Combining genetic analysis and machine learning can improve the identification value of biopsies used to detect pneumonia that progressively scars lungs — a classic marker of pulmonary fibrosis — without doctors having to resort to surgery, according to a study.
The findings apply to the transbronchial biopsies, or TBBs, that doctors use to diagnose patients with usual interstitial pneumonia, or UIP. They support the use of Veracyte’s genomic test Envisia Genomic Classifier, researchers said. Usual interstitial pneumonia is a form of lung disease characterized by progressive scarring of both lungs.
The study, “Usual Interstitial Pneumonia Can Be Detected in Transbronchial Biopsies Using Machine Learning,” was published in the Annals of the American Thoracic Society.
Introduced in October 2016, the Envisia Genomic Classifier was designed to help doctors differentiate between IPF and other lung diseases, a process that commonly requires surgery. The test combines RNA sequencing analysis to detect UIP in samples collected from bronchoscopy with machine learning computerized analysis. Bronchoscopy involves inserting a tube through the nose or mouth into the windpipe and lungs to collect tissue samples.
Researchers examined 283 TBB samples from 84 patients. Genomic classifier analysis allowed them to do a better job of distinguishing UIP from non-UIP cases. In fact, they were able to identify two-thirds of UIP cases with more accuracy.
“These strong early results” support the development of the Envisia Classifier, “which we believe can help significant numbers of patients with suspected IPF obtain a more timely, accurate, and safer diagnosis,” Bonnie Anderson, chairman and chief executive officer of Veracyte, said in a press release.
To validate the findings, researchers compared the genomic classifier-generated results with those obtained in the traditional manner. That approach combines surgical sample collection with experts’ classifications of pneumonia as UIP or non-UIP.
“These newly published findings also reinforce the scientific and clinical rigor that went into the Envisia Classifier’s development,” Anderson said. “We believe that this foundational work, along with the robust clinical validation data we have shared in recent months, will build the body of published evidence needed to drive physician adoption and payer reimbursement of the Envisia Genomic Classifier test.”
Veracyte made several presentations on the Envisia Genomic Classifier at the American Thoracic Society International Conference in Washington, May 18-23.