Leveraging Enriched Genomic Data to Develop a Next-Generation Afirma Gene Expression Classifier
Interview with Giulia C. Kennedy, Ph.D., chief scientific officer, and Richard T. Kloos, M.D., senior medical director, endocrinology, at Veracyte
Dr. Giulia Kennedy is chief scientific officer at Veracyte where she led development of the Afirma Gene Expression Classifier (GEC). Most recently, she has overseen development of the company’s next-generation Afirma Genomic Sequencing Classifier (GSC), designed to enable even more patients to avoid an unnecessary surgery in thyroid cancer diagnosis. Dr. Richard Kloos is senior medical director, endocrinology, at Veracyte where he draws upon his extensive academic and clinical experience in endocrinology to educate physicians about the clinical use and impact of the company’s genomic technology. Here they discuss the new Afirma GSC and pivotal clinical validation study data, which are being unveiled for the first time at a Product Theater event at the AACE annual meeting.
Q: What is the next-generation Afirma test designed to do?
Dr. Kennedy: The current Afirma GEC has a high sensitivity and reclassifies nearly half of indeterminate thyroid nodules as benign, which enables most of these patients to avoid unnecessary diagnostic surgery. While this is a huge improvement for patients compared to diagnostic surgery, we wanted to take advantage of recent advances in sequencing and machine learning technology to further increase the test’s specificity so that we can save even more thyroids and keep patients out of the operating room.
Q: How is the technology behind the enhanced Afirma GEC different from the current test?
Dr. Kennedy: The current test combines RNA-based gene expression, which is measured on a microarray platform, with machine learning algorithms to recognize benign thyroid nodules among those diagnosed indeterminate by cytology. The rates of transcription from expressed genes are the “features” that feed into the algorithm, which uses a form of pattern recognition to determine if a nodule is benign.
With the enhanced Afirma GEC, we are combining RNA sequencing and newer machine learning techniques to leverage more enriched, previously undetectable genomic information. This includes not only gene expression, but also the presence of DNA variants, fusions, copy number variants and other information that may be predictive of thyroid cancer. This powerful combination of machine learning and RNA sequencing gives us much richer genomic content that enhances the classifier’s ability to distinguish benign from malignant nodules.
Q: What do the data show about the test’s performance?
Dr. Kloos: New data from a pivotal clinical validation study show that the test maintains the current test’s high sensitivity – of 91 percent – while increasing its specificity from 52 percent to 68 percent. This means that the next-generation test is expected to identify 30 percent more benign nodules compared to the current Afirma GEC, with a negative predictive value of 96 percent. These findings suggest that the new test can enable significantly more patients to undergo monitoring in lieu of diagnostic surgery. We’ll be sharing these data during a product theater event at the AACE meeting.
Q: Tell us more about the pivotal clinical validation study.
Dr. Kloos: The new Afirma GSC was validated on a prospective, multicenter, blinded sample set of 191 indeterminate thyroid FNAs. These are from the same sample set that was used to validate the current test – in a study published in The New England Journal of Medicine. In addition, we found that in a consecutive series of 324 indeterminate nodules, nearly 100% of indeterminate nodules that were previously called benign by the current Afirma test were classified as benign by the enhanced classifier.
Q: How is this approach different from other molecular tests for indeterminate thyroid nodules?
Dr. Kennedy: Other tests look at targeted sets of gene mutations in a small number of genes or microRNAs. These approaches have limited clinical utility because numerous studies show that individual gene mutations and microRNA rarely tell the whole story and frequently miss cancers. Our approach to RNA sequencing interrogates the entire genome and takes advantage of newer methods in machine learning to combine valuable features that provide a higher-resolution picture of thyroid nodules. . We are not aware of anyone else who is using machine learning and RNA sequencing in a single commercially available test for any complex diagnostic indication. We are excited about this work because it takes the same machine learning methods that are being used in other fields such as financial modeling, social media and self-driving cars and uses it to take patient care to a new level. We now have the capability to utilize really complex genetic information in ways that have not been previously used in healthcare diagnostics.
Q: Is the new Afirma GSC available yet?
Dr. Kloos: We plan to begin making the test available to physicians during the next few weeks.
Q: What else should physicians know about the next-generation Afirma GSC?
Dr. Kloos: Multiple long-term studies have shown that the current Afirma test is enabling many patients with benign nodules to avoid unnecessary thyroid surgery – and that the test’s results are durable, meaning patients stay out of the operating room over the long term. This is significantly improving patient care and is saving significant healthcare costs. We think that the Afirma GSC is going to go even farther in improving patient outcomes and removing waste from the healthcare system.