Early detection of colorectal neoplasms, including colorectal cancers (CRCs) and advanced colorectal adenomas (AAs), is crucial to improve patient survival. Circulating microRNAs (miRNAs) in peripheral blood are emerging as noninvasive diagnostic markers for multiple cancers, but their potential for screening colorectal neoplasms remains ambiguous.

To identify candidate circulating cell-free miRNAs as diagnostic biomarkers in patients with colorectal neoplasms.

The study was divided into three phases: (1) Candidate miRNAs were selected from three public miRNA datasets using differential gene expression analysis methods; (2) an independent set of serum samples from 60 CRC patients, 60 AA patients and 30 healthy controls (HCs) was included and analyzed by quantitative real-time polymerase chain reaction for miRNAs, and their diagnostic power was detected by receiver operating characteristic (ROC) analysis; and (3) the origin and function of miRNAs in cancer patients were investigated in cancer cell lines and tumor tissues.

Based on bioinformatics analysis, miR-627-5p and miR-199a-5p were differentially expressed in both the serum and tissues of patients with colorectal neoplasms and HCs and were selected for further study. Further validation in an independent cohort revealed that both circulating miR-627-5p and miR-199a-5p were sequentially increased from HCs and AAs to CRCs. The diagnostic power of miR-672-5p yielded an area under the curve (AUC) value of 0.90, and miR-199a-5p had an AUC of 0.83 in discriminating colorectal neoplasms from HCs. A logistic integrated model combining miR-199a-5p and miR-627-5p exhibited a higher diagnostic performance than either miRNA. Additionally, the levels of serum miR-627-5p and miR-199a-5p in CRC patients were significantly lower after surgery than before surgery and the expression of both miRNAs was increased with culture time in the culture media of several CRC cell lines, suggesting that the upregulated serum expression of both miRNAs in CRC might be tumor derived. Furthermore, in vitro experiments revealed that miR-627-5p and miR-199a-5p acted as tumor suppressors in CRC cells.

Serum levels of miR-199a-5p and miR-627-5p were markedly increased in patients with colorectal neoplasms and showed strong potential as minimally invasive biomarkers for the early screening of colorectal neoplasms.

Colorectal cancer (CRC) is one of the most common malignant digestive tumors, ranking fourth regarding incidence and mortality rates among all cancer types in 2018 and posing a tremendous threat to global population health[1]. The 5-year survival rate of CRC patients highly depends on the tumor stage at diagnosis, approximately 90% for stage I disease and less than 8% for stage IV disease, highlighting the crucial importance of early cancer detection and diagnosis to reduce the burden of this disease[2,3]. Most CRCs start with aberrant crypt foci, evolving into benign premalignant lesions (adenomas), and ultimately progressing to colorectal tumors (adenocarcinomas) over the span of years[4]. The classical process of colorectal tumorigenesis lends itself to screening. Notably, advanced colorectal adenoma (AA) is considered the protracted and treatable preclinical phase and is suitable for population screening due to its significant correlation with a high risk of progression to invasive cancer[5]. Thus, developing cost-effective screening tools for colorectal neoplasms, including CRCs and AAs, is crucial.

Although several advances in diagnostic techniques have been confirmed to effectively decrease the mortality of CRC patients, several challenges persist regarding current CRC screening tools, such as the lack of a universally accepted strategy, the lack of a high specificity and sensitivity strategy, and the high cost of screening advanced adenoma[6]. For example, colonoscopy is currently the most effective and straightforward examination to diagnose advanced lesions and prevent cancer progression by resecting precancerous polyps in an en bloc manner[4]. However, this modality comprises prior bowel cleansing, an invasive procedure, with a relatively high expense, leading to poor compliance of patients and exposing patients to potentially procedural risks[7]. Alternative, non­invasive stool-based screening tools, such as the fecal occult blood test, have been criticized for the poor detection of advanced adenomas and high false-positive rates due to other confounding illnesses, such as gingival bleeding[8]. Additionally, circulating tumor markers, such as carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA199), have been emerging as potential alternatives for the minimally invasive detection and prevention of CRC. Unfortunately, their less-than-desirable sensitivity and specificity have hampered their clinical use in the early detection of CRC and premalignant lesions[9]. Therefore, identifying a simple, inexpensive, noninvasive or minimally invasive screening approach with good sensitivity and specificity to improve the detection of early colorectal neoplasms is required.

CRC originates from a number of genomic instabilities, epigenetic alterations, and genetic aberrations, which offer the foundation for detecting serum or plasma to identify tumor-specific fingerprints[10]. MicroRNAs (miRNAs) are endogenously derived non-coding RNA sequences of 19-25 nucleotides that regulate the expression of target genes through either miRNA degradation or translational repression[11]. Since their discovery in 1993, every cancer type detected by miRNA sequencing has demonstrated that miRNA expression profiles differ between normal tissues and tumor tissues, including CRC[12]. Altered miRNA expression exerts a tumor-suppressive or oncogenic function in regulating signaling pathways involved in CRC progression from normal to premalignant lesions, thereby contributing to malignancy[13-15]. Furthermore, Chen et al[16] demonstrated that peripheral blood contains a number of stable miRNAs secreted from various tumor tissues, raising the possibility of their potential use as noninvasive diagnostic markers for cancer. In fact, miRNAs are superior to mRNAs due to their small sizes, easy detectability, and stability. As serum or plasma miRNAs are resistant to RNase digestion, they remain stable for extended storage even when subjected to 10 freeze-thaw cycles, boiling, and low or high pH[16]. An increasing number of studies have demonstrated that several circulating miRNAs have exhibited the potential to distinguish individuals with AA and CRC from healthy volunteers, such as upregulated miR-29a, miR-92a, and miR-21[17,18]. Thus, circulating cell-free miRNAs are likely robust and novel molecular markers in early colorectal neoplasm diagnosis. However, few miRNAs have been widely applied in clinical practice because of the failure to distinguish patients with AAs from healthy volunteers, the small sample sizes, and the lack of pathological features of subjects.

As the development of genome-sequencing technology has strongly catalyzed the understanding of cancer biology, an increasing number of publicly available genomic datasets, such as the Gene Expression Omnibus (GEO), has been utilized to identify potential biomarkers for cancer diagnosis. In the present study, the miRNA expression profiles of the GEO GSE25609, GSE115513, and GSE41655 datasets were used to mine differentially expressed miRNAs coexisting in serum and tissues between patients with colorectal neoplasms and healthy controls (HCs). Next, we screened circulating cell-free miR-199a-5p and miR-627 as potential diagnostic biomarkers of colorectal neoplasms, and explored their expression in serum, tissues and cancer cell lines.

Identifying non-invasive tumor markers for the early screening of colorectal tumors and precancerous lesions is critical to reduce cancer incidence and extend the long-term survival of patients through timely recognition and radical surgery of advanced colorectal neoplasia. With the development of genome-sequencing technology and comprehensive bioinformatic analysis, the potential abilities of circulating miRNAs released from cancer cells as non-invasive biomarkers have been recognized by an increasing number of scientists. Notably, using circulating miRNAs as tumor markers is a better strategy than using RNA or protein biomarkers due to their reproducible, highly stable, and consistent characteristics on individuals of the same species. In the present study, we first identified that circulating miR-627-5p and miR-199a-5p were markedly elevated in patients with colorectal neoplasia and appeared to be novel biomarkers in the non-invasive screening of colorectal neoplasia.

In the discovery stage, we screened candidate circulating cell-free miRNAs in serum as liquid biopsy biomarkers of colorectal neoplasm using bioinformatic analysis from three GEO datasets. Two overlapping miRNAs, miR-627 and miR-199a-5p, were identified and subjected to ROC analysis. As the mature sequences of miR-627 were designated as “-5p” and “-3p” and a growing number of studies have focused on the role of miR-627-5p in cancer development[22-24], we chose miR-627-5p for further validation. Thus, we speculated that circulating miR-199a-5p and miR-627-5p have potential as biomarkers of colorectal neoplasia. In the validation stage, we sought to determine their efficiencies in discriminating patients with CRC and AA from HCs by target quantification methods. However, previous studies have demonstrated that not all of the miRNAs detected by next-generation sequencing could be steadily achieved using the qRT-PCR method[25]. Several factors have influenced the successful detection of miRNAs, including the relatively low expression of specific miRNAs, sequence length variations of unique miRNAs, and bias of detection methods. We confirmed that the circulating expression levels of miR-199a-5p and miR-627-5p were markedly increased in patients with precancerous lesions and CRC compared with neoplasm-free controls by qRT-PCR, particularly those diagnosed with CRC, suggesting that the increase in both miRNAs occurred during colorectal carcinogenesis. Additionally, serum miR-199a-5p and miR-627-5p levels were significantly elevated in male CRC patients and in those with larger tumor sizes, but were relatively stable in patients with other clinicopathological characteristics, such as age, tumor location, histological grade, and TNM stage. Colorectal adenomas comprise several histological types, including sessile serrated adenoma, TA, and tubulovillous adenoma (TVA). Considering that the traditional adenoma-carcinoma sequence developed by the accumulation of multiple genomic events contributes to 70%-90% of CRC[4], we mainly included patients with TA and TVA in the present study. Although different subtypes of precursor colorectal lesions may involve different genetic alterations and distinct miRNA changes[26], our study demonstrated no difference in the serum miR-199a-5p and miR-627-5p levels between TA and TVA patients, further confirming the promise of both miRNAs as robust biomarkers for precancerous adenomas. However, considering the small sizes in the subgroups, these results should be further validated in an independent large cohort.

To date, a variety of studies have demonstrated that components of colorectal tumors (such as miRNAs, mRNAs, proteins and exosomes), which are released into biological fluids, can be detected by liquid biopsy and used as circulating biomarkers[21]. However, most studies only focused on the predictive power of circulating biomarkers in the identification of CRC and not in the diagnosis of precancerous lesions. As early as 2009, a study showed that miR-92 was markedly increased in CRC patients and had a high diagnostic accuracy with an AUC value of 0.885[27], but patients with adenoma were not included in the study. Aberrant expression of circulating exosomal miR-125a-3p has been detected in the serum of patients with early stage colon cancer, but its diagnostic accuracy was relatively low (AUC = 0.685)[25]. Overexpression of serum metadherin mRNA from CRC patients when compared to HCs was firstly described by Ghafar et al[28] in 2020 and its diagnostic accuracy (AUC = 0.976) was significantly higher than other routine tumor markers. However, the potential utility of circulating metadherin mRNA in the identification of AAs has not been studied. The most noteworthy results of the present study were that serum miR-627-5p and miR-199a-5p could act as novel minimally invasive biomarkers for colorectal neoplasm screening. By enrolling an independent cohort including CRC, AA and HCs, we conducted a powerful statistical analysis to draw conclusions. Circulating cell-free miR-199a-5p showed promising potential in differentiating CRC patients from HCs (AUC = 0.90), but its diagnostic efficiency in discriminating AAs from controls was relatively low (AUC = 0.76). MiR-627-5p exhibited better performance in differentiating cancer patients from controls (AUC = 0.97) and AA patients from HCs (AUC = 0.84), revealing that the marked upregulated expression of serum miR-627-5p was more likely to reflect the early stage during the progression from precancerous lesions to neoplasia. Additionally, circulating miR-627-5p could discriminate patients with colorectal neoplasms from the neoplasm-free group (AUC = 0.90). Concurrently, the AUCs for the conventional tumor markers CEA and CA199 were lower than those for both miRNAs. Considering that the optimal target lesion for CRC prevention would be AA, which has a higher risk of evolving into a tumor, miR-627-5p could be a better minimally invasive biomarker for the early screening of colorectal neoplasms.

Currently, a number of studies have demonstrated that combined genes can be used to establish integrated diagnostic models that help to improve the diagnostic accuracy. For example, Giraldez et al[19] established a combined signature consisting of miR-19a, miR19-b, and miR-15b, discriminating early colorectal tumors with a specificity of 79.25% and a sensitivity of 80.95%. Another study constructed a serum-based four-miRNA signature (miR-23a-3p, miR-27a-3p, miR-142-5p and miR-376c-3p) that could discriminate patients with colon cancer from control groups with an AUC of 0.92[29]. In the present study, a logistic integrated model combining miR-199a-5p and miR-627-5p exhibited a higher diagnostic performance than using either miRNA alone, and it diagnosed AA with both high sensitivity and specificity as robustly as CRC. Several advantages exist regarding the application of the integrated model as a practical and economic non-invasive screening strategy for colorectal neoplasms: (1) The integration of miR-199a-5p and miR-627-5p built a more reliable diagnostic model than conventional tumor markers and each miRNA alone, reflecting the complex molecular mechanisms of CRC initiation and development; (2) the use of blood samples may lead to better patient compliance than guiding patients to conduct bowel preparation and endoscopy; and (3) the diagnostic model comprising only two miRNAs would save the expense of laboratory testing and lower the training requirements for clinical doctors, thereby promoting its potential usefulness in clinical practice.

Growing evidence has demonstrated that dysregulated expression of miRNAs plays key roles in modulating multiple genes crucial in cancer initiation, progression and metastasis[13]. Hence, we retrieved the target genes of miR-627-5p and miR-199a-5p and functionally annotated them to further understand the biological role of both miRNAs. The target genes were mainly related to multiple critical signaling pathways involved in the multistep carcinogenesis, such as Wnt, mTOR, Hippo, transforming growth factor beta receptor signaling pathway and focal adhesion. Abundant research has documented that constitutive activation of Wnt signaling is the pathological basis of many malignancies, including CRC[30]. Transforming growth factor beta signaling pathways have been proved to be essential for the proliferation, angiogenesis, and invasion of CRC[31]. These pathways represented only a small subset of target genes participating in signaling pathways, supporting the function of miR-627-5p and miR-199a-5p in regulating cancer initiation and progression. In our study, miR-627-5p was demonstrated to be not only sequentially decreased from normal colorectal tissues and precancerous lesions to cancer tissues, but also significantly reduced in the human colon cancer cell lines compared with than in FHC cells. MiR-199a-5p was markedly reduced in CRC tissues and cell lines but not significantly decreased in AA tissues. Additionally, in vivo gain-of-function assays revealed that enhanced expression of both miRNAs suppressed the aggressive behavior of CRC cells, suggesting their suppressive role in CRC development. It is worth mentioning that the same miRNA may exert different functions depending on the circumstances, for example, promoting tumor growth in several cancer types, and acting as a tumor suppressor in other tumor types[32]. MiR-627-5p has been demonstrated to be reduced in several cancer tissues, including CRC[33], glioblastoma multiforme[24], oral squamous cell carcinoma[22], and hepatocellular carcinoma[34], supporting its suppressor role in cancer progression. However, Shin et al[35] mentioned that miR-627-5p was markedly increased in patients diagnosed with gastric cancer when compared with their non-tumor counterparts and was considered a potential oncogene in gastric cancer. The regulatory mechanism of miR-627-5p in CRC has not been documented to date, but a recent publication has revealed that decreased expression of miR-627-5p exerts inhibitory effects on hepatocellular carcinoma possibly by targeting the BCL3 transcription coactivator[34]. An in vitro study reported that miR-627-5p suppressed cell proliferation, retarded migration, and promoted apoptosis by increasing YBX2 expression in oral squamous cell carcinoma[22]. Further research is required to explore the underlying mechanism of miR-627-5p in the carcinogenesis of CRC. MiR-199a-5p, a member of the miR-199 family, has been demonstrated to be a proliferation inhibition gene in a variety of tumors, including non-small cell lung cancer[36], laryngeal cancer[37], breast cancer[38], and hepatocellular carcinoma[39]. Extensive studies have been performed to explore the suppressive function of miR-199a-5p in CRC. For example, in a clinical study with a small sample size, miR-199a-5p expression was markedly decreased in clinical CRC tissues than in the control group and inversely correlated with discoidin domain receptor 1 (DDR1) expression. Additionally, downregulated miR-199a-5p increased the level of DDR1, activating epithelial-to-mesenchymal transition related signaling, thus enhancing the migratory and invasive abilities of human CRC cells[40]. These findings were similar to those of another study with a large sample size, showing that the expression level of miR-199a-5p was sequentially reduced in T2, T3 and T4 colorectal tumors compared to that in normal tissues[39]. The study also demonstrated that miR-199a-5p suppressed cell proliferation, migration and invasion of CRC cells by negatively regulating integrin alpha-3[41]. Zhu et al[42] suggested that miR-199a-5p suppressed the growth of CRC cells potentially by inactivating the phosphoinositide 3-kinase/AKT signaling pathway. These previous studies supported the notion that miR-199a-5p plays a crucial role as a tumor suppressor by targeting various genes in CRC, providing a new therapeutic target to inhibit CRC growth and metastasis.

It is worth noting that the levels of miR-627-5p and miR-199a-5p were increased in CRC serum but decreased in tumor tissues. The inconsistency between the serum level of miRNAs and tissue expression validated in CRC and AA suggested that the release of miRNAs is complex during the course of cancers. In general, the circulating miRNA expression pattern should be closely correlated with the tissue expression pattern in the same cancer type. Hur et al[43] found that circulating miR-203 levels were markedly increased in CRC patients with liver metastasis compared with those in primary CRC patients, a finding that was concordant with the markedly high expression in liver metastasis specimens compared with matched primary CRC tissues. However, discrepancies between the circulating levels and tissue levels have been observed in many miRNAs. miR-145 showed the opposite trend, with upregulated expression in non-small cell lung cancer serum[44] and reduced expression in tissues[45]. Similarly, Lian et al[46] suggested that serum miR-199a-3p level was markedly higher in osteosarcoma patients than in HCs, whereas Duan et al[47] reported that miR-199a-3p was decreased in the same tumor tissues. Thus, the miRNA expression patterns between serum and tissues may be opposite. In our study, the levels of miR-627-5p and miR-199a-5p were significantly decreased in postoperative serum samples compared with those in preoperative samples and their expression levels increased with culture time and cell numbers in the culture media of colon cancer cell lines, which suggested that both miRNAs might be tumor-derived. The conflicting data between intracellular and extracellular profiles of miR-199a-5p and miR-627-5p questioned the existence of a specific secretion mechanism for circulating miRNAs. Accumulating evidence has demonstrated that different sources of miRNAs are present in the biological fluids of cancer patients[21,48]. miRNAs can be actively secreted by tumor cells into extracellular fluids by encapsulation with membrane-bound vesicles, including exosomes and shedding microvesicles, or by binding to Argonaute proteins. Some circulating miRNAs are regarded as merely byproducts of tumor cell apoptosis and necrosis. Additionally, circulating tumor cells and blood platelets are also packed with intracellular miRNAs. Therefore, we hypothesized that circulating miR-199a-5p and miR-627-5p originate from the active secretion of tumor cells and passive release from death cells, likely representing biological processes for miRNA clearance in tumor tissues.

Despite some strengths of the present study, it also has some weaknesses. First, a major issue is that we did not enroll patients with other diseases such as inflammatory bowel disease and other cancer types to confirm the diagnostic efficiency of both miRNAs. Second, the small sample size in the control group may reduce the validity and reliability of our diagnostic study. Indeed, recruiting truly healthy volunteers is practically difficult, requiring the collection of each individual’s personal health information and a thorough medical examination. Further research with a large sample size is anticipated to validate whether both miRNAs can be used in routine clinical diagnosis. Third, an optimal protocol for sample preparation is also a key issue in our study. Arroyo et al[49] suggested that the blood clotting procedure may induce the secretion of a number of miRNAs from platelets, thus influencing the abundance of tumor-specific miRNAs. Thus, the use of blood plasma is preferred in future studies. Finally, the origin of circulating extracellular miR-199a-5p and miR-627-5p in patients with colorectal neoplasms was unclear and a series of in vivo and in vitro experiments should be performed to answer this question and further clarify the underlying mechanism of both miRNAs in the initiation and progression of colorectal neoplasms.

To our best knowledge, our study is the first to validate the markedly elevated levels of serum miR-199a-5p and miR-627-5p in patients with CRC and AA, revealing the potential of both circulating cell-free miRNAs in serum as liquid biopsy biomarkers of colorectal neoplasms. An integrated model combining miR-199a-5p and miR-627-5p obtained an enhanced discriminative capacity and can serve as a practical and economic non-invasive screening strategy for colorectal neoplasms. Furthermore, in vitro gain-of-function experiments suggested that both miRNAs act as tumor suppressors in the tumorigenesis of CRC.

The authors thank Gene Expression Omnibus (GEO) for providing invaluable datasets for statistical analyses.