Improving Diagnostic Accuracy
Genomic Classifier Helps Reduce Ambiguity in Lung Cancer Diagnosis
By Lindy Keane Carter
Every year, approximately 250,000 bronchoscopies are performed to rule out cancer when suspicious lung nodules or lesions are found on computed tomography (CT) scans. But bronchoscopy has limited sensitivity (the ability to detect those patients who are truly positive for cancer) and therefore can be inconclusive, leaving the physician with the dilemma of how best to advise the patient. Should the physician recommend surgery or other invasive diagnostic procedures or wait and monitor with periodic CT scans, accepting the risk that the patient may have cancer?
In the July 16, 2015 issue of the New England Journal of Medicine, Gerard A. Silvestri, M.D., MS, Hillenbrand Professor of Thoracic Oncology in the Division of Pulmonary, Critical Care and Sleep Medicine at MUSC Health, and colleagues reported the results of two studies validating a novel diagnostic test, a bronchial genomic classifier that measures the expression of 23 genes associated with lung cancer. This gene array is detected in cells from the proximal airway but can indicate the presence of malignancy or disease processes in distant sites in the lung. The greater the number of oncogene “signatures” in the airway, the higher the likelihood that the lesion in the lung is malignant. This puts more diagnostic information in the hands of the physician trying to advise a patient of the best next step. If the patient is at low risk of malignancy, he or she can be monitored with CT scans instead of undergoing invasive diagnostic procedures that can be risky.
“We have seen promising results in the ability of this classifier to aid in predicting the absence of lung cancer when evaluating new lung masses, which makes this a potentially valuable test for patients and physicians,” says Silvestri.
The two independent, prospective, multicenter, observational studies were Airway Epithelial Gene Expression in the Diagnosis of Lung Cancer (AEGIS) trials (AEGIS-1 and AEGIS-2) (NCT01309087 and NCT00746759, respectively). The studies enrolled 639 former and current smokers undergoing bronchoscopy for suspected lung cancer at 28 sites in the U.S., Canada, and Ireland. Cytology brushes collected epithelial cells from the mainstem bronchus. Patients were followed 12 months after the procedure or until diagnosis. A diagnosis of lung cancer was established either at the time of bronchoscopy or later by means of biopsy with the use of a transthoracic needle, a surgical biopsy, a second bronchoscopic examination, or other invasive procedure. The combination of the classifier and bronchoscopy increased the sensitivity to 96 percent (95 percent confidence interval [CI], 93 percent-98 percent) and 98 percent (95 percent CI, 96 percent-99 percent) in AEGIS-1 and AEGIS-2, respectively, as compared with 74 percent and 76 percent for bronchoscopy alone (P less than 001).
Silvestri says that there were several important findings in the study. “First, bronchoscopy is not as good as we thought it was at establishing a diagnosis of these lesions. Second, in situations where the bronchoscopy is non-diagnostic and the classifier is negative, you can really change how you practice. If I can shunt even a third of my patients away from having an invasive procedure when they are highly likely to have benign disease, then I certainly want to do that,” he says.
The Percepta Bronchial Genomic Classifier (Veracyte, San Francisco, CA) will be available in 2016 at approximately 30 U.S. academic and community medical centers where a registry trial will be conducted to measure the degree to which physicians will avoid invasive procedures when the test indicates the lesion is benign.