LncRNA MIR22HG and miR-10a-5p: Pioneering serum biomarkers for pancreatic cancer diagnosis and progression assessment Authors: Rowyda M Salman, Abdullah F Radwan, Olfat G Shaker, Ahmed A El-Husseiny Published: 2026 ABSTRACT AbstractBackgroundPancreatic cancer (PC) endures as one of the most lethal malignancies, characterized by a dismal prognosis due to late-stage detection and limited therapeutic options. Circulating noncoding RNAs (ncRNAs) play significant roles in malignancy detection and prognosis. Thus, we evaluated the clinical value of serum long ncRNA MIR22HG and miR-10a-5p in PC cases, clarifying their relationship with downstream targets: β-catenin, C-myc, and E-cadherin.MethodsA case-control study was conducted involving 65 PC cases and 25 healthy controls, utilizing RT-qPCR to measure serum levels of MIR22HG, miR-10a-5p, β-catenin, C-myc, and ELISA to investigate E-cadherin. Besides, CA19.9 and CEA were assessed employing the chemiluminescence approach.ResultsOur findings revealed significant downregulation of MIR22HG, β-catenin, and C-myc, alongside upregulation of miR-10a-5p and E-cadherin in the serum of PC cases compared to controls. MIR22HG and miR-10a-5p demonstrated high diagnostic accuracy for PC with AUCs of 0.95 and 0.97, respectively. The diagnostic efficiency of PC traditional markers, CA19.9 or CEA, was enhanced by integrating with MIR22HG and miR-10a-5p. Moreover, low MIR22HG and β-catenin serum levels were notably associated with advanced lymphatic and distant metastases, distinguishing between metastatic and curative PC cases with AUCs of 0.73 and 0.82, respectively. MIR22HG was notably positively correlated with β-catenin and C-myc, whereas it was negatively correlated with miR-10a-5p.ConclusionThis study highlights the clinical potential of serum MIR22HG and miR-10a-5p as novel diagnostic biomarkers and identifies MIR22HG and β-catenin as prognostic indicators, thereby paving the way for improved PC management.Keywords: Pancreatic cancer, MIR22HG, miR-10a-5p, Metastasis, Biomarker Introduction Pancreatic cancer (PC) has been regarded as one of the most aggressive and life-threatening types of cancer, characterized by a subtle and often asymptomatic onset [1]. It ranks as the fourth primary cause of cancer-related fatalities worldwide [2]. According to GLOBOCAN 2022, PC's global age-standardized incidence and mortality rates are 4.9 and 4.7 per 100.000 persons, which are closely coincident [3]. In spite of improvements in cancer therapy, PC patients still have a poor 5-year survival rate of 5% to 9% due to resistance to therapies, ending up in a dismal prognosis [4]. PC is a malignant tumor that grows without distinct symptoms, resulting in late diagnosis in advanced stages, with only 15–20% of patients undergoing surgery [5]. Current diagnostic biomarkers for PC, such as cancer antigen 19.9 (CA 19.9) and carcinoembryonic antigen (CEA), are typically utilized but have limited specificity and sensitivity. Although CA 19.9 is useful for monitoring advanced PC and assessing treatment response, it can also be elevated in non-malignant conditions like pancreatitis and cholangitis [6,7]. Carcinoembryonic antigen (CEA) is another biomarker that, when combined with CA 19.9, can improve diagnostic accuracy. However, its sensitivity for early-stage diagnosis is constrained [8]. These limitations display the requirements for additional biomarkers, particularly epigenetic markers, to enable earlier detection and better prediction of prognosis. Epigenetic changes, including aberrant noncoding RNAs (ncRNAs), histone alterations, and DNA methylation, possess a critical role in PC development and progression [[9], [10], [11]]. ncRNAs, comprising long noncoding RNAs (lncRNAs) and microRNAs (miRNAs), participate in regulating key biological functions such as tumor repression, cell migration, and metastasis [12,13]. In PC, both miRNAs and lncRNAs have been discovered as interesting therapeutic and diagnostic targets; in addition, their reciprocation results in regulatory RNA-crosstalk, which is implicated in numerous carcinogenesis and metastatic processes. Thus, the understanding of this interaction is still emerging to reveal new and novel diagnostic biomarkers for PC [[14], [15], [16]]. The miR-10a family, an atypical group of microRNAs, performs a regulatory action in cell proliferation and migration across various malignancies [17]. Moreover, miR10a acts as an oncogenic driver in glioma cells, and the miR10a/MTMR3/β-catenin axis downregulates transcription factors as C-myc, resulting in inhibiting the proliferation of glioma cells [18]. Additionally, it has been delineated to be a key modulator of PC's metastatic nature [19]. In pancreatic juice samples, it was shown that miR-10a-5p appeared to be a biologically valuable biomarker of intraductal papillary mucinous neoplasm [20], while the exact molecular mechanism and diagnostic utility of miR-10a-5p as a serum tumour marker for PC are still under interrogation. Remarkably, it was revealed that lncRNA MIR22 host gene (MIR22HG) acts as a sponge of miR-10a-5p and increases nuclear receptor co-repressor2 (NCOR2) expression, which downregulates β-catenin and E-cadherin levels in hepatocellular carcinoma (HCC) tissues; thus it has suggested that MIR22HG/miR-10a-5p/NCOR2/β-catenin axis could repress proliferation, invasion, and migration of HCC [21]. Also, it participates as a tumor suppressor in colorectal cancer (CRC) [22]. However, in certain types of cancer, as in esophageal adenocarcinoma, it can act as an oncogenic driver [23]. Wnt/β-catenin signalling participates in PC progression by increasing cell proliferation, survival, and epithelial-mesenchymal transition (EMT), all crucial to metastasis [24]. It was reported that the persistent induction of β-catenin is critical for pancreatic intraepithelial neoplasia (PanIN) growth, which serves as a precursor to pancreatic carcinoma [25]. Elevated cytoplasmic and nuclear β-catenin levels are demonstrated in these lesions, implying their contribution in the early stages of carcinogenesis in pancreatic ductal adenocarcinoma (PDAC) tissues. β-catenin is particularly localized in the nucleus and cytoplasm, in contrast to its existence at the membrane in normal pancreatic cells [25,26]. From the data mentioned earlier, we speculated that MIR22HG/miR-10a-5p could affect β-catenin expression levels, inhibiting the proliferation and invasion of different kinds of cancer tissues. Nonetheless, the serum expression levels of the MIR22HG/miR-10a-5p/Wnt/β-catenin axis in PC remains unexplored. This study aimed to fill the current knowledge gap by exploring the serum expression levels of lncRNA MIR22HG and miR-10a-5p and their downstream targets in PC patients. Besides, their association with clinicopathologic features and their utility as biomarkers for diagnosis, prognosis, and prediction of PC were clarified. Additionally, we investigated the Wnt/β-catenin signaling axis, offering further insights into its possible use as an indicator for PC detection and prognosis. Patients and methods Study participants A case-control study was conducted, including 90 participants: 65 treatment-naïve PC cases and 25 healthy controls. All participants, aged over 35 years, attended the Gastrointestinal Endoscopy Unit at Kasr Al-Ainy Hospital, Cairo University, between May and October 2024, presenting with non-specific GI issues such as nausea, vomiting, abdominal discomfort, as well as severe symptoms in some cases, including sudden weight loss and jaundice. The PC cases' age mean ± standard deviation (SD) was 53 ± 8 years. PC diagnosis and disease staging were determined using a combination of endoscopic ultrasound findings, computed tomography (CT) scan results, and histopathological biopsies. [27]. PC staging was conducted adhering to the American Joint Committee on Cancer (AJCC) criteria [28]; notably, stage IV is characterized by distant metastasis. According to Hruban and Fukushima (2007), histologic cancer grading was evaluated [29]. All clinical routine data for patients were collected from medical record documents, including their complete medical history, blood count, biochemical liver profile, and demographic data such as gender and age. Patients were excluded from the study if they had undergone prior treatments (chemotherapy, radiotherapy, immunotherapy, or surgical tumor resection, or had been previously diagnosed with chronic diseases or disorders concurrent with PC. Smokers and individuals with a history of alcohol consumption were also excluded. On the other hand, healthy control individuals were randomly selected during routine physical examinations at the same hospital. Their test results and endoscopic evaluations were negative, and they were matched to the patient cases based on age and gender. This study was approved by the Ethics Committee of the Faculty of Pharmacy, Cairo University, under approval number BC3571, and conducted in compliance with the ethical principles established in the Declaration of Helsinki. Also, all methods were implemented in compliance with relevant regulations and guidelines. Formal informed consent was provided from all patients and controls or their legal representatives. Sampling Venous blood samples (5 mL) were provided from both controls and PC cases using clot activator polymer gel vacutainers. The samples were left to clot for 30 min before centrifugation at 8000 rpm for 10 min to isolate the serum from the coagulated blood. The resulting serum was then divided into 2 aliquots: one designated for extracting RNA and the second was assigned for the E-cadherin, CA19.9, and CEA levels assay. All aliquots were stored at −80 °C until further analysis. Selection of MIR22HG/miR-10a-5p and target genes (β-catenin and C-myc) co-expression network According to the miRWalk database v3 (http://mirwalk.umm.uni-heidelberg.de/), miR-10a-5p was recognized as a candidate target of the lncRNA MIR22HG. The analysis revealed that the CoDing Sequence (CDS) region of MIR22HG is the predicted binding site for miR-10a-5p, consistent with data obtained from RNA22 V2 [30]. Moreover, an experimental study by Wu et al. has confirmed the interaction between MIR22HG and miR-10a-5p in HCC [31]. The Human MicroRNA Disease Database v4 (http://www.cuilab.cn/hmdd) further indicated an association between miR-10a-5p and pancreatic carcinoma, supported by experimental evidence in both cell lines and tissue samples [32]. Additionally, the expression levels of β-catenin fluctuated based on miR-10a-5p levels in different cancer types [18,33]. Using the miRDB database v6.0 (https://mirdb.org/cgi-bin/target_detail.cgi?targetID=481490), β-catenin, with gene symbol catenin beta interacting protein 1 (CTNNBIP1), was predicted to be a downstream target of miR-10a-5p, with a target score of 61 and seed location 511. This interaction has been experimentally validated in cell lines [34]. Furthermore, miR-10a-5p was shown to target several metastasis-related genes, leading to the upregulation of nuclear β-catenin and C-myc expressions in both cell lines and tissue samples [35]. Differential expression investigation RNA-seq data for pancreatic adenocarcinoma (portal ID: TCGA-PAAD) were retrieved from The Cancer Genome Atlas (TCGA) using the TCGAbiolinks package (v2.26) in R (v4.2.1) [36]. To facilitate differential expression analysis, RNA-seq data from normal pancreatic tissues were provided from the Genotype-Tissue Expression (GTEx) portal. The analysis was conducted utilizing the DESeq2 package [37]. Due to the unavailability of hsa-miR-10a-5p expression data in TCGA, miRNA expression levels were evaluated using the GSE226762 dataset [38]. Methods Total RNA extraction and purity evaluation Utilizing the miRNeasy Mini Kit (Cat. No. 217004; Qiagen, Hilden, Germany), total RNA was extracted from 200 μL of hemolysis-free serum in compliance with the manufacturer's instructions. Additionally, the purity and concentration of the isolated RNA were assessed employing the Nanodrop® 2000c spectrophotometer (Thermo Scientific, Wilmington, USA). RNA samples with a purity ratio (A260/A280) between 1.9 and 2.1 were deemed suitable and used for cDNA synthesis targeting the selected lncRNA and miRNAs and related (messenger RNAs) mRNAs. Assessment of serum MIR22HG and miR-10a-5p, C-myc, and β-catenin using RT-qPCR After accomplishment of reverse transcription employing the miScript II RT kit (Qiagen, Valencia, CA, USA) for RNA samples in an ultimate volume of 20 μL RT reactions, adhering to the directions supplied by the manufacturer, the expression levels of lncRNA MIR22HG were analyzed utilizing glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as the reference housekeeping gene for normalization. Custom-designed primers (Invitrogen), created through the Primer3web software tool (https://primer3.ut.ee/), were utilized for the analysis. Primer specificity was confirmed prior to synthesis using the NCBI Primer-BLAST tool (https://www.ncbi.nlm.nih.gov/tools/primer-blast/). The primer sequences for MIR22HG were (F; 5′-CCTGTATCTGAGACCCTTAG-3′, R; 5′-AGTAGCAGCAGTTAGAATCC-3′) and GAPDH were (F; 5′-CCCTTCATTGACCTCAACTA-3′, R; 5′-TGGAAGATGGTGATGGGATT-3′). Quantitative PCR (qPCR) was conducted on the Qiagen Rotor-Gene Q System utilizing 20 μL reaction mixtures developed with the Maxima SYBR Green PCR kit (Thermo Scientific, USA), in combination with the forward and reverse primers. The thermal cycling protocol included an initial denaturation for 10 min at 95 °C, 45 amplification rounds of denaturation at 95 °C for 15 s, and annealing/extension at 56/72 °C subsequently for 60 s. Additionally, miR-10a-5p expression levels were evaluated by qPCR utilizing Qiagen miScript SYBR Green PCR kit. The primers for miR-10a-5p (Cat. no. MS00031262) and the housekeeping control miRNA SNORD68 (Cat. no. MS00033712) were provided by Qiagen. SNORD68 has been extensively validated in multiple studies as a reliable internal control for miRNA normalization, demonstrating stable and consistent expression across serum samples from both patients and healthy controls [39,40]. This validation supports its utility as a reference gene for the relative quantification of miRNAs. Shortly, RT-PCR was conducted using 20 μL reaction mixtures developed following the established protocols. The reactions were applied on the Qiagen Rotor-Gene Q system under the subsequent circumstances of thermal cycling: an initial step at 95 °C for 30 min, 40 cycles of 15 s at 94 °C, and annealing/extension at 55/70 °C subsequently for 30 s. For the evaluation of C-myc and β-catenin expression levels, the QIAGEN OneStep RT-PCR Kit (Cat. No./ID: 210212; Qiagen), designed for reverse transcription and real-time PCR in a single step, was utilized. The reactions were applied using the QIAGEN Rotor-Gene Q Real-Time PCR System, strictly adhering to the manufacturer's protocol. The primer sequences for C-myc (F; 5′-CAGCTGCTTAGACGCTGGATTT-3′; R; 5′-ACCGAGTCGTAGTCGAGGTCAT-3′), β-catenin (F; 5′-AATACCATTCCATTGTTTGTGCAG-3′, R; 5′-AGCTCAACTGAAAGCCGTTT-3′) were utilized, subsequently. GAPDH was employed as the internal control for normalization of gene expression. Finally, melting curve assessment was implemented to establish the PCR products' validity. Gene expression levels were adjusted to the internal reference using the 2−ΔCt method. For relative quantification, the fold change in expression was determined by implementing the 2−ΔΔCt formula. Quantitative estimation of E-cadherin protein level by ELISA The human E-cadherin ELISA kit developed by Abcam (Trumpington, Cambridge, UK, Catalog number: ab233611) was utilized to quantify E-cadherin levels in the sera of study participants, as indicated by the manufacturer. Negative and positive controls were included, every sample was assessed in duplicate, and the variance in the repeatability was less than 10%. Assessment of serum CEA and CA19.9 Carcinoembryonic antigen and CA19.9 serum levels were assessed employing the chemiluminescence approach by the Access 2 Immunoassay System (Beckman Coulter, Brea, CA, USA). Sample size computation To evaluate the sample size required for assessing the studied indicators, comprising lncRNA MIR22HG, miR-10a-5p, and related targets such as C-myc, E-cadherin, and β-catenin, the statistical software G∗Power was employed. The computation incorporated several critical factors, including the effect size, which quantifies the variation in expression levels of the biomarkers between patient and control groups, and the type I error rate (α), adopted at 0.05, representing a 5% probability of falsely detecting a variation. The type II error rate (β) was set at 0.2, matching to a statistical power of 80%, ensuring an 80% likelihood of detecting a true effect if present. Additionally, the standard deviation of serum biomarker levels, derived from prior studies [[41], [42], [43]], was integrated into the calculation. These parameters were utilized to accurately estimate the sample size required to reliably detect significant variations in the expression of MIR22HG, miR-10a-5p, and the related targets. Statistical analysis GraphPad Prism 9.0 (San Diego, CA, USA) and SPSS version 25 (Chicago, IL, USA) were implemented for the data analysis. Qualitative data were presented as frequencies and percentages, while quantitative variables were compiled employing median with interquartile range (IQR) or mean ± standard deviation (SD). Fischer's exact test was implemented to compare categorical data. The variables' normality was checked employing Kolmogorov–Smirnov, D'Agostino, and Pearson tests. For non-normally distributed variables, comparisons were implemented employing the Mann-Whitney or Kruskal-Wallis tests, with subsequent Dunn's test applied when necessary. Additionally, the student's t-test was implemented for contrasts of variables with a normal distribution. To examine relationships between variables, Spearman correlation coefficients were calculated. The diagnostic accuracy of studied parameters was investigated using receiver operating characteristic (ROC) curve analysis, with the area under the curve assessed [44]. Significant variables identified in univariate analysis were incorporated into a stepwise forward multivariate model to evaluate the final predictors for the likelihood of PC. Lastly, a p-value <0.05 was adopted for the statistical significance level. Our work complies with the Reporting Recommendations for Tumor MARKer Prognostic Studies (REMARK) prerequisites [45]. Results Study groups' demographic, laboratory, and clinicopathological data A comparative analysis of demographic, laboratory, and clinicopathological characteristics of patients with PC and healthy controls was displayed in Table (1). The mean age of PC patients was 53 ± 8 years, compared to 50 ± 6.2 years in the control group, though this difference did not reach statistical significance (p = 0.087). The distribution of sex showed a higher proportion of males among PC patients (66.15%) compared to controls (56%), but this difference was also not statistically significant (p = 0.465). Table 1Demographic and clinicopathological variables of PC cases and control groups.Table 1Demographic dataPC cases (n = 65)Controls (n = 25)PAge at diagnosis (years)53 ± 850 ± 6.20.087Age range(38–64)(40–61)Sex n (%)male43 (66.15%)14 (56%)0.465female22 (33.85%)11 (44%)Hb (g/dL)11.4 (10.1–12.5)12 (11–12.7)0.133TLC × 103/mm37.8 ± 26.8 ± 1.70.027Platelet count × 103/mm3275 ± 98.2284 ± 58.40.652INR1.1 (1–1.2)1 (0.97–1.1)<0.0001ALT (IU/L)35 (29–84)13 (10–17)<0.0001AST (IU/L)40 (28–82)16 (12–21)<0.0001AFP (ng/mL)3.3 (1.7–7.8)1.3 (0.55–1.9)<0.0001Albumin (g/dL)3.6 (3.3–3.8)4.5 (4.3–4.6)<0.0001Creatinine (mg/dL)0.90 (0.60–1)0.69 (0.60–0.93)0.027CA19.9 (U/mL)163 (46–573)12 (9.2–18)<0.0001CEA (ng/mL)2.9 (1.6–4.7)1.2 (0.65–1.9)<0.0001Clinicopathological dataTNM stagesCurative stages (ll + lll)56 (86.15%)N/AMetastatic stage (IV)9 (13.85%)pT stagesT229 (44.6%)N/AT316 (24.6%)T420 (30.8%)pN stagesN114 (21.5%)N/AN251 (78.5%)Tumor gradeG252 (80%)N/AG313 (20%)Lymphatic metastasisPresent65 (100%)N/AAbsent0 (0%)Distant metastasisPresent9 (13.85%)N/AAbsent56 (86.15%)AFP: Alpha-fetoprotein; ALT: Alanine transaminase; AST: Aspartate transaminase; CA19.9: Cancer antigen 19.9; CEA: Carcinoembryonic antigen; Hb: hemoglobin; INR: International normalized ratio; PC: Pancreatic cancer; pN: pathological node; pT: pathological tumor; TLC: Total leukocyte count; TNM: Tumor-node-metastasis. Data of the studied groups are compiled as median (IQR), mean ± SD, or number (percentage). Bold P values (p < 0.05) imply statistical significance. Demographic and clinicopathological variables of PC cases and control groups. AFP: Alpha-fetoprotein; ALT: Alanine transaminase; AST: Aspartate transaminase; CA19.9: Cancer antigen 19.9; CEA: Carcinoembryonic antigen; Hb: hemoglobin; INR: International normalized ratio; PC: Pancreatic cancer; pN: pathological node; pT: pathological tumor; TLC: Total leukocyte count; TNM: Tumor-node-metastasis. Data of the studied groups are compiled as median (IQR), mean ± SD, or number (percentage). Bold P values (p < 0.05) imply statistical significance. Notably, PC patients exhibited significantly elevated serum levels of ALT, AST, CA19.9, CEA, and AFP (p < 0.0001), as well as lower albumin levels and higher INR values in comparison to controls, indicating substantial alterations in liver functions and tumor markers. Additionally, serum creatinine levels and total leukocyte count (TLC) were notably higher in the PC cases when contrasted to controls (p = 0.027 for both). Clinically, a majority of PC patients were diagnosed at advanced stages (stage III and IV, 86.15%), with most tumors being moderately differentiated (80%). All PC cases showed lymphatic metastasis, and 13.85% had distant metastases. Serum expression levels of MIR22HG, miR-10a-5p, and related targets in PC patients vs. healthy controls A significant downregulation of MIR22HG in the PC cases compared to controls (median [IQR] fold change: 0.0043 [0.0022–0.0091] vs. 1.22 [0.35–2.53]; p < 0.0001, Fig. 1A) whereas a marked upregulation of miR-10a-5p in PC patients (202 [26.52–414.4] vs. 0.63 [0.54–2.20]; p < 0.0001, Fig. 1B). Besides, a significant downregulation of β-catenin and C-myc in the PC group (β-catenin: 0.0046 [0.0007–0.149] vs. 1.3 [0.47–3.24], p < 0.0001, Fig. 1C; C-myc: 0.22 [0.047–1.1] vs. 1.80 [0.46–1.84], p = 0.0017, Fig. 1D). Comparative serum expression levels of MIR22HG, miR-10a-5p, and related targets in PC cases and controls. Variables were expressed as a box plot; the line inside the box displays the median; the box borders display the 25%–75%. p < 0.05 reflects statistical significance. Additionally, the upregulation of serum E-cadherin protein levels was observed in PC patients compared to controls (95 [87–107.5] vs. 88 [80.5–96.7]; p = 0.0083, Fig. 1E). Furthermore, a notable overexpression of classical tumor markers, CEA and CA19.9, was displayed in the PC group (CEA: 2.9 [1.6–4.7] vs. 1.2 [0.65–1.9], p < 0.0001, Fig. 1F; CA19.9: 163 [46–573] vs. 12 [9.2–18], p < 0.0001, Fig. 1G). Diagnostic reliability of the studied biomarkers regarding PC The precision of diagnosis of the studied biomarkers (MIR22HG, miR-10a-5p, β-catenin, C-myc, E-cadherin, CA19.9, and CEA) for PC was conducted using ROC curve analysis. Concerning ncRNAs, MIR22HG and miR-10a-5p exhibit the highest diagnostic performance with AUCs of 0.95 and 0.97, sensitivity of 93.9% and 95.4%, and specificity of 88% and 92% at cutoff values of <0.21 and >3.05, respectively (Fig. 2A and B). Also, β-catenin had high diagnostic potential for PC with 0.93 AUC, 92% sensitivity, and 84.6% specificity at a cutoff value of <0.17 (Fig. 2C). In contrast, C-myc and E-cadherin demonstrated relatively lower diagnostic accuracy with AUCs of 0.71 and 0.68, sensitivity of 63.1% and 64.6%, and specificity of 92% and 60% at cutoff values of <0.40 and >89.25, respectively (Fig. 2D and E). On the other hand, PC standard biomarkers CA19.9 and CEA depicted strong diagnostic potential with AUCs of 0.86 and 0.83, respectively (Fig. 2F and G). Diagnostic accuracy of serum MIR22HG, miR-10a-5p, and related targets for PC using ROC curve analysis. AUC: Area under the curve. p < 0.05 means statistical significance. Integrative evaluation of the diagnostic performance of various models of the studied biomarkers for PC The diagnostic efficiency of combined biomarkers panels was summarized in Table (2). Notably, the combination of MIR22HG and miR-10a-5p demonstrated enhanced diagnostic performance, with an improved AUC for MIR22HG and specificity for both, compared to their individual application as singular markers. Moreover, panels B and C, which integrate MIR22HG and miR-10a-5p with CA19.9 or CEA, yielded significant improvements in the diagnostic efficacy of these conventional markers. Specifically, the inclusion of MIR22HG, β-catenin, and E-cadherin in panel D achieved an AUC of 0.94 with a sensitivity of 92.3% and specificity of 84%. In comparison to panel D, further enhancement was observed in panel E, where the addition of miR-10a-5p to the panel comprising MIR22HG and E-cadherin improved the AUC to 0.96 with a sensitivity of 95.4% and a specificity of 80%. Similar diagnostic improvements were achieved in panel G when CA19.9 was incorporated with MIR22HG and E-cadherin. These findings highlight the potential of integrating these biomarkers for more precise and reliable diagnostic assessments in PC. Table 2Diagnostic performance of some panels of the investigated serum biomarkers for PC.Table 2PanelAUCSensitivitySpecificityP95% CIA: MIR22HG + miR-10a-5p0.9793.8%100%< 0.010.932-1B: MIR22HG + miR-10a-5p + CA19.90.9793.8%100%< 0.010.938-1C: MIR22HG + miR-10a-5p + CEA0.9893.8%100%< 0.010.956-1D: MIR22HG + β-catenin + E-cadherin0.9492.3%84%< 0.010.890-0.993E: MIR22HG + E-cadherin + miR-10a-5p0.9695.4%80%< 0.010.916-0.996G: MIR22HG + CA19.9 + E-cadherin0.9695.4%80%< 0.010.916-0.996AUC: Area under the curve; CI: Confidence interval. Statistical significance is outlined as p < 0.05. Diagnostic performance of some panels of the investigated serum biomarkers for PC. AUC: Area under the curve; CI: Confidence interval. Statistical significance is outlined as p < 0.05. Logistic regression analysis for predicting the risk of pancreatic cancer To evaluate the independent effects of various parameters that predict the risk of PC, univariate logistic regression was applied. This analysis identified expression levels of MIR22HG and β-catenin as significant predictors, with both showing strong statistical significance (p < 0.0001) and odds ratios of 0.194 and 0.247, respectively, suggesting a protective association. In a more comprehensive analysis, forward stepwise multivariate logistic regression was employed to refine the predictors further. In this model, MIR22HG remained the only significant predictor for PC with a protective association (p = 0.006), highlighting its potential as a robust biomarker for assessing PC risk (Table 3). Additionally, CEA showed a strong association with elevated PC risk, revealing as a significant predictor in the multivariate model (p = 0.048) with a significantly higher OR of 5.40. Table 3Logistic regression analysis for predicting the pancreatic cancer risk.Table 3Parameter(β) CoefficientSEpOROR (95% CI)Univariate analysisMIR22HG−1.6420.414<0.00010.1940.086–0.436miR-10a-5p0.3550.1190.0031.4261.129–1.801E-cadherin0.060.0230.11.0621.014–1.111C-myc0.0790.0670.2401.0820.949–1.235β-catenin−1.3980.375<0.00010.2470.119–0.515CA19.90.0500.0170.0041.0511.016–1.088CEA1.3860.383<0.00013.9991.888–8.468Multivariate analysisMIR22HG−1.4150.5180.0060.2420.088–0.670CA19.90.07910.04210.06021.080.996–1.17CEA1.6870.5980.0485.401.67–17.46Constant1.360CI: Confidence interval; OR: Odds ratio. The stepwise multivariate logistic regression model yielded a log-likelihood of −17.966 and a −2 log-likelihood of 35.705, with an overall model p-value <0.0001, indicating a highly significant fit. p < 0.05 means statistical significance. Logistic regression analysis for predicting the pancreatic cancer risk. CI: Confidence interval; OR: Odds ratio. The stepwise multivariate logistic regression model yielded a log-likelihood of −17.966 and a −2 log-likelihood of 35.705, with an overall model p-value <0.0001, indicating a highly significant fit. p < 0.05 means statistical significance. Differential serum expression levels of MIR22HG, miR-10a-5p, and related targets β-catenin, C-myc, and E-cadherin in PC cases with various clinicopathological features The PC patients were subclassified utilizing T staging to determine the association between expression levels of MIR22HG, miR-10a-5p, and their related targets β-catenin, C-myc, and E-cadherin with cancer progression. In contrast to PC with the T2 stage, serum E-cadherin level was notably elevated in those with the T4 stage (P = 0.04), proving that the increase in serum E-cadherin levels is evidence of cancer aggressiveness. While the other studied parameters did not differ significantly between various T stages. Also, the cases were examined for regional lymph node invasion and subclassified based on their N-stage to elucidate the relationship between these investigated parameters and lymphatic metastases (Table 4). Serum expression levels of MIR22HG and β-catenin demonstrated a notable downregulation in PC with N2 stage when compared to those with N1 stage, with p-values of 0.047 and 0.004, respectively, indicating low expression levels of MIR22HG and β-catenin were associated with advanced lymphatic metastases. Table 4Serum levels of MIR22HG, miR-10a-5p and related targets β-catenin, C-myc, and E-cadherin in pancreatic cancer cases with various clinicopathological features.Table 4Clinicopathological featuresMIR22HG (fold change)miR-10a-5p (fold change)β-catenin (fold change)C-myc (fold change)E-cadherin (ng/mL)pT stagesT20.0044 (0.0022–0.009)124.5 (11–416)0.0004 (0.0007–0.133)0.209 (0.024–0.682)89.2 (84.3–96.4)T30.0035 (0.0026–0.0132)291.5 (187–564.3)0.027 (0.002–0.184)0.461 (0.04–4.09)98 (89.3–116.5)T40.0044 (0.0022–0.0091)149 (45–304)0.00012 (0.0005–0.133)0.231 (0.084–3.681)97.2 (89.1–134) aP0.9080.0600.1760.1830.029pN stagesN10.0052 (0.0040–0.0255)151 (10.6–298.1)0.133 (0.035–0.17)0.574 (0.025–2.79)96.3 (85–112.5)N20.0033 (0.0021–0.0066)201 (27.3–427.6)0.003 (0.0006–0.03)0.209 (0.059–0.96)95 (87–101)P0.0470.2430.0040.9720.859TNM stagesCurative (II & III)0.0047 (0.0023–0.0099)221 (30.4-426)0.0071 (0.0011–0.1574)0.215 (0.037–0.981)93.6 (86.5–97.2)Metastatic (IV)0.0023 (0.0017–0.0033)49.5 (22.6–149)0.0005 (0.0004–0.0008)0.198 (0.087–2.585)95 (86.4–113.5)P0.0240.0500.0010.8850.341Cancer gradesG20.0044 (0.0021–0.0094)176.8 (22.23–417.3)0.004 (0.0006–0.135)0.191 (0.032–0.952)94 (86.45–98.45)G30.0037 (0.0025–0.0090)203.7 (88.26–415.3)0.005 (0.002–0.17)0.361 (0.21–9.58)90.6 (84.85–223.4)P0.7180.5110.4480.0440.843G: Grade; pN: Pathological node; pT: Pathological tumor; TNM: Tumor-node-metastasis. Data are expressed as median (IQR). Bold P values (p < 0.05) imply statistically significant differences between two groups assessed using the Mann–Whitney U test, or among three groups assessed using the Kruskal–Wallis test followed by Dunn's post hoc test for pairwise multiple comparisons. a: A significant difference from PC cases with T2 stage at P < 0.05. Serum levels of MIR22HG, miR-10a-5p and related targets β-catenin, C-myc, and E-cadherin in pancreatic cancer cases with various clinicopathological features. G: Grade; pN: Pathological node; pT: Pathological tumor; TNM: Tumor-node-metastasis. Data are expressed as median (IQR). Bold P values (p < 0.05) imply statistically significant differences between two groups assessed using the Mann–Whitney U test, or among three groups assessed using the Kruskal–Wallis test followed by Dunn's post hoc test for pairwise multiple comparisons. a: A significant difference from PC cases with T2 stage at P < 0.05. Concerning the comparison of the PC cases with different TNM stages, the cases were subsequently divided into those in the metastatic (IV) and curative (II and III) stages to evaluate whether these investigated parameters have a diagnostic relevance for metastasis. Statistical findings demonstrated that the expression of MIR22HG and β-catenin was significantly downregulated in the metastatic group, with median fold changes of 0.0023 (p = 0.024) and 0.0005 (p = 0.001), respectively, when compared to the curative group. The serum CA19.9 levels were significantly upregulated in the metastatic group, with a median (IQR) fold change of 470 (428.5–2037) (p = 0.006). On the other hand, no notable difference was observed in metastatic cases concerning the serum levels of miR-10a-5p, C-myc, and E-cadherin. Among the studied parameters, serum C-myc levels were significantly elevated in PC cases with grade 3 in contrast to those with grade 2 (P = 0.044) (Table 4). Prognostic significance of the studied biomarkers regarding the presence of metastasis MIR22HG and β-catenin serum expression levels can distinguish between metastatic and curative PC cases with an AUC of 0.73, 0.82, sensitivity of 77.8%, 88.9%, and specificity of 67.9%, 78.6% at cutoff values < 0.0031, <0.00095, respectively (Fig. 3A–C). Serum CA19.9, as a standard biomarker for PC had the prognostic potential for metastasis with 0.78 AUC, 88.9 sensitivity, and 69.6% specificity at cutoff value > 391.5 U/mL (Fig. 3F). On the other hand, serum levels of miR-10a-5p, C-myc, E-cadherin, and CEA observed insignificant prognostic value for metastasis (Fig. 3B–D, E, G). Prognostic values of serum MIR22HG, miR-10a-5p, and related targets between metastatic vs curative PC patients. AUC: Area under the curve. p < 0.05 means statistical significance. Correlation between the investigated parameters and against the clinicopathological parameters within PC patients Serum MIR22HG observed significant positive correlations with serum levels of β-catenin (r = 0.47, p < 0.0001) and C-myc (r = 0.25, p = 0.039), whereas it exhibited a negative correlation with miR-10a-5p (r = −0.16, p = 0.21). Besides, a notable positive correlation was displayed between β-catenin and C-myc serum levels (r = 0.46, p < 0.0001). Moreover, serum miR-10a-5p exhibited a marked positive relationship with CEA (r = 0.29, p = 0.02) (Fig. 4 and Table 5). Correlation between the investigated parameters vs each other and against the clinicopathological parameters within PC patients. A correlation map with a blue-red (cold-hot) gradient is used, where blue reflects a correlation near 1, red signifies a relationship near −1, and white indicates a correlation close to 0. Correlations are calculated using Spearman's method. Table 5Spearman correlation between the different investigated parameters in the PC patients.Table 5VariablesrMIR22HGmiR-10a-5pβ-cateninC-mycE-cadherinCA19.9CEAMCancer stagesGMIR22HGr1−0.160.47∗0.25∗0.08−0.03−0.07−0.26∗−0.240.03miR-10a-5pr−0.1610.030.27∗0.120.150.29∗−0.24−0.070.08β-cateninr0.47∗0.0310.46∗0.06−0.24−0.090.39∗0.38∗0.10C-mycr0.25∗0.27∗0.46∗10.14−0.12−0.080.02−0.090.25∗E-cadherinr0.080.120.060.141−0.100.08−0.12−0.160.03CA19.9r−0.030.15−0.24−0.12−0.1010.54∗0.33∗0.27∗0.28∗CEAr−0.070.29∗−0.09−0.080.080.54∗10.05−0.050.33∗Correlation was done by Spearman correlation. G: Cancer grade; M: Distant metastasis; r: Spearman rho coefficient. ∗: reflects statistically significant at p < 0.05. Spearman correlation between the different investigated parameters in the PC patients. Correlation was done by Spearman correlation. G: Cancer grade; M: Distant metastasis; r: Spearman rho coefficient. ∗: reflects statistically significant at p < 0.05. Bioinformatics findings for the molecular interactions of miR-10a-5p with MIR22HG and β-catenin RNA22 V2 was utilized as a bioinformatic tool to identify miR-10a-5p binding sites and their corresponding heteroduplexes in MIR22HG and β-catenin, using 10 as a minimum number of paired-up bases in heteroduplex. Interestingly, miR-10a-5p was identified to bind with β-catenin in 22636 leftmost positions of the predicted target site (folding energy = −12.40 Kcal/mol, p = 0.0255; Fig. 5A) and MIR22HG in 4522 leftmost positions of the predicted target site (folding energy = −13.10 Kcal/mol, p = 0.0336; Fig. 5B). The molecular interactions of miR-10a-5p according to RNA22 V2. The predicted binding sequence of miR-10a-5p with β-catenin (A) and MIR22HG (B). Bioinformatics findings for differential expression of MIR22HG, miR-10a-5p, and some related targets in PC tissues The differential expression analysis identified significant alterations across all genes examined. E-cadherin and β-catenin exhibited significantly higher expressions in PC samples contrasted to normal tissues (p < 0.001). Conversely, MIR22HG and C-myc were significantly downregulated in PC samples (p < 0.001) according to TCGA and GTEx datasets (Fig. 6A). Additionally, miR-10a-5p expression levels were notably elevated in PC samples contrasted to normal tissues (p < 0.001) according to the GSE226762 dataset (Fig. 6B). Differential expressions of MIR22HG, β-catenin, C-myc, E-cadherin (TCGA/GTEx) and miR-10a-5p (GSE226762) for PC tissues. TCGA: The Cancer Genome Atlas; GTEx: Genotype-Tissue Expression. ∗∗∗: reflects statistically significant difference. Discussion The aggressiveness and devastating behavior of the PC and its heterogeneous complex environment compel researchers to dig deeper into the PC microenvironment [46]. Several reports have underscored genetic mutations as a cornerstone in pancreatic carcinogenesis, augmenting the risk of PC initiation and progression, but they cannot entirely account for the diverse behaviors associated with pancreatic malignancies [46]. Current therapeutic options are incompetent for early diagnosis and PC curing [47,48]. Consequently, new strategies and biomarkers are essential to unfold the complexity between genetic and epigenetic factors driving PC progression and improve long-term survival rates. Interestingly, various studies have investigated the serum ncRNAs as potential tumour markers for PC detection and prognosis [49,50]. Based on our information, the present study is the first to evaluate the relation between the serum levels of LncRNA MIR22HG and miR-10a-5p in PC patients. Many studies have underscored the regulatory role of MIR22HG in various cancers, including CRC, HCC, and glioblastoma [33,51,52]. While it has been shown to act as both an oncogene and a tumor suppressor, most current studies emphasize its tumour-suppressive role with its optimal diagnostic potential. In the present study, serum MIR22HG level demonstrated a significant downregulation in the PC cases compared to the control group and showed an AUC of 0.95 at a fold change cutoff value of <0.21 (Fig. 2A), suggesting its potential utility as a diagnostic indicator in the serum of PC cases and an appealing target for therapeutic modulators. Also, MIR22HG displayed a promising discriminative ability in lung adenocarcinoma (LUAD) tissues with an AUC value of 0.79, as reported by Su et al. [53]. Several studies reported that miR-10a and its close counterpart, miR-10b, are vital in enhancing metastasis and/or cellular invasion in pancreatic and breast cancers [19,54]. Besides, miR10a revealed a significant over-expression pattern in non-small cell lung cancer (NSCLC) and glioma tissues and cell lines and enhanced tumorigenesis through targeting phosphatase and tensin homolog (PTEN) and Wnt/β‐catenin signaling pathway, respectively [18,55]. Moreover, miR-10a-5p demonstrated overexpression in PC tissues and cell lines and enhanced cellular proliferation and invasion [56]. Accordingly, we hypothesized that miR-10a-5p functioned as oncomiR, contributing to tumor progression. However, miR-10a-5p has been described as a malignant indicator of invasive intraductal papillary mucinous neoplasm (IPMN) in preoperative pancreatic juice samples, revealing significantly higher expression in IPMN associated with invasive carcinoma as compared to those with noninvasive [20]. The exact potential diagnostic accuracy of miR-10a-5p as a novel diagnostic marker in the serum of PC cases has not been investigated yet. In the current study, miR-10a-5p showed significant up-regulation, discerning PC cases from healthy controls (p < 0.0001). Also, we provided evidence that miR-10a-5p can act as a potential diagnostic biomarker in PC serum samples exhibiting (AUC = 0.97 at a cutoff value > 3.05, p < 0.0001; Fig. 2B). In harmony with this finding, serum miR-10a-5p showed a good discriminative ability for acute myeloid leukemia and NSCLC with an AUC of 0.83 and 0.709 reported by Zhi et al. [57] and Bao et al. [58] studies, respectively. Spearman correlation demonstrated a negative correlation between miR-10a-5p and MIR22HG (r = −0.16, p =0.21). Besides, bioinformatic analysis revealed that miR-10a-5p may interact with MIR22HG (Fig. 5B), which warrants future tissue-based studies of PC using experimentally validated techniques, such as luciferase reporter assays, to confirm this molecular interaction. These findings come in harmony with a previous study that indicated MIR22HG is inversely correlated with miR-10a-5p in HCC tissues (r = −0.365, p = 0.18) and acts as a competing endogenous RNA (ceRNA) for miR-10a-5p, resulting in growth inhibition by targeting NCOR2 [21]. Expression levels of serum MIR22HG observed a notable downregulation in advanced lymphatic metastasis, differentiating between pN stages, and in distant metastasis, differentiating between PC patients with stage IV from those with curative stages (II & III), with an AUC value of 0.73 (Fig. 3A), displaying that MIR22HG may act as a potential prognostic marker in the serum of PC cases. This finding aligns with previous studies, which have observed varied expression patterns of MIR22HG through different lymphatic and distant metastasis stages in tissues or cell lines of CRC, LUAD, and HCC [21,52,53]. Also, serum MIR22HG differs but does not reach a significance level regarding PC pathological grades, which is inconsistent with previous studies that revealed a significant association of MIR22HG expression levels with pathological grades in CRC and HCC [21,52]. The obtained sample type and cancer location could be the origin of this discrepancy. According to mounting data, miR-10a-5p is raised in a variety of solid tumors, including PC, and this upregulation is associated with poor survival, tumour metastasis, and a worse prognosis [[58], [59], [60], [61]]. However, we observed that there was no notable relationship between miR-10a-5p serum levels and aggressive clinical outcomes. This lack of difference may be attributed to the utilized sample type and complex regulatory mechanisms of miRNAs, which could vary depending on the tissue context and stage of disease progression. Wnt signaling pathway aberrations are widespread across multiple types of pancreatic cancer [62,63]. Nearly 65% of pancreatic cancers revealed canonical Wnt/β-catenin pathway activation [64]. Also, it has implied that the Wnt/β-catenin signaling axis endorses the progression and aggressiveness of PC [65]. Interestingly, some studies reported evidence that MIR22HG might act as a regulatory tumor suppressor by suppressing the EMT process via the modulation of the Wnt/β-catenin pathway in HCC and cholangiocarcinoma, reducing mRNA levels of β-catenin, C-myc, as downstream targets, and up-regulating protein levels of epithelial marker E-cadherin [21,66]. While our findings revealed that serum β-catenin and C-myc levels were significantly downregulated and positively correlated with MIR22HG, whereas serum levels of E-cadherin were markedly upregulated in PC cases compared to controls. These findings were incoherent with reported results in HCC and cholangiocarcinoma tissues, as when MIR22HG was up-regulated, diminishing mRNA levels of C-myc and β-catenin, and up-regulating E-cadherin [21,66]. Such contradictory results can be attributed to the type of samples involved in the study regarding β-catenin expression level tissues is often up-regulated and localized in cytoplasm and nucleus, associated with PC progression and poor prognosis, also provide mechanistic insights into tissues, implying its potential as a prognostic indicator for tumor aggressiveness [64], but in serum samples was downregulated as, which indicate systemic response to the tumor, in normal pancreas, β-catenin is predominantly located at the plasma membrane, where it facilitates cellular attachment through cadherins [26], which supports our findings. Although the observed relationships among MIR22HG, miR-10a-5p, and β-catenin are biologically plausible and supported by bioinformatic interaction predictions, the current study design is primarily correlative, especially in serum-derived analyses. Therefore, definitive causal regulatory interactions cannot be established. As Witwer noted, circulating serum biomarkers reflect systemic correlations that may not correspond to tissue-specific mechanisms [67]. Accordingly, future functional and mechanistic studies, comprising cellular and molecular validation experiments, are necessary to confirm direct regulatory pathways linking these molecules. One of the prominent hallmarks of PC is its early systemic dissemination and its peculiar local tumor progression [68]. Previous reports indicated that abnormal expression level of β-catenin correlates with tumor progression, metastasis, and worse prognosis across various cancer types, including PC [69], and this may be attributed to β-catenin's involvement in cellular adhesion and its impact on the EMT process [70,71]. In this study, low serum β-catenin level was notably associated with advanced lymphatic and distant metastases, distinguishing between metastatic and curative PC cases with an AUC of 0.82 (Fig. 3C), and showed a notable positive correlation with C-myc as its downstream target. Hence, this finding suggests that reduced serum β-catenin may act as a prognostic indicator for PC cases assessing disease progression. These results are compatible with previous reports, emphasizing that decreased expression of β-catenin correlated with tumor dedifferentiation, a crucial step in tumor invasion [72], and emerged β-catenin as a prognostic indicator for PDAC patients [73]. In the present study, the upregulation of serum E-cadherin protein level in the PC group showed no significant difference between the metastatic and curative stages. These findings are consistent with earlier studies that reported that aberrant cytoplasmic upregulation of E-cadherin signifies abnormalities in cell polarity and differentiation in PanIN lesions, which could increase cell growth and invasion [74,75]. Other studies revealed that soluble E-cadherin levels are upregulated in blood serum of breast cancer and CRC [76,77], with an insignificant relationship with lymph node metastasis and TNM staging in CRC. However, a significant correlation was displayed in breast cancer. Such contrast may be related to the difference in cancer type. Serum–tissue discrepancies in β-catenin and E-cadherin likely arise because circulating β-catenin levels do not necessarily reflect activation of the canonical intracellular Wnt signaling pathway. Pathway activation, typically assessed by nuclear β-catenin accumulation or induction of downstream target genes, is inherently a cell-specific event and is therefore not directly captured by systemic measurements [78]. Moreover, circulating E-cadherin more plausibly reflects proteolytic ectodomain shedding or vesicular export rather than membrane expression, representing its release into the circulation, a process documented across multiple cancer types and correlated with tumor burden and disease progression [79,80]. Additionally, it has been reported that circulating β-catenin and E-cadherin levels are likely indicators of extracellular shedding and systemic activities rather than a result of direct intracellular signaling pathways [79]. Although C-myc is a well-characterized oncogene involved in cell cycle regulation and tumorigenesis [81], its serum level did not exhibit a clear association with metastasis in this study. This may suggest that C-myc expression alone is not a reliable indicator of metastatic potential in PC. Indeed, the CA19.9 biomarker can't be used as a standalone diagnostic biomarker in evaluating a patient's condition, which may result in an incomplete or inaccurate assessment [8,82]. Therefore, it has been recommended to incorporate combined biomarkers for enhancing diagnostic accuracy in several studies. In this study, integrating complementary biomarkers into panels provided a synergistic approach that improves diagnostic performance for accurate PC detection and addresses their limitations when used alone [83]. For example, the combination of MIR22HG and miR-10a-5p (Table 2, panel A) demonstrated 0.97 AUC, 93.8% sensitivity, and 100% specificity, surpassing the diagnostic efficacy of these markers when used individually. This improvement can be attributed to their complementary roles in the molecular pathways associated with PC [21]. Furthermore, adding traditional markers like CA19.9 and CEA augmented diagnostic accuracy achieved AUC values of 0.97 and 0.98, respectively (Table 2, panels B and C), highlighting the benefit of integrating traditional biomarkers with emerging molecular markers. CA19.9, while widely used in clinical settings, suffers from low specificity, particularly in distinguishing PC from benign conditions [84]. Combining it with novel biomarkers like MIR22HG and miR-10a-5p addresses this limitation, as evidenced by the improved AUC, sensitivity, and specificity. Interestingly, the inclusion of MIR22HG, β-catenin, and E-cadherin (AUC = 0.94; Table 2, panel D) underscores the importance of markers associated with EMT in PC progression. Both proteins are critical regulators of cell adhesion and migration, processes that are frequently dysregulated in cancer [85]. Furthermore, Panel E, which incorporates miR-10a-5p, MIR22HG, and E-cadherin, demonstrated the highest diagnostic sensitivity (95.4%) among the evaluated panels. This finding supports the hypothesis that combining EMT-related markers with miRNAs enhances diagnostic precision [86]. The univariate logistic regression analysis provides an initial evaluation of each parameter's independent contribution to the risk of PC. Significant predictors such as MIR22HG and β-catenin were identified, with strong statistical associations (p < 0.0001) and ORs less than 1, indicating a potential protective effect against PC. This step is crucial for identifying individual biomarkers that could be relevant in PC diagnosis, but it doesn't account for the interactions or combined effects of multiple factors. The multivariate logistic regression analysis addresses these limitations by assessing the influence of several predictors simultaneously. In this analysis, MIR22HG remained a significant predictor (p = 0.006), suggesting its independent role in assessing PC risk, even when other variables are considered. This strengthens the evidence for MIR22HG as a potentially critical biomarker for PC diagnosis. Notably, β-catenin, which was significant in the univariate analysis, did not retain its significance in the multivariate model, indicating that its effect may be confounded by or dependent on other variables. Moreover, CEA emerged as a significant predictor in the multivariate model (p = 0.048) with a substantially higher OR of 5.40, pointing to a strong association with increased PC risk when other factors are controlled. This highlights how multivariate analysis can reveal the true impact of predictors, offering a more comprehensive understanding of the complex interactions between biomarkers in PC. The combination of univariate and multivariate analyses provides a layered approach for identifying and validating biomarkers for PC, with the multivariate analysis offering a more refined and reliable prediction by accounting for confounders and interactions among variables. In addition, these findings suggest that MIR22HG, alongside other biomarkers such as CEA and CA19.9, could play a crucial role in enhancing the predictive accuracy for PC diagnosis, emphasizing the need for further validation in larger cohort studies. These findings can be evidenced by previous studies in CRC and HCC logistic regression analysis, which displayed that lower expression of MIR22HG is an independent prognostic indicator for patient outcomes and is associated with poorer outcomes [87,88]. Based on bioinformatics analysis, which revealed mirrored differential expression patterns of the four studied genes in both PC tissue samples and investigated serum samples from PC cases in this study, MIR22HG and C-myc were notably downregulated. In contrast, miR-10a-5p and E-cadherin showed notably increased expression levels. These findings may support our hypothesis that MIR22HG and miR-10a-5p could serve as valuable biomarkers for PC diagnosis, necessitating larger cohorts to validate their diagnostic potential. Specifically, future investigations should aim to elucidate the precise molecular relationship between MIR22HG and miR-10a-5p, including their roles in the progression and pathophysiology of PC, to establish their utility in diagnostic and therapeutic applications. Conclusion This study sheds light on the promising role of serum lncRNA MIR22HG and miR-10a-5p as novel diagnostic indicators for PC. By combining MIR22HG and miR-10a-5p with traditional markers like CA19.9 or CEA, a more accurate and comprehensive diagnostic tool for PC detection could be achieved. Besides, the strong link between low serum levels of MIR22HG and β-catenin with advanced lymphatic and distant metastases offers valuable insights into prognosis, helping to predict how the disease will advance and potentially guiding treatment decisions. Moreover, our findings support the correlation between MIR22HG and the Wnt/β-catenin axis, particularly β-catenin and C-myc, highlighting their importance in cancer metastasis. However, further tissue-based studies are needed to validate mechanistic interpretations. Finally, this study paves the way for further research, hoping these findings will eventually be translated into real-world clinical practice, leading to earlier diagnoses and better outcomes for patients facing this challenging disease. CRediT authorship contribution statement Rowyda M. Salman: Writing – original draft, Visualization, Software, Project administration, Investigation, Formal analysis, Data curation, Conceptualization. Abdullah F. Radwan: Writing – review & editing, Writing – original draft, Validation, Methodology, Formal analysis, Data curation, Conceptualization. Olfat G. Shaker: Writing – review & editing, Software, Resources, Project administration, Investigation, Data curation, Conceptualization. Ahmed A. El-Husseiny: Writing – review & editing, Writing – original draft, Visualization, Validation, Supervision, Data curation, Conceptualization. Ethics approval and consent to participate This study was approved by the Ethics Committee of the Faculty of Pharmacy, Cairo University, under approval number BC3571, and conducted in compliance with the ethical principles established in the Declaration of Helsinki. Also, all methods were implemented in compliance with relevant regulations and guidelines. Formal informed consent was provided from all patients and controls or their legal representatives. Availability of data and materials The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Funding This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of competing interest The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Acknowledgements The author expresses sincere appreciation to the members of the Gastrointestinal Endoscopy Unit, Kasr Al-Ainy Hospital, and Medical Biochemistry and Molecular Biology Department of the Faculty of Medicine, Cairo University, Cairo, Egypt, for cooperating and providing the lab equipment. Footnotes Peer review under the responsibility of Editorial Board of Non-coding RNA Research. List of abbreviations AbbreviationsFull nameAUCArea under the curveCA 19.9Cancer antigen 19.9CEACarcinoembryonic antigenCIConfidence intervalCRCColorectal cancerCTNNB1β-cateninEMTEpithelial-mesenchymal transitionGAPDHGlyceraldehyde 3-phosphate dehydrogenaseGIGastrointestinalGTExGenotype-Tissue ExpressionHCCHepatocellular carcinomaINRInternational normalized ratioIPMNIntraductal papillary mucinous neoplasmIQRInterquartile rangelncRNAsLong noncoding RNAsLUADLung adenocarcinomamiRNAsmicroRNAsmRNAMessenger RNANCOR2Nuclear receptor co-repressor 2ncRNAsNoncoding RNAsNSCLCNon-small cell lung cancerOROdds ratioPCPancreatic cancerPDACPancreatic ductal adenocarcinomapNPathological nodepTPathological tumorqPCRQuantitative PCRrCorrelation coefficientROCReceiver operating characteristicRTReverse transcriptionSDStandard deviationTCGAThe Cancer Genome AtlasTLCTotal leukocyte countTNMTumor-node-metastasis References Associated Data Data Availability Statement The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.