Abstract Oral squamous cell carcinoma (OSCC) has high recurrence and mortality rates despite advances in diagnosis and treatment. Therefore, it is necessary to identify new biomarkers for early detection, efficient monitoring, and prognosis prediction. Since microRNA (miRNA) is stable and detectable in serum, it has been reported to inform the diagnosis and monitor disease progression through liquid biopsy. In this study, a circulating specific miRNA panel in OSCC patients was developed, and its usefulness as a dynamic monitor was validated. Small RNAs were extracted from the serum of OSCC patients (n = 4) and normal controls (n = 6) and profiled using next-generation sequencing. NGS identified 42 differentially expressed miRNAs (DEmiRNAs) in serum between patients with OSCC and healthy controls, with threefold differences (p < 0.05). Combining the 42 DEmiRNAs and The Cancer Genome Atlas (TCGA) databases OSCC cohort, 9 overlapping DEmiRNAs were screened out. Finally, 4 significantly up-regulated miRNAs (miR-92a-3p, miR-92b-3p, miR-320c and miR-629-5p) were identified from OSCC patients via validation in the Chungnam National University Hospital cohort. Application of the specific miRNA panel for distinguishing OSCC patients from healthy controls produced specificity and sensitivity of 97.8 and 74%, respectively. In addition, the serum levels of these 4 miRNAs significantly decreased after complete surgical resection and increased after recurrence. We suggest that circulating 4-miRNA panel might be promising non-invasive predictors for diagnosing and monitoring the progression of patients with OSCC. Subject terms: Cancer, Computational biology and bioinformatics Introduction Oral squamous cell carcinoma (OSCC) is the most common type of oral cancer, accounting for approximately 350,000–400,000 cases per year. OSCC is twice as common in male than female due to risk factors, such as tobacco, alcohol and HPV. According to statistics, OSCC is the 6th and 8th particularly for incidence and mortality in both men, respectively^[42]1. Due to the high occurrence of secondary carcinoma and tumor heterogeneity, OSCC is often diagnosed in an advanced state with a poor prognosis^[43]2. Even though most cases of OSCC could be managed with complete surgical resection alone or a combination of ionizing radiation or chemo-radiation therapy, a certain proportion of advanced OSCCs remain unresponsive to treatment or exhibit loco-regional recurrence, resulting in a mortality rate of 50%^[44]3,[45]4. In the diagnosis and prevention of OSCC, emphasis is placed on identifying potential malignant lesions of the oral mucosa and local diseases that promote chronic inflammation, mainly relying on objective clinical examinations and surgical biopsy^[46]5,[47]6. Although surgical biopsy is the gold standard for the diagnosis of OSCC, it is somewhat invasive and can sometimes be harmful to patients^[48]7. Moreover, conventional biopsy is temporally and spatially limited and often provides a brief snapshot of a single region of a heterogeneous tumor^[49]8. Therefore, it is crucial to find promising non-invasive biomarkers for monitoring or patient surveillance and further illuminate the pathogenesis of OSCC regarding tumor behavior at the molecular level. Blood samples are relatively easy to collect in a minimally invasive manner, and increasingly many recent studies have suggested that circulating microRNAs (miRNAs) are promising as potential biomarkers for disease diagnosis and monitoring^[50]9,[51]10. miRNAs are small, non-coding RNAs of 18–25 nucleotides in length that have been linked to essentially all known pathological and physiological processes, including cancer. Recent studies have reported that miRNAs can not only be utilized for diagnosis and prognosis, but also play integral and convoluted roles in the regulatory network of cancer. miRNA have been reported as diagnostic biomarkers for many cancers, including head and neck cancer^[52]11,[53]12. However, the approach of using tissue-derived miRNA in surveillance or prognosis is commonly invasive in nature which may impede the screening. Furthermore, previous studies have demonstrated that miRNA can be quite stable in serum due to its protection from endogenous RNase activity and that it is readily detected by various assays^[54]13,[55]14, presents the possibility to exploiting circulating miRNAs as biomarkers for early-stage cancer. Therefore, serum miRNA panel signatures have recently been identified as promising candidate biomarkers for liquid biopsy. However, studies have rarely examined circulating miRNA expression in patients with OSCC, leading to little noticeable and reliable signatures. The aim of this study was to explore and validate the possibility that circulating miRNAs could overcome the limitations of tissue biopsy and act as potential biomarkers in liquid biopsy for the early diagnosis and dynamic monitoring of disease progression in OSCC patients. Materials and methods Patient and sample collection Serum samples from 27 patients with OSCC and 21 age- and sex-matched healthy individuals were obtained at the Chungnam National University Hospital (CNUH) (Daejeon, Republic of Korea), between January 2017 and December 2019. The clinical information of patients with OSCC were summarized in Table [56]S1. Tumor tissues and adjacent non-tumor tissues were collected from 7 patients with OSCC. All patients with OSCC were enrolled at the initial diagnosis, and the pathological diagnoses were subsequently confirmed. The study participant provided an informed consent form before participating. The Institutional Review Board of CNUH approved this study (CNUH 2019-07-041). All methods were performed in accordance with the Institutional Review Board of CNUH guideline and regulation. Next-generation sequencing and analysis Serum samples from 4 OSCC patients and 6 age- and sex-matched healthy controls were selected from CNUH cohort for next-generation sequencing^[57]10. The clinical information of patients with OSCC for NGS were presented in Table [58]S2. Whole-transcriptome next-generation sequencing was performed by Macrogen Inc. (Seoul, Republic of Korea). Briefly, extracted RNA samples were used to prepare small RNA libraries using SMARTer smRNA-Seq Kit protocol and sequenced using a HiSeq 2500 sequencer (Illumina, San Diego, CA, USA), following the HiSeq 2500 System User Guide Document #15035786 v02 HCS 2.2.70 protocol. After sequencing, the raw sequence reads were filtered based on quality determined by the phred quality score at each cycle (Table S3). Both the trimmed reads and non-adapter reads as processed reads were used, to do analyzing long target (≧ 50 bp).The processed reads were gathered forming a unique cluster. In order to eliminate the effect of large amounts of ribosomal RNA (rRNA) from this study, the read was aligned to the rRNA sequence. rRNA removed reads were sequentially aligned to reference genome (UCSC Homo sapiens reference genome (GRCh37/hg19)), miRBase v21 and non-coding RNA database, RNAcentral 10.0 to classify known miRNAs and other type of RNA such as tRNA, snRNA, snoRNA etc. Novel miRNA prediction was performed by miRDeep2. The read counts for each miRNA were extracted from mapped miRNAs, differentially expressed miRNAs (DEmiRNAs) were determined through comparing across conditions each miRNA using statistical methods. Detailed work flow of sequencing and analysis were additionally described in the supplementary material. Figure [59]S1 represents the small RNA composition of each sample. Bioinformatics Differentially expressed miRNAs (DEmiRNAs) between the evaluated groups were estimated using DESeq2 and edgeR^[60]15. The screening criteria were a fold change > 3 and p < 0.05. All genomic data of OSCC from The Cancer Genome Atlas (TCGA) were obtained from a specific portal ([61]https://tcga-data.nci.nih.gov) and cancer browser ([62]https://genome-cancer.ucsc.edu). To select miRNA differentially expressed between patients with OSCC and normal controls, false discovery rate-adjusted p values (< 0.05) were used to correct, using the Benjamin-Hochberg method. A volcano map, heatmap, and cluster analysis were conducted using an online analysis tool ([63]https://www.chiplot.online/), a free online platform for data analysis and visualization. The target genes of miRNAs were predicted with the TargetScan 8.0 database ([64]www.targetscan.org). Functional annotation was performed using the Database for Annotation, Visualization and Integrated Discovery ([65]https://david.ncifcrf.gov/), a web-accessible tool for Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis. A network analysis of miRNA-mRNA interactions was carried out using Cytoscape (version 3.7.1), an open bioinformatics software. RNA extraction and quantitative real time polymerase chain reaction (qRT-PCR) Circulating miRNA was isolated from 200 μL of serum for RNA purification using miRNeasy serum/plasma kits (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. Total RNA was extracted from tissue samples using the TRIzol reagent (Invitrogen, Waltham, MA, USA). The SYBR Green qRT-PCR assay was used for miRNA quantification. Total miRNA was used as the template for cDNA synthesis with miScript II RT Kit (Qiagen, Hilden, Germany) according to the manufacturer’s instructions. For miRNA analysis, qRT-PCR was performed using the miScript SYBR Green PCR kit (Qiagen, Hilden, Germany) with the manufacturer-provided miScript assays, using the universal primer and miRNA-specific forward primers with the 7500 system. The miRNA-specific primers were obtained from the miScript primer assays, the miRCURY LNA miRNA PCR Assay (Qiagen, Hilden, Germany), and Bioneer (Daejeon, Korea). All primer sequences used for qRT-PCR are listed in Table [66]S4. miR-16 and miR-423-5p were used as references for serum miRNA