Analytical Performance of Afirma GSC: A Genomic Sequencing Classifier for Cytology-Indeterminate Thyroid Nodule FNA Specimens


Interview with Zhanzhi (Mike) Hu, Associate Director of Product Development, Veracyte, Inc.

Zhanzhi (Mike) Hu is associate director of product development at Veracyte. He has led the development and launch of multiple molecular diagnostics products, including the Afirma® Genomic Sequencing Classifier (GSC). At the 2017 American Thyroid Association Annual Meeting, Mike presented results from a recent study evaluating the analytical performance of this next-generation classifier.

Q: Can you describe the Afirma GSC? 

M. Hu: The Afirma GSC is the only test currently available for thyroid nodule diagnosis that combines cutting-edge, RNA next-generation sequencing and big data-based machine learning to reveal subtle but robust genomic signals that are otherwise missed by traditional methods such as DNA-based sequencing. For optimal power, the classifier model is built on the RNA expression profiles of more than 1,000 genes.

Q: What was the purpose of the study you presented at the ATA meeting this year? 

M. Hu: This was a comprehensive evaluation designed to provide analytical verification for the Afirma GSC. Analytical verification is a critical assessment for molecular diagnostic products. It ensures that any diagnostic test we launch is analytically robust and will provide physicians with results that are reliable.


Q: How would you summarize the key findings?  

M. Hu: In terms of the test’s performance, we confirmed that the Afirma GSC is highly reproducible when tested by running the same samples with different operators, reagents and instruments, etc. We confirmed that the test is highly sensitive analytically with a limit of detection of 5 percent, i.e. a malignant sample diluted with up to 95 percent of normal adjacent tissue or benign tissue is still expected to be called positive by the Afirma GSC classifier. This helps minimize false negative results by mis-sampling of normal adjacent tissue. The Afirma GSC test is also effective against commonly seen assay interferents present in FNA specimens, such as blood and genomic DNA.

Q: Why was it important to conduct this research? What specifically do the findings tell physicians who might use the Afirma GSC?

M. Hu: This study demonstrates that the Afirma GSC provides sound and reproducible results that physicians can utilize to make confident treatment decisions.


Q: Based on this research, what advantage does the Afirma GSC offer over other methods for classifying thyroid nodules? 

M. Hu: Cancer is highly complicated. Data have shown that relying on single genes to diagnose malignancy is either not sensitive enough clinically, not specific enough clinically, or both. These findings contribute to the body of evidence demonstrating that the combination of RNA next-generation sequencing and machine learning enable the Afirma GSC to better classify thyroid nodules that are indeterminate by cytopathology, and thereby help more patients avoid unnecessary diagnostic surgeries.

Read the Abstract Here