Concordance of ctDNA and tissue mutations in NSCLC: A meta-analysis

Concordance of ctDNA and tissue mutations in NSCLC

Abstract

Systematic evaluation of the consistency between circulating tumor DNA (ctDNA) in plasma and tumor tissue samples in mutations in non-small cell lung cancer (NSCLC) patients. To collect publicly published literature from numerous important international medical databases through computer retrieval. To compare the differences in literature data related to gene mutations between plasma ctDNA and tumor tissue samples, a meta-analysis was performed using Stata 12.0 software while taking into account the inclusion and exclusion criteria. This article includes a total of 15 research data and collected data reports from 15 groups of NSCLC patients. The results are displayed using tissue samples as the gold standard. The Pearson correlation coefficients of sensitivity and specificity were used to calculate rho=0.044, and P=0.893. The Q-test found poor homogeneity and high heterogeneity in sensitivity and specificity among research data from various literature studies (I²>50%, P<0.1). The area under the SROC curve is 0.97 (95% CI: 0.96~0.99). The small sample subgroup has high heterogeneity, and the combined diagnosis effect size is 26[6~111], lower than the large sample subgroup 185320 [0~2.7×10¹²]. When taking 200 as the cut-off point, the combined effect size of the small sample subgroup is 46 [12~183], still lower than that of the large sample subgroup 429 [52~3574]. From this, it can be concluded that the consistency of small-sample studies is higher than the quality of large-sample studies, and the heterogeneity is relatively low. From the perspective of mutation types, the heterogeneity of literature with EGFR gene mutations alone is higher than that of literature with non-EGFR mutations alone, and the consistency is lower. The consistency of using plasma ctDNA to detect mutations in NSCLC patients with tumor tissue samples is influenced by the type of mutation gene and sample size measured by the patient, and there is a significant bias in related studies.