Genomic sequencing classifier may improve management of indeterminate pulmonary nodules
NEW ORLEANS — A genomic sequencing classifier can help identify the probability of malignancy for patients with lung nodules and improve the sensitivity of bronchoscopy overall, according to a study presented at the CHEST Annual Meeting.
“Management of lung nodules usually starts by estimating the probability that the nodule is cancerous. Many nodules are in the in-between range, which is when other tests are often performed,” Peter J. Mazzone, MD, MPH, FCCP, director of the lung cancer program and director of education for the Respiratory Institute at Cleveland Clinic, said during a presentation. “When a bronchoscopy is recommended, despite fantastic advances in navigation systems to get to those nodules, we often come back without a solid answer and that leaves the clinician in a bit of a predicament. What’s the next step?”
Mazzone and colleagues sought to validate the accuracy of the second-generation Percepta genomic sequencing classifier (Veracyte) in classifying lung nodules. The study included 412 patients from three independent cohorts — the AEGIS I and II cohorts (n = 246) and the Percepta registry (n = 166) — who underwent a standard-of-care bronchoscopy for a lung nodule, had a history of smoking and did not have a history of concurrent or prior cancer. The patient’s treating clinician determined the pretest probability of malignancy as low, intermediate or high.
Overall, the study population included more men than women, more white patients than those of other races and more former smokers than active smokers, according to Mazzone. He added that there was a wide distribution of nodule size, with more peripheral than central nodules.
Clinician-determined pretest probability of malignancy was low in 19% of nodules, intermediate in 46% and high in 35%. Among these, 5% of low-risk nodules, 28% of intermediate-risk nodules and 74% of high-risk nodules were malignant.
After an inconclusive bronchoscopy, the genome sequencing classifier was run. The classifier down-classified patients with low pretest risk with greater than 99% negative predictive value and intermediate pretest risk with a 91% negative predictive value, which would allow a clinician to monitor these patients, according to Mazzone. Further, the classifier up-classified patients with intermediate pretest risk with a 65% positive predictive value and a high pretest risk with a 91% positive predictive value.
“How often do those results potentially impact patient care? If you started at low risk, you got a negative result about 55% of the time. If you started at intermediate risk, 29% of the results were negative, allowing you to shift classes from intermediate to low risk. If you start at intermediate, about 12% were high-risk classifiers, shunting you to the high-risk category. And if you were already at high risk, you went to very high risk about 27% of the time,” Mazzone said.
Additionally, results indicated that the sensitivity of bronchoscopy alone was improved by the inclusion of the classifier (95.5% vs. 41% in low- and intermediate-risk pretest groups) — a finding that held true across the spectrum of nodule sizes, whether the nodule was central or peripheral and regardless of the histology of the cancer, he noted.
“Overall, the Percepta genome sequencing classifier is clinically validated to be able to up- or down-classify the probability of malignancy in a subset of lung nodules after a nondiagnostic bronchoscopy and improve the sensitivity of bronchoscopy overall. It can also serve as a complement to bronchoscopy and could enable improved management in the sticky situation when your bronchoscopy is negative,” Mazzone said. – by Melissa Foster
Mazzone P. Late Breaking Abstracts. Presented at: CHEST Annual Meeting; Oct. 19-23, 2019; New Orleans.
Disclosures: Mazzone reports he is an advisory committee member for Grail and a consultant for SEER, and he has received research support to his institution from Exact Sciences, Oncocyte and Veracyte.