br Results In early stage NSCLC
Results: In early-stage NSCLC, the average Imipenem allele frequency (MAF) in pre-op plasma ctDNA was lower than that in tissue DNA (tDNA). The concordant gene variations between pre-op ctDNA and tDNA were diﬃcult to detect. However, we found the tissue- pre-op plasma concordant ctDNA mutation detection ratio in lung squamous cell carcinoma (LUSC) was much higher than that in lung adenocarcinoma (LUAD). We also estab-lished a LUSC-LUAD classification model by a least absolute shrinkage and selection operator (LASSO) based approach to help separate LUAD from LUSC based on ctDNA profiling. This model included 14 gene mutations and extracted an accuracy of 89.2% in the training set and 91.5% in the testing set. Correlation analysis showed tDNA-ctDNA concordant ratio was related to histologic subtype, gene mutations and tumor size in early-stage NSCLC.
Conclusion: This study suggests histology subtype and gene mutations could aﬀect ctDNA detection in early-stage NSCLC. NGS-based ctDNA profile has the potential utility in LUSC-LUAD classification.
Corresponding author at: Department of Anesthesiology, Tianjin Medical University Cancer Institute and Hospital, Huan-Hu-Xi Road, Ti-Yuan-Bei, He Xi District, Tianjin, 300060, China.
Corresponding author at: BGI-Shenzhen, Beishan Industrial Zone, Yantian District, Shenzhen, 518083, China.
Corresponding author at: Department of Lung Cancer, Tianjin Lung Cancer Center, Tianjin Medical University Cancer Institute and Hospital, Huan-Hu-Xi Road, Ti-Yuan-Bei, He Xi District, Tianjin 300060, China.
E-mail addresses: [email protected] (C. Wang), [email protected] (M. Ye), [email protected] (W. Liu).
1 These authors contributed equally to this study.
Lung cancer is the leading cause of cancer-related deaths worldwide . LUAD and LUSC are the most common histologic subtypes of NSCLC. Several genetic alterations drive tumor growth in adenocarci-noma, such as EGFR mutations, ALK and ROS1 translocations are found in adenocarcinoma histology and these can be successfully targeted with small molecule tyrosine kinase inhibitors [2–5]. In East Asians close to 50% of patients with adenocarcinoma harbor EGFR mutations, and most of these mutations are found in never-smokers. In contrast, squamous cell lung cancer still represents a tumor subtype that is most common in heavy smokers and no targetable driver mutations have so far been identified in this tumor type. Because of the association with smoking, squamous cell histology has a higher number of mutations than adenocarcinomas [6,7]. Genomic characterization of lung cancer based on tumor tissue of patients with resected disease has been com-prehensively studied, but limited analysis of blood ctDNA in early NSCLC is available.
Circulating tumor DNA (ctDNA) is composed of small fragments of nucleic acid that are released from apoptotic or necrotic tumor cells and circulate in blood [8,9]. Detection of ctDNA has several advantages over conventional tumor biopsies. Sampling ctDNA from blood is noninvasive and can be repeated over time, which allows potentially for early diagnosis of cancer, identifying minimal residual disease or re-currence, real-time monitoring of responses to treatment, and pre-dicting the prognosis [10–17]. Considering tumor heterogeneity, ana-lysis of ctDNA could theoretically also provide a more comprehensive and representative information of multiple tumor deposits. ctDNA–-based liquid biopsy is a promising surrogate and potential supplement to tumor biopsies.
Most studies of ctDNA in NSCLC focused on late-stage patients, where the tumor burden is high. Limited data is available on the eva-luation of ctDNA detection in early-stage NSCLC. In this study, we in-vestigated the ctDNA mutation profile using targeted NGS in matched tumor and plasma samples from 48 NSCLC patients who underwent surgery for stage I-IIIA at our institution. We investigated the con-cordance between tissue and plasma mutations in LUSC and LUAD and we analyzed the correlation between ctDNA gene mutations and clinical features. Furthermore, a LUSC-LUAD classification model was built based on gene mutations detected in ctDNA.