Poster T144 - LUNG CANCER DETECTION VIA WHOLE-TRANSCRIPTOME RNA SEQUENCING OF NASAL EPITHELIUMView Session Add to Schedule
Daniel Pankratz, Lori Lofaro, Giulia Kennedy, Jing Huang, Avrum Spira, Janna Chamberlin, Patric Walsh, Jonathan Wilde, Jie Ding, Saeed Saberi, Joshua Babiarz, Sangeeta Bhorade, Christopher Stevenson, Marc Lenburg, Duncan Whitney
We have previously established the clinical utility of a molecular test (Percepta) that utilizes a brushing of normal-appearing bronchial epithelium to improve lung cancer detection in former and current smokers. This 'field of injury' principle has been extended to the nose with evidence that genomic changes associated with cancer can be detected in the nasal epithelium (Perez-Rogers et al. 2017). We report the feasibility of a molecular test developed from non-invasive nasal brush samples for lung cancer detection in current and former smokers.
Brushings of the nasal epithelium were prospectively collected during the AEGIS clinical trials from current and former smokers undergoing a bronchoscopy for suspected lung cancer lesions found on chest CT. Patients were followed for up to one year until a final diagnosis of lung cancer or benign disease was made. Purified RNA was analyzed using whole-transcriptome RNA sequencing. A total of 675 patients were divided into a training set of 411 nasal samples and an independent test set of 264 nasal samples. Classifiers were developed using gene expression as well as clinical factors such as age, gender, smoking status and nodule characteristics.
The independent test set was analyzed using the developed classifiers and results compared to the final clinical diagnosis (malignant or benign). A nasal genomic-clinical classifier demonstrated sensitivity of 97% (95% CI: 93-99) and a specificity of 46% (95% CI: 32-59). While the prevalence of lung cancer in the overall test set is 78%, reflecting the high-risk clinical setting of this bronchoscopy-based cohort, it also includes a subset of patients with low/intermediate pre-test risk of cancer (n=57). In this subset, sensitivity is 95% (95% CI: 74-100) and specificity is 45% (95% CI: 29-62), similar to that of the overall test set.
A nasal genomic-clinical classifier developed using whole transcriptome RNA sequencing shows early feasibility in detecting lung-cancer related gene expression changes in nasal brushings from high-risk smokers. Further research is warranted to understand its impact on patient care in a real-world clinical setting.
A nasal genomic-clinical classifier has the potential to serve as a non-invasive tool for lung cancer risk-stratification among patients with pulmonary nodules.