• Pulmonary Clinical Trials
• Cancer
• Pulmonary Genomics
• Respiratory Cancers

• Pulmonary Clinical Trials
• Pulmonary & Critical Care
• Cancer
• Pulmonary Genomics
• Respiratory Cancers

The addition of gene expression analysis of 17 cancer genes increases sensitivity of bronchoscopy for the diagnosis of lung cancer, regardless of lesion size, location, and histology. Duncan Whitney, PhD, Allegro Diagnostics Corporation, Maynard, Maryland, USA, presented data from the Airway Epithelium Gene Expression in the Diagnosis of Lung Cancer trial [AEGIS; NCT01309087; Whitney D et al. Am J Respir Care Med 2014].

Bronchoscopy is the standard method of lung cancer diagnosis. Although it is a safe procedure associated with few complications, it is unable to rule out cancer because of a relatively high false-negative rate [Ost DE et al. Chest 2011; Ernst A et al. Chest 2010; Wilson DO et al. Am J Respir Crit Care Med 2008]. Gene expression profiles are able to detect damage to lung epithelium from smoking and may be a technique to differentiate cancer from benign disease [Beane J et al. Genome Biol 2007; Spira A et al. Nat Med 2007]. The purpose of this study was to determine if a gene expression–based classifier would increase the sensitivity of the bronchoscopic diagnosis of lung cancer.

In AEGIS, a prospective multicenter cohort trial, mainstem bronchus epithelial cells were collected from patients scheduled for bronchoscopy and then followed for 12 months. All patients were suspected to have lung cancer, were current or former smokers, and did not have any other malignancies or history of lung cancer.

In the study, 597 samples were evaluated, with half assigned to the training set and half to the validation set. The training set was used for gene selection and model fitting. A total of 232 differentially expressed genes were identified, and the selected cancer genes were from residuals of clinical covariates, such as age, smoking, pack-years, and sex. The final algorithm included 17 cancer prediction genes. The independent validation set was used to validate the gene expression biomarkers identified in the training set.

Seventy-four percent of subjects were diagnosed with lung cancer by pathology as the gold standard. The remainder was followed for 12 months to confirm cancer-free status if no malignancy was initially diagnosed. The addition of genomics to bronchoscopy (BronchoGen; Allegro) increased the sensitivity from 74% to 96% but decreased the specificity from 100% to 47%, which is not expected to lead to a high false-positive rate when the genomic test is used to rule out cancer. The increased sensitivity of BronchoGen was not affected by lesion size, location, stage, or histology. The diagnostic yield was most improved in small peripheral lesions. As a result, the combination technique resulted in a negative likelihood ratio of 0.077, compared with 0.260 from bronchoscopy alone, thereby improving the ability to rule out lung cancer and avoid unnecessary invasive follow-up procedures.

In the AEGIS study, 43% of bronchoscopies were negative for cancer, and 34% of surgeries were performed in patients with benign lesions. Therefore, 42% of patients could avoid invasive procedures if BronchoGen were used instead of bronchoscopy alone to diagnose lung cancer.

In conclusion, Dr. Whitney stated that, in his opinion, the results of the AEGIS study suggest that adding a genomic classifier to bronchoscopy increases clinical sensitivity of bronchoscopy, which is independent of lesion size, location, and histology. This increased sensitivity was at the expense of decreased specificity and a higher false-positive rate. BronchoGen is currently under development to be released as a Clinical Laboratory Improvement Amendments laboratory service.

• © 2014 MD Conference Express®