgery and the Clair and John Bertucci Endowed Chair in Thyroid Surgical Oncology at Harvard Medical School. He spoke with us about findings from a new study demonstrating the Afirma Genomic Sequencing Classifier’s (GSC) accuracy in identifying medullary thyroid cancer (MTC) among thyroid nodules deemed indeterminate by cytopathology. He presented the findings in an oral presentation at the 2017 Annual Meeting of the American Thyroid Association.

Q: Why is diagnosing medullary thyroid cancer prior to surgery so difficult – and why is it so important? 

Dr. Randolph: MTC cells have a cellular morphology that can mimic several thyroid conditions. In addition, MTC is relatively rare – accounting for about 5 percent of thyroid cancers, depending on what study you look at. These factors combine to make MTC difficult to interpret with cytopathology.

In medicine generally, a fundamental goal is that the most optimal treatment always follows diagnosis. This is particularly true in the case of MTC. Knowing that a patient has MTC before surgery will enable the surgeon to plan the surgery appropriately. This includes ordering the necessary imaging studies to inform whether the cancer has spread regionally to cervical lymph nodes. It also enables the surgeon to anticipate potential complications such as hypertension that can accompany MTC in certain cases from undiagnosed adrenal tumors which may co-exist.

Q: What was the purpose of the Afirma GSC study you are presenting at the ATA meeting?

Dr. Randolph: The Afirma GSC uses RNA sequencing and advanced machine learning methods to provide more refined distinctions in challenging thyroid cancer subtypes among FNA biopsies that were cytopathologically indeterminate. In our case, we wanted to see how well the next-generation Afirma classifier identified MTC.

Q: How did you do this and what did you find? 

Dr. Randolph: The Afirma GSC’s MTC classifier was developed with a training set of 483 thyroid FNA samples that had been collected from multiple centers and were used to teach the algorithm what MTC looks like compared to non-MTC. The final MTC classifier includes 108 genes.

We then tested the final MTC classifier blindly on an independent test set of 211 thyroid FNA samples, which included 21 MTC cases and 190 non-MTC cases. We found that it correctly identified all 21 MTC cases (i.e., 100 percent sensitivity) and was accurate every time it identified a sample as not MTC (i.e., 100 percent specificity).

Q: What do these findings mean for patient care? 

Dr. Randolph: Our findings suggest that the Afirma GSC’s MTC classifier can give physicians the information they need to provide better care for their patients. Specifically, patients with MTC can get the appropriate and optimally safe surgery to help ensure optimal outcomes – without having to consider a second surgery to finish the job.

Read the Abstract Here