Between May 2018 and June 2020, 80 subjects who were diagnosed with LUAD and 69 control subjects were recruited from The Affiliated Suzhou Hospital of Nanjing Medical University (Suzhou, China). The Ethics Review Board of The Affiliated Suzhou Hospital of Nanjing Medical University approved the study. Informed consent was obtained from all participants. The subjects were treated according to the principles described in the Declaration of Helsinki. Patient clinical parameters, including sex and age, were collected by reviewing the medical records.

For patients with LUAD, blood was drawn after diagnosis and before any treatment. Patients with a history of cancer or severe clinical symptoms and genetic diseases were excluded from the study. The control samples were obtained only from healthy subjects who came to the hospital for a physical examination and patients with benign lung disease, and not from patients with other tumor types. EDTA blood tubes were employed for blood collection. For plasma separation, blood samples were centrifuged at 1,600 × g for 10 min at room temperature, and the plasma samples were stored in a refrigerator at −80°C for later use.

Before EV isolation, plasma samples were centrifuged at 2,000 × g for 10 min at 4°C and 10,000 × g for 30 min at 4°C to separate out the cells and apoptotic bodies. Next, 400 µl of 30% PEG4000 (1 M NaCl) was added and incubated with 800 µl of supernatant for 3 h at 4°C. After centrifugation at 3,000 × g for 15 min at 4°C, the pellet was subjected to further experiments. The Nano platform (IZON Science, Ltd.) was utilized to analyse the size distribution of the pellet.

Western blotting (WB) was employed to detect the signature proteins of EVs, including tumor susceptibility gene 101 protein (TSG101) and cluster of differentiation 9 (CD9), with GAPDH as the reference protein. Exosomes were lysed with RIPA buffer (Enzo Biochem, Inc.), and the protein concentrations were assayed using a BCA protein assay kit (Thermo Fisher Scientific, Inc.). Anti-TSG101 (catalog no. ab133586; 1:5,000), anti-CD9 (catalog no. ab223052; 1:5,000) and anti-GAPDH (catalog no. ab8245; 1:5,000) (all Abcam) were used as the primary antibodies. The detailed WB protocol was in keeping with a previously reported protocol (29). The density of the WB image was analyzed using ImageJ software (National Institutes of Health).

RNA isolation and qPCR were performed in accordance with the studies by Shan et al (30) and Su et al (31). RNAiso Plus (Takara Biotechnology Co., Ltd.) was employed to isolate RNA from EVs and plasma according to the manufacturer's instructions. In detail, 1 ml RNAiso Plus was added to the pellets or 200 µl plasma and mixed with a homogenizer. Next, 200 µl chloroform was added and mixed with a vortex oscillator. The mixture was placed at room temperature for 10 min and centrifuged at 12,000 × g for 15 min at 4°C. The supernatant was transferred to a new 1.5-ml centrifuge tube and mixed with an equal volume of isopropyl alcohol. After overnight incubation at −20°C, the solution was centrifuged at 12,000 × g for 15 min at 4°C. The supernatant was discarded, and 75% ethanol was prepared with DEPC water to wash the RNA. After centrifugation at 12,000 × g for 5 min at 4°C, the remaining liquid was removed with blotting paper, and the inverted position was maintained on new blotting paper for 15–30 min. The RNA was dissolved in 10 µl DEPC water.

A total of 5 control subjects and 5 patients with LUAD were selected to perform a preliminary experiment. A U6 snRNA Normalization RT-PCR Quantitation kit (Shanghai GenePharma Co., Ltd.) was employed to perform reverse transcription of the RNA. The RT-PCR system used a final reaction volume of 20 µl, which contained 4 µl 5X MMLV RT Buffer, 0.75 µl dNTP (10 mM), 1.2 µl miRNA RT primers (1 µM), 0.25 µl RNasin (40 U/µl), 0.2 µl MMLV Reverse Transcriptase, 10 µl RNA sample (1–3 µg) and 3.6 µl RNase-free H₂O. The standard thermocycling conditions were as follows: 25°C for 30 min, 42°C for 30 min, 85°C for 5 min and then storage at 4°C. Next, Cq values were measured via qPCR. Three Hairpin-it™ miRNAs qPCR Quantitation kit reagents boxes (Shanghai GenePharma Co., Ltd.) were employed to perform qPCR. All the primer sequences used in the study are listed in Table I. The qPCR system used a final reaction volume of 20 µl, which contained 10 µl 2X Real-time PCR Master Mix Buffer (FAM), 0.4 µl miRNA specific primer set (10 µM), 0.4 µl miRNA specific probe (10 µM), 0.4 µl ROX reference dye (50X), 0.2 µl Taq DNA polymerase (5 U/µl), 2 µl cDNA and 7.6 µl sterilized H₂O. The standard thermocycling conditions were as follows: 95°C for 3 min, followed by 40 cycles of 95°C for 12 sec and 62°C for 40 sec. The signal was then collected at 62°C. U6 snRNA was used as an internal reference for normalization, relative to which the expression of miRNAs was calculated using the comparison 2^−∆∆Cq method (32). The EV-associated and plasma miR-10b, miR-200a and miR-141 expression levels in the LUAD and control groups were compared. Synthetic miR-10b was diluted to 10⁻³, 10⁻⁴, 10⁻⁵, 10⁻⁶, 10⁻⁷ and 10⁻⁸ µM. RNA (2 µl) from samples and different concentrations of synthetic miR-10b were reverse transcribed into cDNA and then were used to establish standard curve. The concentration of miR-10b in samples was calculated according to the established standard curve.

The AFP, NSE, CEA, CYFRA211, pro-gastrin-releasing-peptide (Pro-GRP), CA125, CA153, CA199 and CA724 were detected by the Department of Laboratory Medicine, The Affiliated Suzhou Hospital of Nanjing Medical University. Blood was drawn after diagnosis and before any treatment. After standing for 20 min, the serum was separated by centrifugation at 1,600 × g for 10 min at room temperature and electrochemiluminescence detection was used. The serum samples were tested on the same day. Results were obtained from the Department of Pathology.

The Cancer Genome Atlas (TCGA) (https://www.cancer.gov/about-nci/organization/ccg/research/structural-genomics/tcga) survival analysis was performed with Kaplan-Meier Plotter (https://kmplot.com/analysis/) (33). The expression data of the target miRNA in the GSE114711 dataset (34) were obtained from the Gene Expression Omnibus (GEO) database (https://www.ncbi.nlm.nih.gov/geo/) using R v3.5.3 (https://cran.r-project.org/bin/windows/base/old/3.5.3/). Target gene prediction was performed using TarBase (35), TargetScan (36) and microT-CDS (37). The function of target genes was analyzed via Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.kegg.jp/kegg/kegg1.html), Gene Ontology (GO; http://david.ncifcrf.gov/) annotation and protein-protein interaction (https://string-db.org/) analyses. The expression of genes was analyzed using UALCAN (38). The networks of target genes were constructed with Cytoscape v3.6.0 (http://chianti.ucsd.edu/cytoscape-3.6.0/) and GeneMANIA (39).

Data are presented as the mean ± standard deviation. Experiments were repeated in triplicate. Data were analyzed using SPSS v.19 (IBM Corp.). A non-parametric Mann-Whitney U test was employed to compare miRNA expression levels between the control and LUAD groups. A Wilcoxon signed-rank test was employed to compare miRNA expression levels between EVs and plasma. A non-parametric Kruskal-Wallis test was employed to compare miRNA expression levels at different stages. Bonferroni's correction was used for the correction of multiple comparisons. P<0.05 was considered to indicate a statistically significant difference. Pearson's correlation analysis was employed to analyze the correlation between miRNA expression levels and age. A receiver operating characteristic (ROC) curve and the area under the ROC curve (AUC) were established to assess the potential of miR-10b to distinguish LUAD patients from individuals in the control group. Kaplan-Meier plotter (https://kmplot.com/analysis/) was used to analyze the survival curve of miRNAs (33). The OS from 0 to 120 months of patients with LUAD and control subjects was calculated from data recorded in the TCGA database. The distributions of OS time were analyzed using the Kaplan-Meier method and compared using a two-sided log-rank test.

In total, 149 patients were enrolled in this study, including 69 control subjects and 80 patients with LUAD (Table II). The age of the patients ranged from 21 to 82 years old. There were 66 female patients and 83 male patients. Of the 80 patients with LUAD, 64 had early stage disease (stage I + II). The number of patients with lesions located in the right or left lobe was similar. A total of 70 patients had only 1 lesion, 2 patients had 2 lesions and only 1 patient had 3 lesions.

The procedure for isolating EVs is shown in Fig. 1A. The size distribution analysis indicated that the pellets contained vesicles ranging in size from 80 to 280 nm (Fig. 1B). Signature proteins, including TSG101 and CD9, and the reference GAPDH, were detected by WB analysis (Fig. 1C). The results of the density analysis are shown in Fig. S1. The WB results indicated that the pellets obtained by PEG precipitation did contain EVs, and there was no significant difference in relative protein expression between the LUAD group and control group.

Obtained from the GEO database, the GSE114711 dataset contains small-RNA sequencing data for plasma-derived EVs from a cohort of patients with NSCLC at different stages (34). These patients included 7 control smokers, 11 patients with early stage NSCLC and 8 patients with late-stage NSCLC. Among the 3,762 detected miRNAs, there were no miRNAs that with significantly increased expression. However, 7 miRNAs in early stage disease and 33 miRNAs in late-stage disease exhibited significantly decreased expression compared with the control group, and 16 miRNAs showed significantly decreased expression in the late stages compared with the early stages. Using the expression in the control group as a reference, the expression of these miRNAs is shown in Fig. 2A. As the 80 outpatients with LUAD were primarily early stage lung cancer patients, a further comparison was made of the expression levels of the 7 miRNAs in the early stage and control groups (Table SI). The results showed that miR-10b, miR-141 and miR-200a were the 3 miRNAs that exhibited the most change in expression. The expression levels of miR-10b, miR-141 and miR-200a were then measured in 5 control subjects and 5 patients with LUAD (Fig. 2B-D). The non-parametric Mann-Whitney U test showed that the expression levels of plasma or EV miR-10b and miR-141 in the LUAD group were significantly higher than those in the control group (P<0.05), while there was no significant difference in miR-200a expression level. The Wilcoxon signed-rank test showed that there was no significant difference between the plasma and EV miR-10b and miR-141 expression levels in the control or LUAD groups (P>0.05). For miR-200a, there was no significant difference between plasma and EV in the control group, but there was a significant difference in the LUAD group (P=0.043). After Bonferroni's correction, there was no significant difference in miR-200a level in the LUAD group (P>0.05). Kaplan-Meier plotter was used to analyze the survival curve of miR-10b and miR-141 (Fig. 2E and F). miR-10b had a hazard ratio (HR) of 1.39 (P=0.039), while miR-141 had an HR of 1.32 (P=0.082). Finally, miR-10b was selected for further research.

Synthetic miR-10b was diluted for standard curve establishment (Fig. S2). The expression level of miR-10b in EVs and plasma was significantly increased in the LUAD group compared with that in the control group (Fig. 3A and B). The expression levels of miR-10b at different stages were also compared and the non-parametric Kruskal-Wallis test results showed that the expression levels of miR-10b did not exhibit a significant difference between stages I, II and III in EVs or plasma (Fig. 3C and D). The expression levels of miR-10b were also analyzed in patients with different sexes and ages. The results showed that the expression levels of miR-10b in EVs and plasma did not show a significant difference between males and females (Fig. S3A and B). The correlation between miR-10b and age was also analysed, and the results showed that there was no linear correlation. The correlation coefficients were 0.084 and 0.073, respectively (Fig. S3C and D).

A total of 5 volunteers from the 69 patients of the control group and all LUAD subjects underwent the TM tests, which included AFP, NSE, CEA, CYFRA211, Pro-GRP, CA125, CA153, CA199 and CA724 assays (Fig. 4). The threshold of these markers is marked by dotted grey lines in the figures. Among these markers, CYFRA211 (22/80) and NSE (20/80) had the highest detection rates in the LUAD group (Fig. 4C and I). CA125, CA724 and NSE were determined to be abnormal in the control groups (Fig. 4D, G and I). The normal range of each marker is shown in Table III.

ROC curve analysis was employed to compare the diagnostic efficiencies of miR-10b and TMs. miR-10b in EVs had a significantly higher AUC (AUC=0.998) than that in plasma (AUC=0.693) (P<0.01; Fig. 5A). The Youden index was employed to determine the cut-off value. The sensitivity and specificity of EV-associated miR-10b (sensitivity, 98.75%; specificity, 98.55%) and plasma miR-10b (sensitivity, 37.5%; specificity, 94.2%) reached a maximum when the cut-off values were 9.087×10⁻⁷ and 7.389×10⁻⁷ µM, respectively. According to the cut-off value, the new AUCs of EV and plasma miR-10b were 0.986 and 0.658, respectively (Fig. 5B). The ROC curves of TMs are shown in Fig. 5C. Among these markers, CA153 (AUC=0.844), CA199 (AUC=0.7275) and GRP (AUC=0.655) had the top 3 AUCs. After threshold modification, CYFRA211 had the highest AUC of 0.635 (Fig. 5D).

The target genes of miR-10b were predicted using TargetScan, microT-CDS and TarBase. There were 14 interaction target genes (Fig. 6A). These genes formed a network that included coexpression, colocalization and genetic interactions (Fig. 6B). Next, KEGG pathway enrichment analysis was performed and it was found that the ‘circadian rhythm’ pathway and the ‘longevity regulating pathway’ were the most enriched pathways (Fig. 6C). GO analysis was also performed. ‘Gland development’, ‘thymus development’ and ‘positive regulation of protein complex assembly’ were the top 3 biological processes (Fig. 6D). The ‘RNA polymerase II transcription factor complex’ and the ‘nuclear transcription factor complex’ were the top 2 cellular components (Fig. 6E). ‘Transcriptional activator activity, RNA polymerase II transcription regulatory region sequence-specific DNA binding’ and ‘transcription coactivator binding’ were the most enriched molecular functions (Fig. 6F).