Circulating cell‐free circRNA panel predicted tumorigenesis and development of colorectal cancer

Abstract Background Colorectal cancer (CRC) is reported with high morbidity and mortality. Currently, the sensitivity of diagnostic markers for colorectal cancer is low. Therefore, further exploration of new plasma diagnostic markers for early detection of colorectal cancer is of great value. We aimed to explore potential circRNAs in plasma as biomarkers for early diagnosis of CRC. Methods We employed the circRNA microarray to investigate dysregulated circRNAs in plasma samples of CRC patients, colorectal adenoma patients (CRA), and healthy controls. Through in‐depth analysis, significantly differentially expressed circRNAs were screened as candidate targets. Results Eight circRNAs (hsa_circ_104885, hsa_circ_100185, hsa_circ_103171, hsa_circ_001978, hsa_circ_105039, hsa_circ_103627, hsa_circ_101717, and hsa_circ_104192) were obtained as candidate circRNAs with upregulation in CRC comparing with both CRA and healthy control. Through detecting the plasma expression levels of eight candidate targets, we identified three circRNA (hsa_circ_001978, hsa_circ_105039, and hsa_circ_103627) with increased level which were consistent with the microarray results in training set. Further validation found the circRNA panel was consistent with training set. The ROC curve also revealed a high diagnostic ability of hsa_circ_001978, hsa_circ_105039, and hsa_circ_103627 in predicted the CRC from CRA patients (AUC = 0.966) as well as healthy controls (AUC = 0.969). Conclusion Our data suggest that hsa_circ_001978, hsa_circ_105039, and hsa_circ_103627 might be a CRC‐specific biomarker for early diagnosis.

developed countries has been ranked in the third and second malignant tumors. In China, with economic development, the incidence of colorectal cancer is increasing year by year. 5 Most patients have been diagnosed with local advanced or distant metastasis, especially liver metastasis, with poor treatment effect and poor prognosis. 2 In recent decades, researchers have developed many effective methods for the diagnosis and treatment of colorectal cancer, which significantly improve the survival of colorectal cancer patients. However, the overall prognosis of colorectal cancer patients is still poor. 6 Due to the lack of specific clinical manifestations, simple, and fast diagnostic methods and accurate evaluation methods, diagnosis, and treatment of colorectal cancer is still difficult. There is a long way to go in the diagnosis and treatment of colorectal cancer in China. Therefore, to improve the prognosis of patients with colorectal cancer, the key point is to improve the efficiency of early diagnosis of colorectal cancer. Developing effective minimally invasive diagnostic markers to detect colorectal cancer at an earlier stage, patients with colorectal cancer can receive timely treatment, effectively improve the survival of patients, which has important clinical research value.
Recently discovered circRNAs consist mainly of exon-derived transcripts produced by non-collinear reverse splicing. 7 Based on whether they can be translated, circRNAs can be divided into noncoding circRNAs and coding circRNAs. [7][8][9] The circRNA is widely expressed in a variety of human cells and plays a key role in the regulation of post-transcriptional gene expression. 10 Different from linear RNA, circRNA harbored a closed-loop structure, stable expression, is not easy to degrade, and is affected by RNA exonucleases. 11 The circRNA has miRNA binding sites and acts as a miRNA sponge, blocking or reducing miRNA expression and promoting the expression of target mRNA. 12 There is an increasing evidence that circRNA can be used as biomarkers for early diagnosis and prognostic prediction in various diseases. 13 Abnormal circRNA expression in peripheral blood can be used as a biomarker of rectal cancer. 14 Similarly, circRNA can also be used as an early diagnostic marker for some non-neoplastic diseases, such as Parkinson's disease, and can be used to monitor the severity of the disease. 15 Besides, recent colorectal cancer tumor markers, such as CEA and CA19-9, have been confirmed to have obvious limitations in their diagnostic sensitivity and specificity, especially in patients with early colorectal cancer, whose diagnostic efficacy is weak. However, it can predict postoperative metastasis and recurrence of colorectal cancer. 16 Based on this, it is necessary to explore novel biomarker for the early diagnosis of colorectal cancer.
In this project, we anchored circRNA as the research object and combined with the characteristics of stable expression of circRNA, we believe that circRNA might be used as a potential biomarker. We

| Study design
The circRNA microarray was employed by using Arraystar Circular Next, we conducted an in-depth analysis of the differentially expressed circRNA detected by microarray. Intersection analysis was conducted for circRNA with high differential expression compared with CRA group and circRNA with elevated expression compared with healthy people, and the final differential circRNA was obtained as a candidate. Subsequently, plasma samples from 20 CRC patients and 20 healthy controls were randomly selected for preliminary validation to obtain circRNA with differential expression in a small sample. Based on the above results, a large sample (80 cases in each group) was used for revalidation. Finally, the sensitivity and specificity of candidate patients and their combinations in CRC diagnosis were calculated by setting training set and validation set. The receiver operating characteristic (ROC) curve analysis was applied.

| RNA isolation and qRT-PCR
All blood samples were stored by EDTA anticoagulant tube, and plasma samples were obtained by 1000 g centrifugation for 10 min within half an hour after blood samples were taken. All plasma samples were stored in the −80°C refrigerator. Total RNA was extracted by Trizol method, and circRNA was enriched by RNase treatment after RNA quality control detection. The RNase-treated RNA were retrotranscribed using random primers method, and the reverse transcribed cDNA was amplified using a SYBR ® Premix Ex Taq™ kit (Takara). Finally, the amplification curve was compared with the internal reference to obtain the relative expression level, and triplicates were used to detect each samples. ABI7900 (ABI) was used for amplification.

| Statistical analysis
In this study, continuous variables were analyzed using the student t test, and categorical variables were analyzed using the chi-square test. All the experiment assay was repeated three times in triplicate.
We have also tested the homoscedasticity in all the data obtained.
After confirmed the homoscedasticity, after that, a parametric test (t test, ANOVA) was applied. In analyzing the sensitivity and specificity of diagnostic markers, we quantified the area under the receiver operating characteristic (ROC) curve. Specific methods of risk stratification analysis are described in previous reports. STATA14.0 was used for statistical analysis, and GraphPad Prism 8.0 was used for graph presentation. In all statistical analyses, p < 0.05 was considered statistically significant.

| High-throughput microarray detection of plasma circRNAs
Totally, 100 patients diagnosed with CRC were enrolled with 100 paired healthy controls. 100 patients diagnosed with colorectal adenoma was also participated. The detailed clinical information is listed in Table 1. Based on clinical information analysis, it was found that there were no significant statistical differences in age and gender among the groups in this study. We further presented subgroup information based on clinical characteristics of tumor patients, including tumor size, tumor differentiation degree, and tumor TNM analysis.
After obtained the dysregulated circRNA, the following parameters were used for difference analysis for all circRNA: (1) p < 0.05, (2) the absolute value of CT detection <35, and (3) 75% detectable in all the samples. After the above screening, we obtained 134 circRNA with high expression in CRC compared with the CRA group.120 cir-cRNA were elevated in CRC compared with healthy subjects. Further, the enrichment of differentially expressed circRNA in each group was presented by clustering map, and the general distribution of differentially expressed circRNA was shown by volcanic map (Figure 1A-C).
Further cross comparison was conducted for the above two groups of differentially expressed circRNAs, and 8 circRNAs with the highest expression in CRC were obtained ( Figure 1D). We further selected the eight circRNAs as candidate biomarkers.
We next employed an independent cohort to further validate

| Risk score analysis for colorectal cancer biomarkers
The panel of three circRNAs (hsa_circ_001978, hsa_circ_105039 and hsa_circ_103627) was further validated in an independent cohort including 80 samples in each group. Comparing with both CRA group and healthy control group, we found the three circRNA panel was evaluated in the plasma samples of CRC patients ( Figure 3).
As mentioned above, based on the above study and multi-step verification, we preliminarily found three circRNA panel might be potential biomarkers of CRC. We next employed the risk score analysis Data was presented as mean ± SD, **indicated p < 0.01. *indicated p < 0.05, n.s, indicated no significance obtained the sum of sensitivity and specificity as the node of the maximum risk score, regarding as the cutoff. We classified the population as high-risk group and low-risk group. The high-risk group was suggested to be colorectal cancer, and the low-risk group was the control group (value = 4.31). We found the positive predicting value for CRC from healthy was 1.00 and the negative predicting value was 1.00.
The same method was applied in validation set, and the obtained the positive predictive value and negative predictive value were 0.90 and 0.91, respectively, based on a cutoff value of 8.72 ( Table 2).
The specific diagnostic efficacy of the above fingerprints was presented by ROC curves. As shown in Figure 4, the AUC of hsa_

| Expression stability validation
To verify the stable presence of the three circRNAs in human plasma, plasma samples from randomly selected healthy patients were treated in different ways to test the stability of circRNA expression in different environments. The samples were placed at room temperature for 12 and 24 h. In addition, the plasma samples were subjected to three consecutive freeze-thaw treatments. It was found through detection that the expression levels of these three circRNAs in the above plasma did not altered significantly indicating that the three circRNA panel could be stable expressed in human plasma samples ( Figure 6).

| DISCUSS ION
At present, the diagnosis of CRC mainly relies on colonoscopy, imaging and tumor biomarker detection. However, due to the unsatisfactory results of these methods, patients with CRC are usually diagnosed as intermediate to late stage. In addition, as the gold standard, colonoscopy has not been widely promoted due to the high cost of examination, the painful process of examination, and the poor health awareness of the public. 17 Similarly, imaging examination is difficult to be used for early diagnosis of colorectal cancer due to its poor identification. In terms of diagnostic markers, carcinoembryonic antigen and CA19-9 has been widely applied in the diagnosis of tumor, but previous studies show that CEA is more suitable for being used in the detection of for tumor recurrence and metastasis; in addition, the CEA levels in intestinal inflammation and adenoma has increased in the tumor, for early colorectal cancer its poor sensitivity and specific degree, CA19-9 is a good predictor of tumor metastasis. 18 Therefore, finding new tumor markers with high sensitivity and specificity is very important for the diagnosis of CRC.
It is worth noting that in the current research, more and more researchers have found that circRNAs play an important role in many human diseases. In addition, many circRNAs can be expressed in plasma. For example, circ_0124554 blocked the ubiquitination of AKT promoting the skip lymphovascular invasion on hepatic metastasis in colorectal cancer. 19 Due to the circRNA specific circular structure, they can be stably expressed in human body fluids, including plasma and urine, and thus have the ability to serve as biomarkers. 20 It has been shown that circRNA can be used as an important marker of early tumor occurrence and predict the occurrence of gastric cancer in gastrointestinal tumors. 21

This work was supported by the Science Foundation of Anhui
Medical University project (2020xkj038).

CO N FLI C T O F I NTE R E S T
No potential conflict of interest was reported by the authors.

AUTH O R CO NTR I B UTI O N S
XC conceived and designed the experiments; LQ and YP performed the experiments and contributed to reagents/materials/analysis tools; LQ AND MT analyzed the data; All authors contributed to the article and approved the submitted version.

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.