Identification of clinical features to predict the consistency of mutational profiles obtained from plasma and tissue of advanced non-small cell lung cancer (NSCLC) patients.

Abstract

e21537 Background: Due to its non-invasiveness and advantage in overcoming tumor heterogeneity, plasma-based genomic profiling is widely used in advanced NSCLC patients for treatment guidance. However, whether the mutational profile derived from ctDNA comprehensively represents that in tumor tissue is the major concern for the application. It has been reported that a variety of factors influence the consistency of mutational profile between plasma and tissue samples. This study explores clinical features which predict the concordance between tissue and plasma samples. Methods: Paired tissue and plasma samples were collected from 79 (stage III and IV) NSCLC patients. Samples were profiled using a 520-cancer-related gene panel. Serum markers such as CEA, CA199, SCC and NSE were collected. Results: All tissue samples except for 1 (98.7%) had mutations detected from this panel. 68 plasma samples (86%) had mutations detected from this panel. The overall concordance of mutation between plasma and tissue samples was 43%; and the concordances for point mutation and copy number variations were 60.2% and 15%, respectively. The concordance for 8 classic lung cancer driver genes and druggable mutations were 77.2% and 34.2%, respectively. Next, we explore the correlation of serum biomarkers and the concordance rate. Univariate analyses revealed that CA19-9 (P < 0.001) and CEA (P = 0.01) were significantly associated with the concordance rate of actionable mutations. Using CEA > 5ng/ml as a cutoff, 43.1% patients having higher CEA level harbored an identical mutation profile between plasma and tissue vs only 20.8% patients with lower CEA level had an identical mutation profile (p = 0.01). Furthermore, 75% patients with CA 19-9 > 39U/ml harbored an identical mutation profile between plasma and tissue vs 22.8% patients with CA19-9 £ 39U/ml (P < 0.001). Combining both CEA and CA19-9 can effectively predict the concordance (p < 0.001); 85.7% of patients with CEA and CA19-9 levels above the cutoff have identical mutation profile and only 34.2% patients with either one or both below the cutoff have identical mutation profile (p < 0.001). Conclusions: We identified the clinical and molecular factors which showed promising value in predicting the consistency of mutational profile between plasma and tissue in NSCLC patients. Plasma samples of patients with high CEA ( > 5ng/ml) and CA19-91 ( > 39U/ml) can comprehensively reflect the mutation profile of tumor samples. Tissue biopsy is not necessary for such patients.