Beijing Haidian Hospital Beijing, Beijing, China (People's Republic)
Disclosure(s):
Guotian Pei, n/a: No financial relationships to disclose
Purpose: Multiple primary lung cancer (MPLC) is an increasingly well-known clinical phenomenon, but there is still a lack of methods to effectively distinguish between MPLC and intrapulmonary metastasis (IM). This study aims to establish an effective approach for molecular differential diagnosis of MPLC and IM by next-generation sequencing (NGS). Methods: A total of 74 Chinese lung cancer patients, harboring at least two tumors (N = 168), were selected and randomly divided into a training cohort (N = 51) and a test cohort (N = 23) in a 7:3 ratio. The multifocal lung cancers in each patient were diagnosed as MPLC or IM based on American College of Chest Physicians (ACCP) guidelines. A random forest approach was applied to these training data, resulting in a model optimized for MPLC identification using comprehensive genomic profiling analyzed by Acornmed panel with 808 cancer-related genes, and then evaluated in the test cohort. Finally, 38 patients (103 lesions) were recruited as the independent validation cohort to verify the model. Results: We evaluated the performance of the classifier using the test cohort data and observed an 89.5% percent agreement of MPLC as well as a 100% percent agreement of IM. The diagnostic model showed an excellent Area Under the Curve (AUC) of 0.947, a 91.3% overall accuracy and a κ consistency of 0.75 (p＜0.001) compared with the ACCP criteria. Furthermore, compared to our previous molecular identification method, which detected MPLC only by common mutations and driver gene alterations, the assay achieved improvement in AUC (0.947 vs. 0.921), total accuracy (91.3% vs. 87.0%) and percent agreement of MPLC (89.5% vs. 84.2%) in the test set, with comparable percent agreement of IM patients (100% vs. 100%). Similar results were found in independent validation set with a higher AUC (0.938 vs. 0.875) and a better total accurate rate (89.5% vs. 79.0%) than previous mutations-based method. In addition, the detection algorithm resulted in a higher percent agreement of MPLC than our previous molecular criteria (87.5% vs. 75.0%) and an identical IM percent agreement (100% vs. 100%). What’s more important, the MPLC predictive value of the classification achieved 100% in both test set and validation cohort. Conclusion: These data demonstrated that our novel molecular classifier using comprehensive genomic characteristics of MPLC can accurately classify multifocal lung tumors as MPLC or IM, which suggested that broad panel NGS may be a useful tool for assisting with differential diagnoses.
Identify the source of the funding for this research project: None.
