The average mutation frequency in pre
The average mutation frequency in pre-op ctDNA was lower than that in tDNA, and there were less gene mutations in pre-op ctDNA than in tDNA. Additionally, there were discordant tDNA -pre-op ctDNA gene mutations in 33 sample pairs. This may be caused by lung cancer heterogeneity and the sensitivity of the technology [, , ]. It was reported the mutant allele fractions of ctDNA were higher in metastatic cancer than in earlier stage cancer . Consisting with the observation, we found the mean MAF of ctDNA was higher in patients with stage III than in patients with stage I and II in NSCLC. It was reported EGFR mutation frequency range was 20–76% in LUAD patients in the Asia-Pacific region . EGFR mutation frequency was 69.2% (18/26) in our LUAD tissue samples, higher than the reported 47%, but was within the range. And our sample size (n = 26, LUAD) was so small that could not represent the EGFR mutation rate in whole population. However, the detection of EGFR mutations in pre-op plasma ctDNA was quite low in our cohort. We verified our NGS data by testing EGFR, PIK3CA and TP53 hotspot mutations using ddPCR method in 23 samples. The ddPCR showed good concordant results with NGS. Most of the MAFs detected by the ddPCR approach were comparable to those detected by NGS. The EGFR mutations not detected in ctDNA by NGS were also not detected by ddPCR, suggesting it was not as relative low sensitivity of NGS assay. It could be attributed to low ctDNA input because total cfDNA were extracted from 2 ml plasma. ctDNA often accounts for a small fraction of (<1%) total cfDNA in early stage lung cancer . And, we find most researches use small panel including dozes of ATPγS tetralithium salt to detect plasma EGFR mutations. While we used 546-genes panel in the study. We think it may not appropriate to use big panel to detect driver gene mutations from plasma. Notably, we found the tissue-pre-op plasma concordant ctDNA mutation detection ratio in LUSC was much higher than that in LUAD. Extensive intratumor heterogeneity, the same tumor containing many different subpopulations of cells, is presented in both LUSC and LUAD. Single tumor region analysis cannot cover all gene mutations. But LUSC was reported to carry more clonal mutations which were thought to be present in all cancer cells than did LUAD . So, single-region test from LUSC could have less effect on the tDNA-ctDNA concordant mutation detection ratio than that from LUAD. Additionally, tumor shed DNA fragments containing tumor-specific mutations into the circulation. The release of ctDNA from apoptotic or necrotic tumor cells is related to tumor location, size and vascularity. As we know, tumor location of LUSC is different from LUAD. The majority of LUSC is central type and most LUAD is peripheral type. And, abundant necrotic cells are often observed in LUSC not in LUAD . We indeed observed a lot of necrotic cells in LUSC with high tDNA-pre-op ctDNA concordant mutation detection ratio in our cohort (Supplementary Figure 5). This result is consistent with the most recent report that ctDNA detection is associated with histological subtype . These findings suggest not only the difference of ctDNA release between LUSC and LUAD, but also a potential superior application of ctDNA as a feasible biomarker in LUSC. Pearson correlation analysis also demonstrated that tissue-pre-op plasma concordant ctDNA mutation detection ratio had a relatively high correlation with tumor histology type. More concordant ctDNA mutations were found in LUSC, confirming the report that non-adenocarcinoma histology was an independent predictor of ctDNA detection in early-stage NSCLC . Consistent with most studies [37,38], the concordant ctDNA mutation detection ratio correlated with tumor size. Moreover, we found that the tDNA-ctDNA concordant ratio was related to TP53, ROS1, PIK3CA, KMT2D, EPHB1, and CDKN2A gene mutations, suggesting the gene mutations could represent clonal somatic mutations of tumor evolution . Also, it is possible that the passive release of ctDNA into the bloodstream more often involves these gene mutations.