Whole Blood-Derived circUSP10 Acts as a Diagnostic Biomarker in Patients With Early-Stage Non-Small-Cell Lung Cancer

Abstract

Accumulating evidence has indicated that differentially expressed noncoding circular RNAs (circRNAs) play essential roles in the occurrence and development of various types of cancer. Here, we aimed to identify and explore the diagnostic value of hsa_circ_0003026 (named circUSP10) in patients with early non-small-cell lung cancer (NSCLC). The differentially expressed circRNAs were screened from the microarray-based assay of human NSCLC tissues and their corresponding noncancerous tissues, and the candidate circRNAs were further verified in patients with NSCLC using reverse transcription-quantitative polymerase chain reaction (RT-qPCR). Circulating circUSP10 was isolated from whole blood of healthy people and patients with NSCLC and was detected by RT-qPCR. In addition, the diagnostic value of circUSP10 in early NSCLC was evaluated by receiver operating characteristic (ROC) curve analysis. We found that circUSP10 was upregulated in tumor tissues from patients with early NSCLC and associated with tumor size and tumor-node-metastasis (TNM) stage. Importantly, circUSP10 was obviously upregulated in the whole blood of patients with NSCLC. Additionally, whole blood–derived circUSP10 showed good diagnostic performance for screening early NSCLC and was relatively stable in blood under adverse conditions. These findings demonstrate that circUSP10 may act as a novel biomarker for the diagnosis of early-stage NSCLC, suggesting the potential of circUSP10 in RNA-based therapy for cancer.

Introduction

Lung cancer is one of the most frequently diagnosed cancers and the leading cause of cancer-related death worldwide ¹ . Histologically, approximately 85% of lung cancers are non-small-cell lung cancer (NSCLC). The main subtypes of NSCLC are lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and large cell carcinoma (LCC) ² . Despite significant progresses have been made in the multiple-discipline team (MDT)-based treatment of NSCLC in recent decades, 5-year survival of patient with NSCLC is only approximately 20%^1,3. Therefore, there is an urgent need to identify novel biomarkers for early diagnosis of NSCLC.

Circular RNAs (circRNAs) can be divided into noncoding circRNAs and coding circRNAs^4,5. The noncoding circRNAs are generally produced by backsplicing events. Previously thought to result from abnormal splicing events, these circRNA species were considered “junk” and had no functions in the past few decades ⁶ . However, their covalently closed ring structure can make them more stable than their linear RNAs and prevent these molecules from being digested by RNase R. CircRNAs are also found to be abundant, conserved and involved in the regulation of gene expression ⁷ . Studies suggest biological functions of circRNAs acting as microRNA sponges, transcriptional regulators, and RNA-binding proteins (RBPs). Recently, growing evidence indicates that circRNAs are closely linked to the onset and development of cancers, including lung cancer ⁸ . For example, circSATB2 could promote the progression of NSCLC through regulation of miR-326/actin-bundling protein 1 (FSCN1) axis ⁹ . Additionally, circRNA 100146 played a positive role in NSCLC by binding miR-361-3p and miR-615-5p to regulate multiple targeted mRNAs ¹⁰ . These studies showed that circRNA may act as a potential biomarker for early diagnosis of NSCLC.

In the present study, we aimed to screen and validate some circRNAs as novel biomarkers for the diagnosis of early NSCLC. By analyzing the microarray data (GSE158695), hsa_circ_0003026 (named circUSP10) was identified as a candidate circRNA. The expression of circUSP10 was detected in NSCLC tissues and whole blood. To explore whether circulating circUSP10 could be a novel biomarker for early diagnosis of NSCLC, we analyzed the diagnostic value of circUSP10 using receiver operating characteristic (ROC) curves. In general, these results indicated that circUSP10 might act as a novel biomarker for early diagnosis of NSCLC.

Materials and Methods

Screening of Candidate CircRNAs

CircRNA expression profile of human NSCLC was performed by Agilent-069978 Arraystar Human CircRNA microarray V1 and GSE158695 was chosen from the Gene Expression Omnibus (GEO) database. Comparison between human NSCLC tissues and corresponding noncancerous tissues was reanalyzed normalized microarray data using the GEO2R tool.

Patients and Whole Blood Samples

The tumor-node-metastasis (TNM) stage of all NSCLC specimens was classified according to the Union for International Cancer Control/American Joint Committee on Cancer (UICC/AJCC) Version 8 lung cancer standard system. All patients who participated in this study signed written informed consent. The protocol was in compliance with the Declaration of Helsinki and approved by the Clinical Research Ethics Committee of Ningbo University Health Science Center (approval No. NBU-2018-002). All tissue samples were collected from the Affiliated First Hospital of Ningbo University (Ningbo, China) and the Affiliated Lihuili Hospital of Ningbo University (Ningbo, China). The whole blood samples were obtained from the Affiliated First Hospital of Ningbo University (Ningbo, China) and the Affiliated People’s Hospital of Ningbo University (Ningbo, China) from 2019 to 2021. None of these patients with NSCLC had received chemotherapy, radiotherapy, targeted therapy, or immunotherapy. All 109 pairs of NSCLC and corresponding noncancerous tissues were obtained by surgery and stored within the RNAstore regent (CWBIO, Beijing, China) at −80°C until RNA isolation. Peripheral whole blood from 87 patients with lung cancer and 38 healthy controls were collected by using ethylenediaminetetraacetic acid (EDTA) tubes. Whole blood samples were stored at −80°C until RNA isolation.

Cell Culture

The human normal lung epithelial cell line (BEAS-2B) and four lung cancer cell lines (SPC-A-1, LTEP-a-2, NCI-H1299, and A549) used in this study were purchased from Chinese Academy of Sciences Cell Bank of Type Culture Collection (Shanghai, China). BEAS-2B cells were cultured in Dulbecco's modified eagle medium (DMEM) medium (HyClone, CT, USA) and SPC-A-1, LTEP-a-2, NCI-H1299, and A549 cells were cultured in RPMI-1640 medium (HyClone) supplemented with 10% fetal bovine serum (FBS) (PAN-Biotech, Aidenbach, Germany). All cell lines were placed in humidified air at 37°C with 5% CO₂.

RNA Isolation

Total RNAs were isolated from tissues and cell lines using TRIzol reagent (Invitrogen, Carlsbad, CA, USA) and extracted from whole blood using TRIpure LS Reagent (Invitrogen) according to the manufacturer’s protocol. The quality and concentration of total RNAs were measured by DeNovix DS-11 Spectrophotometer (DeNovix, DE, USA). Finally, total RNAs were stored at −80°C until reverse transcription.

RT-qPCR Analysis

First strand cDNA was synthesized from 1 µg of total RNA using NovoScript® Plus All-in-one 1st Strand cDNA Synthesis SuperMix (gDNA Purge) (Novoprotein, China) following the manufacturer’s instructions. Then the cDNA was further used for real-time quantitative polymerase chain reaction (qPCR) by NovoStart SYBR qPCR SuperMix Plus (Low ROX Premixed) (Novoprotein), and qPCR reaction was operated on an Mx3005P real-time PCR System (Stratagene, CA, USA). GAPDH was used as a normalization control of the expression of circUSP10. The relative expression level of circUSP10 was calculated by the 2^−∆CT method. Primers were shown in Supplementary Table 1.

RNase R and Actinomycin D Assays

To determine the circular structure features of circUSP10, total RNAs from SPC-A-1 or LTEP-a-2 cells (2 μg) were incubated at 37°C for 10 min with or without 2 U/μg of RNase R (Geneseed, China) treatment. To assess the half-life of circUSP10 and its linear RNA USP10, 2 μg/ml actinomycin D (APExBIO, USA) was added into the culture medium in SPC-A-1 or LTEP-a-2 cells. After treatment with RNase R or actinomycin D, RT-qPCR was applied to detect the expression of circUSP10 and USP10.

Detection of Common Tumor Markers

Three milliliters of fasting venous blood was collected and serum was isolated, and the serum was free of hemolysis and lipidemia. The contents of serum cytokeratin 19 fragment antigen 21–1 (CYFR21-1), neuron-specific enolase (NSE), and carcinoembryonic antigen (CEA) were measured by a fully automatic chemiluminescence immunoassay analyzer (Beckman, USA). The threshold values were set 3.3 ng/mL for CYFRA21-1, 17.0 ng/mL for NSE, and 5.0 ng/mL for CEA.

Statistical Analysis

All data were analyzed using GraphPad Prism 8 (GraphPad Software, USA). Student’s t-test was used to compare the means between two groups. Chi-squared test was performed to analyze the relationship between the clinical characteristics of NSCLC patients and the expression of circUSP10. Logistic regression analysis was applied to conduct a diagnostic panel consisting of circUSP10 for NSCLC. ROC and area under curve (AUC) analysis were adapted to evaluate the diagnostic accuracy of circUSP10. The cut-off value of the circUSP10 was calculated by the Youden index. All analyses used two-sided tests, and P values less than 0.05 were considered to indicate statistical significance.

Results

CircUSP10 Was Preliminarily Identified as a Candidate Biomarker

To obtain the differentially expressed circRNAs (DEcircRNAs), we sought for the circRNA microarray datasets with the expression levels of NSCLC tissues and the corresponding noncancerous tissues from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). Subsequently, the microarray data (GSE158695) was chosen to screen the DEcircRNAs using the GEO2R tool ¹¹ . According to the screening conditions (P < 0.05 and fold change >1.0), 109 DEcircRNAs were obtained, including 76 upregulated circRNAs and 33 downregulated circRNAs (Fig. 1A). Among these DEcircRNAs, 3 unreported circRNAs (hsa_circ_0003026 [ubiquitin-specific peptidase 10 (USP10]-derived circRNA, named circUSP10), hsa_circ_0001947 (ALF transcription elongation factor 2 [AFF2]-derived circRNA, named circAFF2), and hsa_circ_0063526 (Ran GTPase activating protein 1 [RANGAP1]-derived circRNA, named circRANGAP1)) were selected for further validation as potential biomarkers (Fig. 1B). To initially verify the expression of the candidate circRNAs, we detected their expression levels in 20 pairs of NSCLC tissues and their corresponding noncancerous tissues by RT-qPCR. The results showed that circUSP10 was significantly upregulated (P = 0.0195, Fig. 1C). However, circAFF2 (P = 0.4083, Fig. 1D) and circRANGAP1 (P = 0.8513, Fig. 1E) did not show statistically significant difference in the paired tissue samples. Therefore, circUSP10 was preliminarily identified as a candidate biomarker.

Figure 1.

CircUSP10 was preliminarily identified as a candidate circRNA. (A) Volcano plot showed the differential expression of circRNAs between human NSCLC tissues and the corresponding noncancerous tissues (GSE158695). According to fold change ≥1.0 (log₂ scaled) and P < 0.05 (log-₁₀ scaled), the red and green points represented the upregulated and downregulated circRNAs, respectively. (B) The information of three circRNAs (hsa_circ_0003026, hsa_circ_0001947 and hsa_circ_0063526) in the GSE158695 dataset. The relative expression of circUSP10 (C), circAFF2 (D), and circRANGAP1 (E) in the NSCLC tissues and the matched adjacent normal tissues (n = 20). Data represent the mean ± SD. NSCLC: non-small-cell lung cancer. *P < 0.05.

CircUSP10 Was Upregulated in Early NSCLC and Lung Cancer Cell Lines

To detect the expression of circUSP10 in a large number of tissue samples, 109 pairs of NSCLC tissues and matched adjacent normal tissues were used for RT-qPCR analysis. As expected, the results showed that the relative expression of circUSP10 was significantly upregulated in NSCLC tissues (P = 0.0035, Fig. 2A). Additionally, the fold change of circUSP10 was −0.85777 ~ 2.07121 (Fig. 2B) in 109 pairs of tissues. To further examine the expression level of circUSP10 in early NSCLC, we analyzed the matched clinicopathological information of patients with NSCLC and performed TNM staging according to the UICC/AJCC TNM staging standard. As a result, we found that circUSP10 was upregulated in tumor tissues of NSCLC patients with TNM stage I compared to that in adjacent normal tissues (P < 0.0100, Fig. 2C). However, there was no significant difference of circUSP10 in tumor tissues of NSCLC patients with TNM stages II–IV compared to the adjacent normal tissues (P = 0.2190, Supplementary Fig. 1). Consistently, circUSP10 was highly expressed in four lung cancer cell lines (SPC-A-1, LTEP-a-2, NCI-H1299, and A549) compared to that in human normal lung epithelial cells (BEAS-2B) (Fig. 2D). Together, the expression of circUSP10 was significantly elevated in NSCLC and even in early NSCLC.

Figure 2.

CircUSP10 was upregulated in early NSCLC tissues and lung cancer cell lines. (A) The relative expression of circUSP10 in NSCLC tissues and the matched adjacent normal tissues (n = 109). (B) The fold-change of circUSP10 expression in NSCLC tissues compared to adjacent normal tissues (n = 109). (C) The relative expression of circUSP10 in NSCLC patients with TNM stage I and the matched adjacent normal tissues (n = 56). (D) The relative expression of circUSP10 in human lung epithelial cell (BEAS-2B) and four lung cancer cell lines (SPC-A-1, LTEP-a-2, NCI-H1299, and A549). Data represent the mean ± SD. NSCLC: non-small-cell lung cancer. **P < 0.01; ****P < 0.0001.

CircUSP10 Was a Stable Circular RNA Molecule

To confirm the structural features of circUSP10, we detected the closed-loop and stability of circRNA. The genomic locus of circUSP10 was shown in Fig. 3A. The mature spliced circUSP10 was 1102 nt in length and was derived from exons 3 and 4 of USP10. After RNase R digestion, the relative expression of circUSP10 had no significant change whereas the relative expression levels of its cognate gene USP10 mRNA and the reference gene GAPDH mRNA were significantly reduced (Fig. 3B, C), suggesting the closed-loop structure of circUSP10. After actinomycin D treatment, it was found that circUSP10 was able to resist actinomycin D-induced suppression of mRNA, whereas the relative expression of USP10 mRNA were significantly repressed (Fig. 3D, E), indicating a longer half-life of circUSP10 compared to its cognate leaner USP10 mRNA. These results demonstrated that circUSP10 was a circular RNA molecule with high stability.

Figure 3.

CircUSP10 was a stable circular RNA molecule. (A) CircUSP10 is derived from USP10 exons 3 and 4. The relative expression of circUSP10, USP10 mRNA and GAPDH mRNA was detected by RT-qPCR in SPC-A-1 (B) and LTEP-a-2 (V) cells treated with or without RNase R. The relative expression of circUSP10 and USP10 mRNA was detected by RT-qPCR in SPC-A-1 (D) and LTEP-a-2 (E) cells treated with actinomycin D at the indicated time points. RT-qPCR: reverse transcription-quantitative polymerase chain reaction. **P < 0.01; ***P < 0.001; ****P < 0.0001.

The Upregulated CircUSP10 Was Associated With Tumor Size and TNM Stage

To investigate the clinical significance of circUSP10 expression in NSCLC, the clinicopathological features of 109 cases were collected and analyzed. As summarized in Table 1, total cases of NSCLC were divided into low expression group (n = 54) and high expression group (n = 55) by the median expression of circUSP10. Notably, the results showed that circUSP10 expression was closely correlated to tumor size (P = 0.0013) and TNM stage (P = 0.0097). However, the expression of circUSP10 was found to have no obvious correlation with the patients’ sex (P = 0.506), age (P = 0.327), smoking history (P = 0.134), pathological classification (P = 0.562), differential degree (P = 0.937), T stage (P = 0.735), N stage (P = 0.720), or M stage (P = 0.775). These findings indicated that the upregulated circUSP10 was involved in the tumor growth of NSCLC.

Table 1.

Correlation Between CircUSP10 Expression and Clinical Characteristics of 109 Patients With NSCLC.

Characteristics	Case no.	CircUSP10 relative level		χ2 test	P value
		Low	High
Total cases	109	54	55
Gender
Male	66	31	35	0.443	0.506
Female	43	23	20
Age (years)
≤60	33	14	19	0.960	0.327
>60	76	40	36
Smoking history
Yes	27	10	17	2.245	0.134
No	82	44	38
Histology
Adenocarcinoma	73	35	38	1.151	0.562
Squamous carcinoma	27	13	14
Other type	9	6	3
Differentiation
High and moderate	65	32	33	0.006	0.937
Poor	44	22	22
Tumor size (cm)
≤3	67	25	42	10.401	0.0013**
>3	42	29	13
TNM stage
I	56	21	35	6.680	0.0097**
II–IV	53	33	20
T stage
I	68	31	37	0.115	0.735
II–IV	31	13	18
N stage
N0	77	39	38	0.129	0.720
N1-3	32	15	17
M stage
M0	98	49	49	0.082	0.775
M1	11	5	6

NSCLC: non-small-cell lung cancer; TNM: tumor-node-metastasis.

^**P < 0.01.

CircUSP10 May Serve as a Diagnostic Biomarker for Early NSCLC Patients

To assess the diagnostic value of circUSP10 in NSCLC patients, ROC curves were developed. The data showed that the AUC analysis of circUSP10 in all 109 cases with NSCLC was 0.6033, 95% confidence interval (CI) = 0.5280–0.6783, P = 0.0085, sensitivity = 47.71%, and specificity = 100.00%, (Fig. 4A). To further evaluate the diagnostic value of circUSP10 in early NSCLC, we analyzed the AUC of circUSP10 in 56 pairs of patient tissues with TNM stage I. As shown in Fig. 4B, the AUC of circUSP10 in NSCLC patients with stage I was 0.6087 (95% CI = 0.4728–0.6844, P = 0.0472, sensitivity = 46.43%, and specificity = 75.00%). Taken together, the results indicated that circUSP10 could be used as biomarker in early diagnosis for NSCLC patients.

Figure 4.

Diagnostic value of circUSP10 in early NSCLC patients. (A) ROC curve analysis of circUSP10 in NSCLC tissues and the matched adjacent normal tissues (n = 109). (B) ROC curve analysis of circUSP10 in NSCLC patients with TNM stage I and the matched adjacent normal tissues (n = 56). NSCLC: non-small-cell lung cancer; ROC: receiver operating characteristic; TNM: tumor-node-metastasis; AUC: area under curve.

CircUSP10 Was Upregulated in Whole Blood From Patients With NSCLC

To investigate the possibility of circUSP10 as a minimally invasive biomarker, we detected the expression levels of circUSP10 in whole blood from 78 NSCLC patients and 38 healthy controls, respectively. Consistent with the expression levels in the NSCLC tissues and cell lines, circUSP10 was significantly upregulated (P = 0.0084, Fig. 5A) in whole blood from patients with NSCLC compared to that in healthy controls. Specifically, the CT value range of circUSP10 in patients with NSCLC is 26.11 to 33.99 (average 29.36), while the one in healthy people is 29.30 to 39.89 (average 33.91). This may provide a reference threshold (CT value less than 29.36) for clinical practice to predict the occurrence of NSCLC.

Figure 5.

CircUSP10 was obviously upregulated in whole blood of patients with NSCLC. (A) The relative expression of circUSP10 in whole blood from NSCLC patients (n = 78) compared to that from health group (n = 38). (B) The relative expression of circUSP10 in whole blood from NSCLC patients with TNM stage I (n = 58) compared to that from health group (n = 38). (C) The relative expression of circUSP10 in whole blood from NSCLC patients with TNM stages II–IV (n = 20) compared to that from health group (n = 38). (D) The relative expression of circUSP10 in whole blood from NSCLC patients with TNM stages II–IV (n = 20) compared to those with TNM stage I (n = 58). Data represent the mean ± SD. NSCLC: non-small-cell lung cancer; TNM: tumor-node-metastasis. **P < 0.01; ****P < 0.0001.

To further explore the expression of circUSP10 in NSCLC patients with early or advanced stages, we analyzed the expression of circUSP10 in whole blood from patients with different TNM stages and from healthy controls. The results showed that circUSP10 was upregulated in NSCLC patients with TNM stage I (n = 58) compared to healthy controls (n = 38) (P = 0.0084, Fig. 5B), as well as in NSCLC patients with TNM stage II-IV (n = 20) compared to healthy controls (P < 0.0001, Fig. 5C). However, no significant difference of circUSP10 expression was found between patients with TNM stage I and II-IV (P = 0.3395, Fig. 5D). Thus, these findings displayed that circulating circUSP10 was significantly upregulated in whole blood from patients with early NSCLC.

Whole Blood-Derived CircUSP10 Showed Better Diagnostic Performance Than Common Tumor Biomarkers

To evaluate the diagnostic performance of circulating circUSP10 in NSCLC, the ROC analysis was conducted in patients with early and advanced stages and in healthy people. For distinguishing NSCLC from healthy controls, the AUC analysis of circUSP10 was 0.9207 (CI = 0.8730–0.9684, P < 0.0001, sensitivity = 88.46% and specificity = 81.58%) (Fig. 6A). To explore whether circUSP10 could serve as a minimally invasive biomarker to screen NSCLC with early stage, the ROC analysis was performed in patients with TNM stage I and in healthy controls. The results showed that the AUC analysis of circUSP10 was 0.9295 (CI = 0.8818–0.9772, P < 0.0001, sensitivity = 83.05%, and specificity = 89.47%) to distinguish patients with TNM stage I from healthy controls (Fig. 6B). In addition, the AUC analysis of circUSP10 was 0.8987 (CI = 0.8457–0.9859, P < 0.0001, sensitivity = 75.00%, and specificity = 94.74%) between patients with TNM stages II to IV and health controls (Fig. 6C). These findings indicated that circulating circUSP10 might serve as a novel biomarker to screen NSCLC at the early stage.

Figure 6.

Whole blood-derived circUSP10 showed good diagnostic performance for screening early NSCLC. (A) ROC curve analysis of whole blood-derived circUSP10 from NSCLC patients (n = 78) and from healthy people (n = 38). (B) ROC curve analysis of whole blood-derived circUSP10 from NSCLC patients with TNM stage I (n = 58) and from healthy people (n = 38). (C) ROC curve analysis of whole blood-derived circUSP10 from NSCLC patients with TNM stages II–IV (n = 20) and from healthy people (n = 38). NSCLC: non-small-cell lung cancer; ROC: receiver operating characteristic; TNM: tumor-node-metastasis; AUC: area under curve.

To further investigate the diagnostic role of circUSP10 combined with common tumor markers, we performed the ROC analyses of CYFR21-1, NSE, and CEA. The results showed that the AUC values of CYFR21-1, NSE, and CEA was 0.5077 (CI = 0.3700–0.6455, P = 0.9174, sensitivity = 15.56%, and specificity = 85.69%), 0.6329 (CI = 0.5007–0.7650, P = 0.0747, sensitivity = 48.89%, and specificity = 86.96%), 0.5932 (CI = 0.4583–0.7282, P = 0.2110, sensitivity = 57.78%, and specificity = 69.57%), respectively (Supplementary Fig. 2A–C). In addition, the combined diagnostic models were designed through logistic regression analysis. The results showed that the AUC values of circUSP10 combined with CYFR21-1, NSE, and CEA were 0.9275 (CI = 0.8676–0.9875, P < 0.0001, sensitivity = 88.89%, and specificity = 86.96%), 0.9324 (CI = 0.8747–0.9900, P < 0.0001, sensitivity = 80.00%, and specificity = 95.65%), 0.9391 (CI = 0.8843–0.9940, P < 0.0001, sensitivity = 95.56%, and specificity = 78.26%), respectively (Fig. 7A–C). Importantly, circUSP10 combined with these three tumor markers together showed a very high AUC value of 0.9372 (CI = 0.8808–0.9936, P < 0.0001, sensitivity = 82.22%, and specificity = 95.65%) (Fig. 7D). Together, not only circUSP10 showed a good diagnostic value, but also circUSP10 improved the diagnostic performance of the common tumor markers.

Figure 7.

CircUSP10 combined with tumor markers improved the diagnostic value of NSCLC. ROC curve analysis of circUSP10 combined with CYFR21-1 (A), NSE (B), CEA (C), and the three tumor makers (D) from NSCLC patients (n = 78) and from healthy people (n = 38). NSCLC: non-small-cell lung cancer; ROC: receiver operating characteristic; NSE: neuron-specific enolase; CEA: carcinoembryonic antigen; AUC: area under curve.

CircUSP10 Was Relative Stable in Whole Blood Under Adverse Conditions

To confirm the stability of circUSP10 in whole blood, the fresh whole blood samples from patients with NSCLC were collected and divided into 4 groups (3 in each group). Prior to RNA isolation, these four groups of whole blood samples were treated with different conditions: groups 1 and 2 were repeated for different freeze-thaw cycles (0, 2, 4, 6, and 8 cycles) at −20°C and −80°C, respectively; groups 3 and 4 were placed at 4°C and room temperature for different time (0, 2, 6, 12, and 24 h), respectively. The results indicated that the relative expression levels of circUSP10 had no significant change under different conditions (Fig. 8A–D). As a reference gene, although GAPDH mRNA did not change under different freeze-thaw cycles at −80°C (Fig. 8A), the linear GAPDH mRNA was significantly changed when the whole blood was repeated for freeze-thaw cycles at −20°C (Fig. 8B), or was placed for different time at 4°C (Fig. 8C) or at room temperature (Fig. 8D). The results demonstrated that circUSP10 was stable under harsh conditions, suggesting a promising biomarker for early diagnosis of NSCLC.

Figure 8.

CircUSP10 was stably expressed in whole blood under harsh conditions. The Cq values of circUSP10 and GAPDH mRNA were compared before the whole blood samples were repeated for different freeze-thaw cycles (0, 2, 4, 6, and 8 cycles) at −20°C (A) and −80°C (B), and were placed at 4°C (C) and room temperature (D) for different time (0, 2, 6, 12, and 24 h). *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001.

Discussion

The main reason for the high mortality rate of lung cancer is that most of the NSCLC patients are diagnosed at advanced stages ¹² . Therefore, it is extremely necessary to improve the effectiveness of early diagnosis for lung cancer. In the present study, we identified circRNA circUSP10 as a noninvasive biomarker with good diagnostic performance for early detection of NSCLC.

CircRNAs, a class of RNAs with single strand and covalent closed-loop structure, are predominantly produced by backsplicing events from corresponding precursor mRNAs^13,14. The continuous expansion of functional studies of circRNA has provided new insights into our understanding of disease development. Growing evidence shows that the dysregulation of circRNA has been involved in the pathogenesis of lung cancer^14,15. For example, circular RNA HIPK3 (circHIPK3) acted as an oncogene and a regulator for autophagy through the miR124-3p-STAT3-PRKAA/AMPKα axis in STK11 mutant lung cancer ¹⁶ . Whereas, circNOL10 was significantly downregulated in lung cancer and affected mitochondrial function by regulating the humanin polypeptide family to inhibit the lung cancer cell growth ¹⁷ . In the present work, circUSP10 was found to be dramatically upregulated in NSCLC tissues and whole blood. Furthermore, considering the close correlation of circUSP10 with tumor size and TNM stage, circUSP10 may serve as an oncogene in NSCLC. Thus, the functional role and molecular mechanism of circUSP10 in tumor development could be further investigated in future studies.

As ideal candidates for biomarker development, circRNAs should be more stable than their cognate linear transcripts ¹⁸ . The most common methods for evaluating the stability of circRNAs are exoribonuclease and transcription inhibitor treatments ¹⁹ . Similarly, our data demonstrated very clearly that the relative expression of circUSP10 was not significantly changed while its cognate linear mRNA was obviously reduced. Through further comprehensive studies, most of circRNAs remain stable in body fluids, including plasma, serum, exosomes, and urine ²⁰ . Consistently, our previous study indicated that the expression of circTOLLIP was fairly stable in whole blood even if the storage condition changed ²¹ . In this work, we did not find a change of the relative expression of circUSP10 in whole blood from patients with NSCLC under different adverse conditions. Hence, circulating circUSP10 may be considered an appropriate biomarker for early diagnosis of NSCLC because of the advantage of its high stability in peripheral blood.

The diagnostic values of circRNAs have been found in multiple diseases ²² . For instance, circRNA_10223 expression was significantly upregulated and acted as a potential biomarker in LUAD ²³ . In addition, the combination of several circRNAs showed better diagnostic accuracy for cancers^24,25. In this study, the ROC analysis showed that circUSP10 possessed great potential for screening NSCLC by analyzing the expression of circUSP10 in whole blood and tissues. Interestingly, the diagnostic performance of circUSP10 in whole blood was better than that in tissues for distinguishing NSCLC from healthy people. Importantly, the AUC value of circulating circUSP10 in NSCLC patients with TNM stage I for distinguishing from healthy controls was slightly greater than that in NSCLC patients with TNM stage I for distinguishing from TNM stages II-IV. However, circUSP10 could not be used to distinguish patients with TNM stage I from those with TNM stages II to IV. In addition, we also analyzed the diagnostic performance of three common tumor markers (CYFR21-1, NSE, and CEA) in whole blood and found that their diagnostic values were much lower than circUSP10. However, circUSP10 would improve the diagnostic values for NSCLC when combined with these biomarkers. Therefore, circUSP10 can act as a promising biomarker for the early diagnosis of NSCLC.

Conclusion

In summary, circUSP10 was upregulated in NSCLC tissues, cell lines, and whole blood. The stably expressed circulating circUSP10 could be used to distinguish NSCLC from healthy people, suggesting the potential as an ideal biomarker for early screening of lung cancer. Admittedly, further work needs to be done to distinguish early NSCLC from noncancerous lung diseases (pneumonia, pulmonary nodules, etc.) in the near future.