Stroma-associated extracellular vesicles snRNAs and piRNAs as non-invasive diagnostic and prognostic biomarkers in colon and pancreatic cancer

Stroma-associated extracellular vesicles snRNAs and piRNAs as non-invasive diagnostic and prognostic biomarkers in colon and pancreatic cancer | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Stroma-associated extracellular vesicles snRNAs and piRNAs as non-invasive diagnostic and prognostic biomarkers in colon and pancreatic cancer Manuel Collado, Víctor Medina-Chico, María E Castillo, Mercedes Herrera, and 14 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8948517/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Colon cancer (CC) and pancreatic ductal adenocarcinoma (PDAC) are among the most lethal gastrointestinal (GI) malignancies, largely due to limitations in early detection and risk stratification. While most circulating biomarkers currently under clinical evaluation are tumor cell–derived, the tumor stroma, and particularly cancer-associated fibroblasts (CAFs), represents an underexplored source of clinically informative signals. Methods: Based on our prior characterization of CAF-derived extracellular vesicles (EVs) cargo, we evaluated two small nuclear RNAs (snRNAs) and one PIWI-interacting RNA (piRNA) as EV-associated biomarkers in peripheral blood samples obtained prior to surgical intervention. EV RNA was isolated from serum samples of patients with colon pathology (including benign, premalignant, and malignant lesions), PDAC patients, and healthy donors. Small non-coding RNA (sncRNA) expression was quantified by RT–qPCR. Multiple machine learning models were applied to assess diagnostic and prognostic performance. The prognostic relevance of predicted target genes regulated by these sncRNAs was further evaluated in a large, publicly available CC cohort from The Cancer Genome Atlas. Results: The sncRNA panel discriminated CC patients from healthy individuals with high diagnostic accuracy (AUC = 0.866) and enabled discrimination of benign and premalignant colonic lesions. In PDAC, the same signature achieved robust diagnostic performance (AUC = 0.782), and combined analysis of CC and PDAC supported its ability to detect GI malignancy. In stage I–III CC patients, pre-operative EV sncRNA levels were associated with disease relapse, allowing stratification into low- and high-risk groups (AUC = 0.725; log-rank p = 0.009). Analysis of 43 predicted sncRNA target genes in a TCGA cohort of 597 CC patients revealed strong prognostic value for overall survival (AUC = 0.813; log-rank p < 0.001). Conclusions: CAF-associated EV snRNAs and piRNAs can be detected in peripheral blood and provide clinically relevant diagnostic and pre-operative prognostic information in CC and PDAC. By capturing stromal-derived signals distinct from tumor cell–intrinsic biomarkers, these molecules offer a complementary liquid biopsy strategy with potential utility for early detection, pre-surgical risk stratification, and personalized patient management. Together, our findings support the tumor stroma as a systemic source of accessible biomarkers and highlight CAF-associated EV sncRNAs as promising tools for improving clinical decision-making in GI cancers. Cancer-associated fibroblasts Extracellular Vesicles Small non-coding RNAs Liquid biopsy Colon cancer Pancreatic cancer Tumor microenvironment Risk stratification Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 HIGHLIGHTS Cancer-associated fibroblast–associated extracellular vesicles snRNAs and piRNAs are detectable in peripheral blood by liquid biopsy. A three-sncRNA signature discriminates healthy individuals from patients with benign colonic lesions, colon cancer, and pancreatic ductal adenocarcinoma, supporting broader applicability across gastrointestinal malignancies. Pre-operative extracellular vesicles sncRNA levels enable risk stratification and relapse prediction in stage I–III colon cancer. Target genes regulated by CAF-associated sncRNAs predict overall survival in a large public colon cancer cohort. INTRODUCTION Gastrointestinal (GI) cancers account for a substantial proportion of cancer incidence and mortality worldwide. In 2022 alone, more than five million new GI cancer cases were diagnosed globally, with close to 3.5 million associated deaths. Among these malignancies, colorectal cancer (CRC) ranks as the third most frequently diagnosed cancer and the second leading cause of cancer-related death, representing the most common GI tumor. Approximately two million new CRC cases and one million deaths are reported each year worldwide. Pancreatic ductal adenocarcinoma (PDAC), while less frequent and ranking twelfth in incidence, remains disproportionately lethal and is currently the sixth leading cause of cancer-related mortality. Together, CRC and PDAC account for nearly 2.5 million diagnoses and 1.5 million deaths annually ( 1 ). Early-stage CRC (stages I–II) is associated with a favorable prognosis, whereas survival declines sharply in advanced disease ( 2 ). Despite curative-intent surgery, nearly 30% of patients with stage I–III CRC experience disease recurrence ( 3 , 4 ). Notably, the risk of recurrence is highest during the first postoperative year and decreases thereafter, with 83% of recurrences occurring within three years of surgery ( 5 , 6 ). These observations underscore the clinical importance of diagnosing CRC at early stages, when tumors are more likely to be localized and associated with a lower risk of recurrence and improved long-term outcomes. In this context, population-based screening programs play a central role in enabling early detection before the onset of advanced disease. Current screening strategies include guaiac-based fecal occult blood testing (gFOBT), fecal immunochemical testing (FIT), and stool DNA tests, followed by colonoscopy when results are positive ( 7 ). Although these approaches are relatively inexpensive and easy to implement, their clinical application remains controversial. Their performance is limited in hospitalized patients, individuals with GI bleeding, or those receiving medications such as acetylsalicylic acid, anticoagulants, antiplatelet agents, non-steroidal anti-inflammatory drugs, selective serotonin reuptake inhibitors, or iron supplements ( 8 ). False-positive results frequently lead to unnecessary invasive procedures, increasing healthcare costs and exposing patients to risks including cardiopulmonary complications related to sedation, bleeding, perforation, and infection ( 9 – 11 ). Importantly, more than 40% of CRC-related deaths occur in patients older than 75 years, a population in whom invasive diagnostic procedures carry increased risk due to comorbidities and age-related vulnerability ( 12 ). In this context, no screening strategy has demonstrated clear superiority, and test selection must be individualized ( 13 ). Screening adherence also remains suboptimal and may benefit from the availability of blood-based biomarkers ( 14 ). PDAC is projected to surpass breast, prostate, and colorectal cancers to become the second leading cause of cancer-related death by 2030 ( 15 ). Pancreatic ductal adenocarcinoma (PDAC) accounts for approximately 90% of pancreatic cancer cases and is characterized by an extremely poor prognosis, with mortality rates closely mirroring incidence ( 16 , 17 ). Five-year survival currently exceeds only 12% ( 2 ), largely due to the absence of early symptoms, resulting in nearly 90% of tumors being diagnosed at advanced or metastatic stages ( 17 ). Surgical resection is feasible in only about 20% of patients and improves five-year survival to 15–25% in this subgroup ( 18 ). The relatively low incidence of PDAC, the lack of clearly defined high-risk populations, and the absence of non-invasive screening tools restrict screening programs to individuals with hereditary or familial risk. Available screening methods rely on imaging modalities such as endoscopic ultrasound (EUS), magnetic resonance imaging (MRI), magnetic resonance cholangiopancreatography (MRCP), computed tomography (CT), abdominal ultrasound, and positron emission tomography (PET), all of which are costly and limited by suboptimal sensitivity and specificity, preventing their application in the general population ( 19 , 20 ). Tumor development and progression depend not only on malignant epithelial cells but also on the surrounding tumor microenvironment (TME), which comprises cellular components, including endothelial cells, pericytes, immune cells, and fibroblasts, and non-cellular elements such as extracellular matrix, growth factors, and extracellular vesicles ( 21 , 22 ). Bidirectional communication between tumor cells and stromal components reshapes the TME and supports tumor growth, angiogenesis, and metastatic dissemination ( 23 ). Fibroblasts, the predominant stromal cell population, undergo phenotypic and functional changes upon interaction with tumor cells, giving rise to cancer-associated fibroblasts (CAFs) ( 21 , 22 , 24 ). Increased stromal content has been consistently associated with adverse outcomes in colon, breast, and esophageal cancers ( 25 – 27 ). Moreover, CAF-related gene signatures and markers have shown prognostic relevance across multiple tumor types, including colorectal, pancreatic, gastric, and lung cancers ( 28 – 32 ). Nevertheless, clinically applicable stroma-derived diagnostic biomarkers remain largely unavailable. Although CRC and PDAC are clinically distinct entities, they share important biological features. Both are epithelial adenocarcinomas arising within the gastrointestinal tract and are characterized by a prominent desmoplastic stroma enriched in cancer-associated fibroblasts. In particular, PDAC is well known for its highly fibrotic and CAF-dense microenvironment ( 33 ), while increasing evidence highlights the prognostic relevance of stromal content in colon cancer ( 34 ). This shared stromal architecture provides a biologically plausible rationale to explore whether stroma-derived molecular signals may represent common and systemically detectable biomarkers across these malignancies. In this context, extracellular vesicles (EVs) emerge as key mediators of intercellular communication within the TME and as potential carriers of stromal-derived signals into the systemic circulation. These vesicles, measuring approximately 40–150 nm, transport bioactive cargo, proteins, lipids, RNAs, and DNA fragments that reflects their cellular origin ( 35 , 36 ). Tumor-derived EVs contribute to CAF activation ( 37 – 39 ), while CAF-derived EVs, particularly through their non-coding RNA (ncRNA) content, enhance tumor cell proliferation, stemness, and chemoresistance ( 40 , 41 ). Our group previously demonstrated a distinct ncRNA cargo in CAF-derived versus normal fibroblast (NF)-derived EVs in colon cancer (CC), with enrichment of small nuclear RNAs (snRNAs), PIWI-interacting RNAs (piRNAs), and microRNAs (miRNAs) ( 42 ). We further showed that gene expression signatures derived from CAF-related ncRNAs predict prognosis in large cohorts of CC patients (N = 1,235) ( 43 ). These findings suggest that CAF-derived EV ncRNAs may represent clinically informative biomarkers. Based on this background, the present study was designed to: (i) evaluate whether two snRNAs (U1 and RNU1-11P) and one piRNA (piR-36249), previously identified as enriched in CAF-derived EVs and representing additional classes of small non-coding RNAs, can serve as diagnostic and prognostic biomarkers in colon pathology using peripheral blood samples obtained prior to surgical intervention; (ii) validate their diagnostic performance in an independent PDAC cohort; and (iii) assess the prognostic value of their target genes in a large, publicly available cohort of CC patients. By integrating stromal biology with liquid biopsy–based modeling strategies, this approach seeks to capture systemically detectable signals of tumor–stroma interaction at a preoperative stage, when clinically actionable information is often limited and largely dependent on postoperative pathological assessment. MATERIALS AND METHODS Patient samples and data collection Peripheral blood samples from CC patients were obtained at Ramón y Cajal University Hospital in collaboration with the Nursing Service. Blood samples from sporadic pancreas cancer patients, were collected from the Spanish Familial Pancreatic Cancer Registry (PANGENFAM)( 44 ). In addition, the Hospital Ramón y Cajal Biobank provided blood samples from healthy donors and from individuals diagnosed with colonic hyperplasia or adenoma, recruited through its established sample collections. Clinicopathological data were retrieved in collaboration with the Medical Oncology, Radiation Oncology, and Pathology Services of Ramón y Cajal University Hospital. All participants provided written informed consent prior to sample collection. The study protocol was approved by the local Ethics and Clinical Investigation Committee (Comité de Ética de la Investigación con Medicamentos del Hospital Universitario Ramón y Cajal)) and conducted in accordance with the Declaration of Helsinki. Blood samples were obtained prior to surgical intervention and subsequently processed and stored by the Biobank. Blood sample processing Peripheral blood samples were collected in red Vacutainer SST™ II Advance tubes (Becton Dickinson). Samples were processed under standardized conditions to obtain the serum fraction, which was aliquoted and stored at − 80°C until further use. Extracellular vesicles isolation and RNA extraction EVs were isolated from 250 µL serum aliquots using polymer-based precipitation with the ExoQuick™ Exosome Precipitation Solution (System Biosciences), following the manufacturer’s instructions. Total RNA was extracted from isolated EVs using the mirVana™ miRNA Isolation Kit (Ambion). To monitor extraction efficiency, the synthetic RNA spike-in UniSP2 was added to each sample together with MS2 carrier RNA, which improves spike-in stability without interfering with downstream applications. Purified RNA was eluted in nuclease-free water and stored at − 80°C until analysis. RNA reverse transcription Reverse transcription was performed using the miRCURY LNA™ RT Kit (Qiagen) according to the manufacturer’s protocol. The synthetic Caenorhabditis elegans miRNA cel-miR-39-3p (Qiagen) was added to each reaction as an internal control for reverse transcription efficiency. Complementary DNA (cDNA) was stored at − 20°C until quantitative PCR analysis. Detection of small non-coding RNAs by qPCR Quantitative PCR (qPCR) assays were performed to assess the efficiency of RNA extraction and reverse transcription using UniSP2 and cel-miR-39-3p controls, respectively, with miRCURY LNA™ miRNA primers (Qiagen). For detection of the small non-coding RNAs (sncRNAs) of interest (piR-36249, U1 and RNU1-11P), custom miRCURY LNA™ miRNA primers specific to each sequence were designed (Qiagen). cDNA samples were diluted 1:11 with nuclease-free water. qPCR reactions were carried out using miRCURY LNA™ SYBR Green Master Mix (Qiagen) on a LightCycler® 480 system (Roche). UniSP2 was used for data normalization, as it was added at the same concentration to all samples. Relative expression levels were calculated using the second-derivative method. Missing expression values due to low transcript abundance were imputed using half of the minimum detected value for the corresponding sncRNA. Predictive modeling and statistical analysis Multiple supervised machine learning algorithms were evaluated to identify the best-performing predictive models for each clinical endpoint and patient cohort. The following datasets were analyzed: Colonic pathology cohort: A total of 183 individuals, including 40 healthy donors, 14 patients with colonic hyperplasia, 19 with adenomas, and 110 with CC were used to analyze the expression of the three CAF-derived extracellular vesicles sncRNAs in peripheral blood. Pancreatic pathology cohort: A total of 172 individuals, including 40 healthy donors, 34 patients with familial PDAC, and 98 with sporadic PDAC were used to analyze the expression of the three CAF-derived extracellular vesicles sncRNAs in peripheral blood. Public CC cohort: A publicly available dataset of 597 CC patients from TCGA COADREAD cohort (TCGA; COADREAD Pan-Cancer Atlas, 2018) was analyzed to evaluate overall survival (OS) in relation to the predicted target genes of the studied sncRNAs ( 42 ). For each dataset, the performance of multiple classification algorithms was evaluated using a five-fold cross-validation approach. In this framework, the data were repeatedly partitioned into training (80%) and validation (20%) subsets across five iterations, such that each sample was used once for validation and multiple times for training. This strategy provides a robust and unbiased estimate of model performance while reducing the risk of overfitting. Given the presence of class imbalance in several cohorts, the Synthetic Minority Over-sampling Technique (SMOTE) was applied to balance the training data. All statistical analyses and machine learning procedures were implemented in Python, using standard scientific computing and machine learning libraries. Functional enrichment analyses were performed using the Gene Set Enrichment Analysis (GSEA) platform and the Molecular Signatures Database (MSigDB) ( https://www.gsea-msigdb.org/gsea/msigdb ), focusing on biological process gene sets. RESULTS Clinicopathological characteristics and EV sncRNA profiling of the study cohorts The cohort of healthy donors included a total of 40 individuals. The CC cohort comprised 110 patients, with a mean age of 74 years (range: 33–93), including 58 males and 52 females. Tumor location was right-sided in 47 patients (43%) and left-sided in 47 patients (43%). Four tumors were located at the rectosigmoid junction (3%), which were classified according to institutional anatomical and clinical criteria and were included in the colon cancer cohort. Information about tumor location was unavailable for 12 cases (11%). Regarding tumor stage, 22 patients were classified as stage I (20%), 35 as stage II (32%), 37 as stage III (34%), and 12 as stage IV (11%); staging information was missing for four patients (3%). During follow-up, disease relapse was observed in 22 patients (20%), whereas 88 patients (80%) remained relapse-free. For relapse prediction analyses, stage IV patients were excluded, resulting in a final dataset of 97 individuals, including 16 relapses (17%) and 81 non-relapses (83%). In addition, the study included 14 patients with colonic hyperplasia and 19 with adenomas, considered benign and premalignant lesions, respectively. The pancreatic pathology cohort consisted of 132 patients with PDAC. The selection of the small non-coding RNAs (sncRNAs) analyzed in this study was based on previous work from our group, which characterized the differential distribution of sncRNAs in cellular and extracellular vesicle (EV) fractions derived from normal fibroblasts (NFs) and CAFs ( 42 ). From this dataset, candidate molecules were prioritized according to the magnitude of deregulation between NFs and CAFs and their high read counts, ensuring both biological relevance and technical robustness. Based on these criteria, one piRNA (piR-36249) and two snRNAs (U1 and RNU1-11P) enriched in CAF-derived EVs were selected for evaluation. Correlation analysis among the three studied sncRNAs showed significant interdependence between variables (Supplementary Table S1 ). A summary of sncRNA expression levels across all study groups is provided in Supplementary Table S2. EV sncRNAs discriminate colon cancer from healthy individuals As an initial exploratory approach, Student’s t -tests were performed to compare healthy donors and CC patients. U1 and RNU1-11P expression levels showed statistically significant differences between groups, whereas piR-36249 did not (Fig. 1 A). To further evaluate diagnostic performance, multiple supervised classification algorithms were applied. A five-fold cross-validation scheme was used for both parameter optimization and performance evaluation. All tested models demonstrated robust discriminative performance, achieving areas under the receiver operating characteristic curve (AUCs) above 0.70. Detailed performance metrics, including accuracy and specificity, are summarized in Fig. 1 B. The Random Forest model outperformed other classifiers in terms of accuracy and AUC, leading to its selection for subsequent analyses. The average AUC across the five folds was 0.773, as illustrated by the ROC curve (Fig. 1 C). Predicted probabilities for individual samples are shown in Fig. 1 D, illustrating clear separation between healthy donors and CC patients. Variable importance estimates derived from the Random Forest model are presented in Fig. 1 E. Although logistic regression exhibited slightly lower performance compared with Random Forest, its interpretability warranted further consideration. Odds ratios derived from this model are shown in Fig. 1 F. Detection of benign and premalignant colonic lesions using EV sncRNAs Given the clinical relevance of early lesion detection, we next assessed whether the sncRNA panel could distinguish healthy donors from individuals with benign or premalignant colonic lesions (hyperplasia and adenoma). Student’s t -test analysis revealed a statistically significant difference for RNU1-11P expression, whereas no significant differences were observed for piR-36249 or U1 (Fig. 2 A). Subsequent machine learning analyses indicated that the Support Vector Machine (SVM) model achieved the best predictive performance in this setting, with an AUC of 0.792 (Fig. 2 B). The average AUC across the five folds using the Random Forest model was 0.764 (Fig. 2 C). Predicted probabilities for individual samples are shown in Fig. 2 D. Variable importance rankings from the Random Forest model and odds ratios derived from logistic regression are displayed in Figs. 2 E and 2 F, respectively. EV sncRNAs as diagnostic biomarkers in pancreatic cancer To evaluate whether the diagnostic potential of the selected sncRNAs extended beyond colorectal malignancy, their expression was analyzed in EVs isolated from peripheral blood samples of PDAC patients. Student’s t -tests demonstrated statistically significant differences for all three sncRNAs when comparing PDAC patients with healthy donors (Fig. 3 A). Following the same analytical workflow established for CC, classification models were evaluated and compared using a five-fold cross-validation approach. All evaluated algorithms achieved AUC values above 0.700 (Fig. 3 B). Random Forest again showed the best overall performance. The average AUC across the five was 0.775 (Fig. 3 C). Sample-level predicted probabilities are shown in Fig. 3 D, with clear separation between healthy individuals and PDAC patients. Variable importance scores and logistic regression-derived odds ratios are presented in Figs. 3 E and 3 F, respectively. EV sncRNA signature identifies gastrointestinal cancer To explore the ability of the sncRNA panel to detect GI malignancy more broadly, CC and PDAC patients were combined into a single cancer group and compared with healthy donors. All three sncRNAs showed significant differential expression between healthy individuals and cancer patients in Student’s t -test analyses (Fig. 4 A). Classification models evaluated via five-fold cross-validation consistently yielded AUC values exceeding 0.700 (Fig. 4 B). The average AUC across the five folds was 0.715 (Fig. 4 C). Predicted probabilities for individual samples are shown in Fig. 4 D. Variable importance rankings and odds ratios derived from Random Forest and logistic regression models are shown in Figs. 4 E and 4 F, respectively. Prognostic value of EV sncRNAs in colon cancer relapse We next assessed whether EV sncRNA expression levels were associated with relapse in CC patients. Student’s t -test analysis did not reveal significant associations between relapse status and expression of all three sncRNAs (Fig. 5 A). To develop a predictive model for three-year relapse, analyses were restricted to the 97 stage I–III CC patients included in the relapse cohort. The dataset comprised 81 non-relapsing patients and 16 patients who experienced relapse. Classification algorithms were evaluated using five-fold cross-validation, and performance metrics are shown in Fig. 5 B. Given its overall performance across evaluation metrics, the Random Forest algorithm was selected for further characterization. When assessed by five-fold cross-validation, the model achieved a modest mean AUC of 0.541 (Fig. 5 C), likely reflecting the limited number of patients experiencing relapse in this cohort. Notably, despite the reduced discriminative capacity, the model retained a relatively high specificity (0.690), indicating a consistent ability to correctly identify patients unlikely to relapse. Predicted probabilities were ranked and visualized to facilitate interpretation (Fig. 5 D). Variable importance estimates and odds ratios derived from Random Forest and logistic regression models are shown in Figs. 5 E and 5 F, respectively. To further stratify patients, a probability-based thresholding approach was applied. Patients were divided into low- and high-risk groups according to predicted relapse probability, and Kaplan–Meier survival analyses were performed iteratively to identify the optimal cut point. The most significant separation was obtained at cut point 52 (indicated by a dotted red line in Fig. 5 D), yielding a log-rank p -value of 0.009 (Fig. 5 G). The extracellular vesicles sncRNAs target gene signature predicts overall survival in public TCGA colon cancer cohorts To validate the biological and clinical relevance of the studied sncRNAs, we analyzed the expression of their predicted target genes, which were previously identified and functionally characterized by our group in CAF-derived EVs ( 42 ). In that study, a total of 64 target genes regulated by these sncRNAs were defined. Functional enrichment analysis of these genes revealed significant involvement in pathways related to cancer progression, including tyrosine kinase signaling, apoptosis, cell proliferation, and DNA repair (Supplementary Table S3). Based on this enrichment analysis, a prognostic gene signature comprising 43 genes was constructed, corresponding to those included in the ten most significantly represented biological process pathways (Supplementary Table S4). Gene expression and overall survival data were obtained from a TCGA CC cohort of 597 patients. Among these, 514 patients had overall survival exceeding 36 months, while 83 patients died within 36 months of diagnosis. After data preprocessing, 43 uncorrelated variables were retained. Their individual associations with survival were evaluated using Student’s t -tests (Supplementary Table S5). A five-fold cross-validation scheme was used for both parameter optimization and performance evaluation. Random Forest and SVM models achieved AUC values above 0.80, with specificities approaching 1, indicating a very high accuracy in identifying patients who will experience early death (Fig. 6 A). Although a small subset of patients who died within 36 months were classified as low risk, the near-perfect specificity indicates that patients classified by the model as high risk almost invariably experienced death within this time frame. This characteristic underscores the clinical value of the model for confidently identifying a subgroup of patients with particularly unfavorable prognosis. Using the Random Forest model, the mean AUC across the five folds was 0.813 (Fig. 6 B). Ranked predicted probabilities are shown in Fig. 6 C, while variable importance scores and logistic regression odds ratios are reported in Supplementary Tables S6 and S7. Using the same probability-based stratification approach described above, patients were divided into low- and high-risk groups for mortality. Kaplan–Meier analysis identified an optimal cut point at position 527 (Fig. 6 C), yielding a log-rank p -value < 0.001 (Fig. 6 D). DISCUSSION In this study, we identify a set of three sncRNAs (two snRNAs, U1 and RNU1-11P, and the piRNA piR-36249) as diagnostic and prognostic non-invasive biomarkers in two highly lethal GI malignancies, CC and PDAC. In addition, the expression of the target genes regulated by these sncRNAs in tumors showed strong prognostic value in a large, independent public CC cohort. Building on our previous work demonstrating that these sncRNAs are deregulated in CAFs and enriched in CAF-derived EVs, this study represents one of the first efforts to translate stroma-associated biomarkers into a non-invasive liquid biopsy setting. By detecting these sncRNAs in serum-derived EVs, our results highlight the clinical potential of tumor stroma–related signals for diagnosis and risk stratification using peripheral blood samples. Most circulating biomarkers currently used or under evaluation in GI oncology are derived directly from tumor cells. These include circulating tumor DNA (ctDNA) ( 45 ), tumor-derived EVs ( 35 ), serum protein biomarkers such as carcinoembryonic antigen ( 46 ) and CA19-9 ( 47 ), as well as circulating miRNAs ( 48 ). Although these markers have contributed substantially to clinical management, their performance in early-stage disease remains limited, and their specificity can be compromised in a range of clinical contexts ( 49 – 51 ). These limitations highlight the need for complementary biomarkers that capture biological processes not fully reflected by tumor cell–intrinsic signals. Within the tumor microenvironment, accumulating evidence indicates that CAFs actively shape tumor behavior through the transfer of ncRNAs via EVs. In both CC and PDAC, CAF-derived EV ncRNAs have been shown to enhance tumor cell stemness, invasiveness, metastatic potential, and resistance to chemotherapy ( 40 , 41 , 52 – 56 ). However, the majority of these studies have focused on miRNAs and long non-coding RNAs, which, despite extensive characterization, have demonstrated limited clinical translation. Challenges related to specificity, reproducibility, and the difficulty of isolating stromal-derived signals from circulation have hindered broader adoption. By contrast, other classes of sncRNAs, including snRNAs and piRNAs, remain comparatively understudied in the context of cancer biomarkers. snRNAs are core components of the spliceosome and play a central role in mRNA processing. Disruption of snRNA biogenesis and processing has recently been linked to cancer progression; notably, increased levels of unprocessed snRNAs resulting from Integrator complex dysfunction have been associated with poor prognosis in CRC ( 57 ). piRNAs, initially described in germ cells as regulators of transposable element activity, are now recognized as contributors to cancer stem cell maintenance and tumor progression ( 58 ). Several studies have suggested that piRNAs may outperform conventional biomarkers in certain oncological settings ( 59 ) yet their stromal origin and EV transport have received little attention. Consistent with these observations, our previous work identified enrichment of specific snRNAs and piRNAs in CAF-derived EVs and linked these molecules to distinct gene expression programs in CC ( 42 , 43 ). Here, we move beyond tissue-based analyses and demonstrate that these stroma-associated sncRNAs are not confined to the local tumor microenvironment but can be detected systemically in peripheral blood. Importantly, their ability to discriminate healthy individuals not only from patients with overt malignancy but also from those with benign and premalignant lesions suggests that they capture early stromal activation and tissue remodeling processes. While differences between early lesions and normal tissue are necessarily subtle, the observed discrimination supports their potential utility for disease detection at very early stages, a clinically relevant window in which intervention is more effective and long-term outcomes are improved. Beyond their diagnostic potential, the three-sncRNA panel also demonstrated clinically relevant prognostic value in CC. The association between sncRNA expression levels and relapse-free survival, together with the risk stratification achieved through probability-based modeling, supports the relevance of these markers beyond disease detection alone. This prognostic dimension was further supported by the analysis of the expression of the genes regulated by the studied sncRNAs, which were used to construct a transcriptional signature and evaluated in a large independent public CC cohort. Although derived from bulk tumor expression data and assessed using overall survival as the clinical endpoint, this gene signature enabled robust patient stratification and provided independent evidence of the biological and clinical relevance of the sncRNA-associated programs. Functional enrichment analyses revealed that these target genes are involved in key pathways related to tumor progression, including tyrosine kinase signaling, apoptosis, proliferation, and DNA repair, supporting a mechanistic link between stromal-derived sncRNAs and malignant behavior. Taken together, these findings support the concept that tumor stroma–associated biomarkers can provide actionable prognostic information that complements, and in certain clinical contexts may add value beyond, traditional tumor cell–derived biomarkers. In addition to their diagnostic and prognostic associations, the most clinically relevant implication of this work lies in the early identification of a subset of patients with particularly high risk. The high specificity achieved by the proposed models indicates that patients classified as high risk are identified with a high degree of confidence, enabling the recognition of individuals who are likely to experience early relapse or unfavorable outcome. Importantly, all analyses were performed using peripheral blood samples obtained prior to surgical intervention, meaning that this risk information is available before tumor resection and before definitive pathological staging. This preoperative, minimally invasive assessment represents a potential shift in clinical management, particularly in colon cancer, where robust molecular prognostic information is typically derived only from the surgical specimen. This capacity for early, pre-surgical risk stratification opens a window for therapeutic intensification or closer surveillance in selected patients, including consideration of more aggressive adjuvant strategies or tailored follow-up schedules. Significantly, rather than replacing established clinical or pathological criteria, this model provides complementary information derived from the tumor stroma. In this context, the integration of stroma-associated EV sncRNAs into clinical workflows could represent a practical and non-invasive tool to support decision-making and improve personalized management of CC patients. Importantly, the diagnostic potential of the sncRNA signature extended to PDAC, a malignancy for which early detection remains a major unmet clinical need. Despite the biological and clinical differences between colon and pancreatic tumors, the ability of the same sncRNA panel to discriminate PDAC patients from healthy individuals suggests that stroma-associated EV sncRNAs may capture shared stromal activation programs across GI cancers. The combined analysis of CC and PDAC further supports this concept and highlights the possibility of developing pan–GI cancer screening approaches based on stromal biomarkers. Several limitations of this study should be acknowledged. First, the retrospective, single-center design may introduce selection bias and limits generalizability. Second, the limited number of patients with colonic benign and premalignant lesions, as well as the relatively small number of relapses in the CC cohort, constrains model robustness and warrants validation in larger series. Third, the PDAC cohort lacked longitudinal follow-up data, precluding the assessment of prognostic performance in this setting. In addition, while previous experimental work supports the enrichment of the studied sncRNAs in CAFs and CAF-derived EVs, the precise cellular origin of circulating EVs in peripheral blood cannot be definitively established in this clinical setting. Finally, although the biological relevance of these sncRNAs is supported by their target gene analyses, additional functional studies will be required to further elucidate their mechanistic roles. Future work should therefore include prospective and multicenter validation and experimental approaches to refine both clinical applicability and biological interpretation. Despite these limitations, our data provide strong evidence that stroma-associated EV snRNAs and piRNAs can be detected in peripheral blood and carry clinically meaningful diagnostic and prognostic information. The consistent associations observed across CC and PDAC support the notion that stromal activation programs are systemically reflected and may be leveraged as accessible biomarkers in GI malignancies. From a biological perspective, targeting the tumor stroma rather than tumor cells alone offers a complementary view of disease behavior. Looking forward, the integration of stromal-derived EV sncRNAs into liquid biopsy strategies holds promise for broader clinical application, including early detection, risk stratification, and personalized patient management. Together, these findings highlight the tumor stroma as a rational and underexplored source of biomarkers with translational relevance in GI cancer. CONCLUSIONS This study demonstrates that a panel of three sncRNAs, two snRNAs and one piRNA, previously identified as enriched in CAF–derived EVs can be detected in peripheral blood and used as diagnostic and prognostic biomarkers in CC, with diagnostic applicability also demonstrated in PDAC. By capturing stroma-associated signals through a non-invasive liquid biopsy approach, these biomarkers provide clinically relevant information that complements conventional tumor cell-centered strategies. In CC, the sncRNA panel discriminated healthy individuals from patients with benign and premalignant lesions and overt malignancy and showed prognostic value for relapse within three years after surgery. Importantly, these molecular signals were obtained from peripheral blood samples collected prior to surgical intervention, highlighting their potential utility for pre-operative risk assessment and therapeutic decision-making. In addition, validation of the predicted target genes regulated by these sncRNAs in a large public CC cohort confirmed their strong association with overall survival, underscoring the biological relevance of the underlying stromal signaling programs. Importantly, the diagnostic potential of this sncRNA signature was not restricted to CC. The same panel identified PDAC patients with good diagnostic performance, supporting its applicability across distinct GI malignancies and reinforcing the concept that stromal activation programs may be shared across different tumor types within the GI system. Together, these findings support the notion that the tumor stroma is not only a key contributor to cancer progression but also a measurable systemic source of clinically informative biomarkers. CAF–associated EV snRNAs and piRNAs represent an underexplored class of circulating molecules with significant translational potential. Their integration into clinical workflows could improve early detection, pre-operative stratification, and patient management, particularly in clinical settings where current biomarkers show limited sensitivity or specificity. Further prospective and multicenter studies will be required to validate these results and to define the optimal clinical contexts for implementation. Nonetheless, this work provides a strong rationale for the development of stromal-based liquid biopsy strategies and highlights the relevance of stroma-associated EV sncRNAs as accessible and biologically meaningful tools in the management of GI cancers. Declarations FUNDING This research was supported by PI20/00602 from the Instituto de Salud Carlos III and co-financed by the European Development Regional Fund (FEDER) “A way to achieve Europe” (ERDF); “CIBER de Cáncer” (CB16/12/00273), and Biobank and Biomodels Platform PT20/0045 and PT23/00098 from the Instituto de Salud Carlos III - FEDER “A way to achieve Europe” (ERDF); CNS2023-144882 from Agencia Estatal de Investigación “Plan de Recuperación, Transformación y Resiliencia; Next Generation EU”; PID2022-136729OB-I00 funded by MICIU/AEI/10.13039/501100011033 and FEDER, UE”; and P2022/BMD7212, Comunidad de Madrid. MC was supported by FI21/00132 from Instituto de Salud Carlos III - FEDER “A way to achieve Europe” (ERDF). MJL and JMG-S belong to the Spanish National Research Council (CSIC)'s Cancer Hub. AUTHOR CONTRIBUTIONS MC, VMC, MEC, MH, CGP and CP made substantial contribution to conception and design of the work; the acquisition, analysis and interpretation of the data; and draft and revision of the work. MJL, CLF, LGB, FHdO, JMGS, AC and VGB have contributed to the conception and design of the work. IM, BM, EC, JV and CdlP contributed to the acquisition of the data. All authors reviewed the manuscript. ETHICS APPROVAL DECLARATION All participants provided written informed consent prior to sample collection. The study protocol was approved by the local Ethics and Clinical Investigation Committee (Comité de Ética de la Investigación con Medicamentos del Hospital Universitario Ramón y Cajal)) and conducted in accordance with the Declaration of Helsinki. ACKNOWLEDGMENTS The authors acknowledge the Hospital Universitario Ramón y Cajal Biobank for providing access to biological samples and for its essential contribution to sample processing and management. We also thank the biobank staff for their professionalism and technical support. ChatGPT was used for English text editing. We are grateful to lab members for help and advice throughout this research. CONFLICTS OF INTEREST The authors declare no conflict of interest. References Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin [Internet]. 2024 May [cited 2026 Jan 23];74(3):229–63. Available from: https://pubmed.ncbi.nlm.nih.gov/38572751/ Siegel RL, Miller KD, Jemal A, Cancer statistics. 2020. CA Cancer J Clin [Internet]. 2020 Jan [cited 2026 Jan 23];70(1):7–30. Available from: https://pubmed.ncbi.nlm.nih.gov/31912902/ Van Der Stok EP, Spaander MCW, Grünhagen DJ, Verhoef C, Kuipers EJ. Surveillance after curative treatment for colorectal cancer. Nat Rev Clin Oncol [Internet]. 2017 May 1 [cited 2026 Jan 23];14(5):297–315. Available from: https://pubmed.ncbi.nlm.nih.gov/27995949/ Boute TC, Swartjes H, Greuter MJE, Elferink MAG, van Eekelen R, Vink GR et al. Cumulative Incidence, Risk Factors, and Overall Survival of Disease Recurrence after Curative Resection of Stage II-III Colorectal Cancer: A Population-based Study. Cancer research communications [Internet]. 2024 Feb 29 [cited 2026 Jan 23];4(2):607–16. Available from: https://pubmed.ncbi.nlm.nih.gov/38363145/ Shi Q, Andre T, Grothey A, Yothers G, Hamilton SR, Bot BM et al. Comparison of outcomes after fluorouracil-based adjuvant therapy for stages II and III colon cancer between 1978 to 1995 and 1996 to 2007: evidence of stage migration from the ACCENT database. J Clin Oncol [Internet]. 2013 Oct 10 [cited 2020 Feb 5];31(29):3656–63. Available from: http://www.ncbi.nlm.nih.gov/pubmed/23980089 Elferink MAG, de Jong KP, Klaase JM, Siemerink EJ, de Wilt JHW. Metachronous metastases from colorectal cancer: a population-based study in North-East Netherlands. Int J Colorectal Dis [Internet]. 2015 Feb 1 [cited 2026 Jan 23];30(2):205–12. Available from: https://pubmed.ncbi.nlm.nih.gov/25503801/ Méndez G, Rivera-Matos L, Shuja A. Faecal occult blood testing: a review of its use and common misutilisation. BMJ Open Gastroenterol [Internet]. 2025 Jul 17 [cited 2026 Jan 23];12(1). Available from: https://pubmed.ncbi.nlm.nih.gov/40675631/ Fecal Occult Blood. Test - PubMed [Internet]. [cited 2026 Jan 23]. Available from: https://pubmed.ncbi.nlm.nih.gov/30725823/ Soin S, Akanbi O, Ahmed A, Kim Y, Pandit S, Wroblewski I et al. Use and abuse of fecal occult blood tests: a community hospital experience. BMC Gastroenterol [Internet]. 2019 Sep 3 [cited 2026 Jan 23];19(1). Available from: https://pubmed.ncbi.nlm.nih.gov/31481027/ Fisher DA, Maple JT, Ben-Menachem T, Cash BD, Decker GA, Early DS et al. Complications of colonoscopy. Gastrointest Endosc [Internet]. 2011 [cited 2026 Jan 23];74(4):745–52. Available from: https://pubmed.ncbi.nlm.nih.gov/21951473/ Coelho-Prabhu N, Forbes N, Thosani NC, Storm AC, Pawa S, Kohli DR et al. Adverse events associated with EGD and EGD-related techniques. Gastrointest Endosc [Internet]. 2022 Sep 1 [cited 2026 Jan 23];96(3):389–401.e1. Available from: https://pubmed.ncbi.nlm.nih.gov/35843754/ Colorectal Cancer — Cancer Stat Facts [Internet]. [cited 2026 Jan 23]. Available from: https://seer.cancer.gov/statfacts/html/colorect.html Udaikumar J, Nimmagadda R, Ingawale S, Lella VV, Vijayakumar K, Faye AS et al. Current Perspectives on Colorectal Cancer Screening and Surveillance in the Geriatric Population. Curr Gastroenterol Rep [Internet]. 2025 Dec 1 [cited 2026 Jan 23];27(1). Available from: https://pubmed.ncbi.nlm.nih.gov/40455318/ Shaukat A, Ladabaum U, Kanth P, Lieberman D. AGA Clinical Practice Update on Current Role of Blood Tests for Colorectal Cancer Screening: Commentary. Clinical Gastroenterology and Hepatology [Internet]. 2025 Aug 1 [cited 2026 Jan 23];23(9):1486–91. Available from: https://pubmed.ncbi.nlm.nih.gov/40267995/ Rahib L, Smith BD, Aizenberg R, Rosenzweig AB, Fleshman JM, Matrisian LM, pancreas cancers in the United States. Projecting cancer incidence and deaths to 2030: the unexpected burden of thyroid, liver, and. Cancer Res [Internet]. 2014 Jun 1 [cited 2026 Jan 23];74(11):2913–21. Available from: https://pubmed.ncbi.nlm.nih.gov/24840647/ Wood LD, Hruban RH. Pathology and molecular genetics of pancreatic neoplasms. Cancer J [Internet]. 2012 Nov [cited 2026 Jan 23];18(6):492–501. Available from: https://pubmed.ncbi.nlm.nih.gov/23187835/ Wood LD, Canto MI, Jaffee EM, Simeone DM. Pancreatic Cancer: Pathogenesis, Screening, Diagnosis, and Treatment. Gastroenterology [Internet]. 2022 Aug 1 [cited 2026 Jan 23];163(2):386–402.e1. Available from: https://pubmed.ncbi.nlm.nih.gov/35398344/ He J, Ahuja N, Makary MA, Cameron JL, Eckhauser FE, Choti MA et al. 2564 resected periampullary adenocarcinomas at a single institution: Trends over three decades. HPB [Internet]. 2014 [cited 2026 Jan 23];16(1):83–90. Available from: https://pubmed.ncbi.nlm.nih.gov/23472829/ Ballehaninna UK, Chamberlain RS. The clinical utility of serum CA 19 – 9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: An evidence based appraisal. J Gastrointest Oncol [Internet]. 2012 Jun [cited 2019 Nov 26];3(2):105–19. Available from: http://www.ncbi.nlm.nih.gov/pubmed/22811878 Lewis D, Jiménez L, Chan KK, Horton S, Wong WWL. A Systematic Review of Cost-Effectiveness Studies on Pancreatic Cancer Screening. Curr Oncol [Internet]. 2025 Apr 1 [cited 2026 Jan 23];32(4). Available from: https://pubmed.ncbi.nlm.nih.gov/40277782/ Kobayashi H, Enomoto A, Woods SL, Burt AD, Takahashi M, Worthley DL. Cancer-associated fibroblasts in gastrointestinal cancer [Internet]. Vol. 16, Nature Reviews Gastroenterology and Hepatology. Nature Publishing Group; 2019 [cited 2020 Oct 26]. pp. 282–95. Available from: https://pubmed.ncbi.nlm.nih.gov/30778141/ Kalluri R. The biology and function of fibroblasts in cancer [Internet]. Vol. 16, Nature Reviews Cancer. Nature Publishing Group; 2016 [cited 2021 Feb 24]. pp. 582–98. Available from: https://pubmed.ncbi.nlm.nih.gov/27550820/ Singh AJ, Gray JW. Chemokine signaling in cancer-stroma communications. J Cell Commun Signal [Internet]. 2021 Sep 1 [cited 2026 Jan 23];15(3):361–81. Available from: https://pubmed.ncbi.nlm.nih.gov/34086259/ Zhang W, Wang J, Liu C, Li Y, Sun C, Wu J et al. Crosstalk and plasticity driving between cancer-associated fibroblasts and tumor microenvironment: significance of breast cancer metastasis. J Transl Med [Internet]. 2023 Dec 1 [cited 2026 Jan 23];21(1). Available from: https://pubmed.ncbi.nlm.nih.gov/37978384/ Mesker WE, Liefers GJ, Junggeburt JMC, Van Pelt GW, Alberici P, Kuppen PJK et al. Presence of a high amount of stroma and downregulation of SMAD4 predict for worse survival for stage I-II colon cancer patients. Cell Oncol [Internet]. 2009 [cited 2026 Jan 23];31(3):169–78. Available from: https://pubmed.ncbi.nlm.nih.gov/19478385/ Downey CL, Simpkins SA, White J, Holliday DL, Jones JL, Jordan LB et al. The prognostic significance of tumour-stroma ratio in oestrogen receptor-positive breast cancer. Br J Cancer [Internet]. 2014 Apr 1 [cited 2026 Jan 23];110(7):1744–7. Available from: https://pubmed.ncbi.nlm.nih.gov/24548861/ He R, Li D, Liu B, Rao J, Meng H, Lin W et al. The prognostic value of tumor-stromal ratio combined with TNM staging system in esophagus squamous cell carcinoma. J Cancer [Internet]. 2021 Jan 1 [cited 2026 Jan 23];12(4):1105–14. Available from: https://pubmed.ncbi.nlm.nih.gov/33442408/ Zheng H, Liu H, Ge Y, Wang X. Integrated single-cell and bulk RNA sequencing analysis identifies a cancer associated fibroblast-related signature for predicting prognosis and therapeutic responses in colorectal cancer. Cancer Cell Int [Internet]. 2021 Dec 1 [cited 2026 Jan 23];21(1). Available from: https://pubmed.ncbi.nlm.nih.gov/34670584/ Wang F, Li Z, Xu T, Zhang Q, Ma T, Li S et al. A comprehensive multi-omics analysis identifies a robust scoring system for cancer-associated fibroblasts and intervention targets in colorectal cancer. J Cancer Res Clin Oncol [Internet]. 2024 Mar 1 [cited 2026 Jan 23];150(3). Available from: https://pubmed.ncbi.nlm.nih.gov/38478111/ Zhang J, Chen M, Fang C, Luo P. A cancer-associated fibroblast gene signature predicts prognosis and therapy response in patients with pancreatic cancer. Front Oncol [Internet]. 2022 Nov 18 [cited 2026 Jan 23];12. Available from: https://pubmed.ncbi.nlm.nih.gov/36465388/ Ucaryilmaz Metin C, Ozcan G. Comprehensive bioinformatic analysis reveals a cancer-associated fibroblast gene signature as a poor prognostic factor and potential therapeutic target in gastric cancer. BMC Cancer [Internet]. 2022 Dec 1 [cited 2026 Jan 23];22(1). Available from: https://pubmed.ncbi.nlm.nih.gov/35739492/ Si Y, Zhao Z, Meng X, Zhao K. RNA-seq and bulk RNA-seq data analysis of cancer-related fibroblasts (CAF) in LUAD to construct a CAF-based risk signature. Sci Rep [Internet]. 2024 Dec 1 [cited 2026 Jan 23];14(1). Available from: https://pubmed.ncbi.nlm.nih.gov/39369095/ Liot S, Balas J, Aubert A, Prigent L, Mercier-Gouy P, Verrier B et al. Stroma Involvement in Pancreatic Ductal Adenocarcinoma: An Overview Focusing on Extracellular Matrix Proteins. Front Immunol [Internet]. 2021 Apr 6 [cited 2026 Feb 18];12. Available from: https://pubmed.ncbi.nlm.nih.gov/33889150/ Qian C, Ren L, Zhu S, Chen F, Shen H, Hu X et al. Current perspectives and trends in colorectal cancer and cancer-associated fibroblasts: a review and bibliometric analysis. Front Immunol [Internet]. 2025 [cited 2026 Feb 18];16. Available from: https://pubmed.ncbi.nlm.nih.gov/40677714/ Kalluri R, LeBleu VS. The biology, function, and biomedical applications of exosomes. Science [Internet]. 2020 Feb 7 [cited 2026 Jan 23];367(6478). Available from: https://pubmed.ncbi.nlm.nih.gov/32029601/ Mathieu M, Martin-Jaular L, Lavieu G, Théry C. Specificities of secretion and uptake of exosomes and other extracellular vesicles for cell-to-cell communication. Nat Cell Biol [Internet]. 2019 Jan 1 [cited 2026 Jan 23];21(1):9–17. Available from: https://pubmed.ncbi.nlm.nih.gov/30602770/ Yoshii S, Hayashi Y, Iijima H, Inoue T, Kimura K, Sakatani A et al. Exosomal microRNAs derived from colon cancer cells promote tumor progression by suppressing fibroblast TP53 expression. Cancer Sci. 2019. Mito I, Takahashi H, Kawabata-Iwakawa R, Horikawa M, Ida S, Tada H et al. Tumor-derived exosomes elicit cancer-associated fibroblasts shaping inflammatory tumor microenvironment in head and neck squamous cell carcinoma. Oral Oncol [Internet]. 2023 Jan 1 [cited 2026 Jan 23];136. Available from: https://pubmed.ncbi.nlm.nih.gov/36462328/ Ringuette Goulet C, Bernard G, Tremblay S, Chabaud S, Bolduc S, Pouliot F. Exosomes Induce Fibroblast Differentiation into Cancer-Associated Fibroblasts through TGFβ Signaling. Mol Cancer Res [Internet]. 2018 [cited 2019 Sep 30];16(7):1196–204. Available from: http://www.ncbi.nlm.nih.gov/pubmed/29636362 Ren J, Ding L, Zhang D, Shi G, Xu Q, Shen S et al. Carcinoma-associated fibroblasts promote the stemness and chemoresistance of colorectal cancer by transferring exosomal lncRNA H19. Theranostics [Internet]. 2018 [cited 2019 Sep 30];8(14):3932–48. Available from: http://www.ncbi.nlm.nih.gov/pubmed/30083271 Hu JL, Wang W, Lan XL, Zeng ZC, Liang YS, Yan YR et al. CAFs secreted exosomes promote metastasis and chemotherapy resistance by enhancing cell stemness and epithelial-mesenchymal transition in colorectal cancer. Mol Cancer [Internet]. 2019 May 7 [cited 2020 Nov 11];18(1). Available from: https://pubmed.ncbi.nlm.nih.gov/31064356/ Herrera M, Llorens C, Rodríguez M, Herrera A, Ramos R, Gil B et al. Differential distribution and enrichment of non-coding RNAs in exosomes from normal and Cancer-associated fibroblasts in colorectal cancer. Mol Cancer [Internet]. 2018 Aug 3 [cited 2018 Oct 24];17(1):114. Available from: https://molecular-cancer.biomedcentral.com/articles/ 10.1186/s12943-018-0863-4 Herrera M, Berral-González A, López-Cade I, Galindo-Pumariño C, Bueno-Fortes S, Martín-Merino M et al. Cancer-associated fibroblast-derived gene signatures determine prognosis in colon cancer patients [Internet]. Vol. 20, Molecular Cancer. BioMed Central Ltd; 2021 [cited 2021 Jun 15]. Available from: https://pubmed.ncbi.nlm.nih.gov/33926453/ Mocci E, Guillen-Ponce C, Earl J, Marquez M, Solera J, Salazar-López MT et al. PanGen-Fam: Spanish registry of hereditary pancreatic cancer. Eur J Cancer [Internet]. 2015 Sep [cited 2019 Nov 26];51(14):1911–7. Available from: http://www.ncbi.nlm.nih.gov/pubmed/26212471 Wan JCM, Massie C, Garcia-Corbacho J, Mouliere F, Brenton JD, Caldas C et al. Liquid biopsies come of age: towards implementation of circulating tumour DNA. Nat Rev Cancer [Internet]. 2017 Apr 1 [cited 2026 Jan 23];17(4):223–38. Available from: https://pubmed.ncbi.nlm.nih.gov/28233803/ Hall C, Clarke L, Pal A, Buchwald P, Eglinton T, Wakeman C et al. A Review of the Role of Carcinoembryonic Antigen in Clinical Practice. Ann Coloproctol [Internet]. 2019 [cited 2026 Jan 23];35(6):294–305. Available from: https://pubmed.ncbi.nlm.nih.gov/31937069/ Ballehaninna UK, Chamberlain RS. Serum CA 19 – 9 as a Biomarker for Pancreatic Cancer-A Comprehensive Review. Indian J Surg Oncol [Internet]. 2011 Jun 1 [cited 2026 Jan 23];2(2):88–100. Available from: https://pubmed.ncbi.nlm.nih.gov/22693400/ Cui M, Wang H, Yao X, Zhang D, Xie Y, Cui R et al. Circulating MicroRNAs in Cancer: Potential and Challenge. Front Genet [Internet]. 2019 Jul 18 [cited 2026 Jan 23];10. Available from: https://pubmed.ncbi.nlm.nih.gov/31379918/ Oh HJ, Imam-Aliagan AB, Kim Y, Bin, Kim HJ, Izaguirre IA, Sung CK et al. Clinical applications of circulating biomarkers in non-small cell lung cancer. Front Cell Dev Biol [Internet]. 2024 [cited 2026 Jan 23];12. Available from: https://pubmed.ncbi.nlm.nih.gov/39239557/ Duffy MJ, Crown J. Circulating Tumor DNA as a Biomarker for Monitoring Patients with Solid Cancers: Comparison with Standard Protein Biomarkers. Clin Chem [Internet]. 2022 Nov 1 [cited 2026 Jan 23];68(11):1381–90. Available from: https://pubmed.ncbi.nlm.nih.gov/35962648/ Fiala C, Diamandis EP. Utility of circulating tumor DNA in cancer diagnostics with emphasis on early detection. BMC Med [Internet]. 2018 Oct 2 [cited 2026 Jan 23];16(1). Available from: https://pubmed.ncbi.nlm.nih.gov/30285732/ Zhang HW, Shi Y, Liu J, Bin, Wang HM, Wang PY, Wu ZJ et al. Cancer-associated fibroblast-derived exosomal microRNA-24-3p enhances colon cancer cell resistance to MTX by down-regulating CDX2/HEPH axis. J Cell Mol Med [Internet]. 2021 Apr 1 [cited 2026 Jan 23];25(8):3699–713. Available from: https://pubmed.ncbi.nlm.nih.gov/33621425/ Zhao J, Zhang J, Liu J, Zan L. Exosomal transfer of miR-223-3p from carcinoma-associated fibroblasts promotes the malignant properties and chemoresistance of colon cancer cells by targeting NF2/Hippo signaling. Med Oncol [Internet]. 2025 Nov 1 [cited 2026 Jan 23];42(11). Available from: https://pubmed.ncbi.nlm.nih.gov/41026317/ Bhome R, Goh RW, Bullock MD, Pillar N, Thirdborough SM, Mellone M et al. Exosomal microRNAs derived from colorectal cancer-associated fibroblasts: role in driving cancer progression. Aging [Internet]. 2017 [cited 2019 Sep 30];9(12):2666–94. Available from: http://www.ncbi.nlm.nih.gov/pubmed/29283887 Qi R, Bai Y, Li K, Liu N, Xu Y, Dal E et al. Cancer-associated fibroblasts suppress ferroptosis and induce gemcitabine resistance in pancreatic cancer cells by secreting exosome-derived ACSL4-targeting miRNAs. Drug Resist Updat [Internet]. 2023 May 1 [cited 2026 Jan 23];68. Available from: https://pubmed.ncbi.nlm.nih.gov/37003125/ Richards KE, Zeleniak AE, Fishel ML, Wu J, Littlepage LE, Hill R. Cancer-associated fibroblast exosomes regulate survival and proliferation of pancreatic cancer cells. Oncogene. 2017;36(13):1770–8. Escrich V, Romero-Aranda C, López R, de Toro M, Metola Á, Ezcurra B et al. Unprocessed snRNAs Are a Prognostic Biomarker and Correlate with a Poorer Prognosis in Colorectal Cancer. Cancers (Basel) [Internet]. 2024 Jul 1 [cited 2026 Jan 23];16(13). Available from: https://pubmed.ncbi.nlm.nih.gov/39001402/ Garcia-Borja E, Siegl F, Mateu R, Slaby O, Sedo A, Busek P et al. Critical appraisal of the piRNA-PIWI axis in cancer and cancer stem cells. Biomark Res [Internet]. 2024 Dec 1 [cited 2026 Jan 23];12(1). Available from: https://pubmed.ncbi.nlm.nih.gov/38303021/ Sohn EJ, Han ME, Park YM, Kim YH, Oh SO. The potential of piR-823 as a diagnostic biomarker in oncology: A systematic review. PLoS One [Internet]. 2023 Dec 1 [cited 2026 Jan 23];18(12). Available from: https://pubmed.ncbi.nlm.nih.gov/38060527/ Additional Declarations No competing interests reported. Supplementary Files ColladoMetalSupplementaryTables.ods Supplementary Table S1. Pairwise correlation analysis of EV sncRNA expression levels in all studied individuals. Pearson correlation coefficients and significance values for U1, RNU1-11P, and piR-36249. Supplementary Table S2. Distribution of EV sncRNA expression levels across all study groups. Summary statistics for EV sncRNA expression in healthy donors, patients with benign and premalignant colonic lesions, and colon and pancreatic cancer. Supplementary Table S3. Functional enrichment analysis of CAF-associated EV sncRNA target genes. 20 most significantly enriched Gene Ontology biological processes terms. Supplementary Table S4. List of predicted target genes regulated by CAF-associated EV sncRNAs used to construct the prognostic gene signature. Supplementary Table S5. Univariate association of sncRNA target gene expression with overall survival in the TCGA colon cancer cohort. Student’s t-test results for genes retained after preprocessing. Supplementary Table S6. Variable importance scores for sncRNA target genes derived from the Random Forest model predicting overall survival in the TCGA colon cancer cohort. Supplementary Table S7. Logistic regression odds ratios for sncRNA target genes associated with overall survival in the TCGA colon cancer cohort. ColladoMelalGraphicalAbstract.png Graphical abstract description Schematic representation of a non-invasive liquid biopsy strategy based on stroma-associated extracellular vesicles (EVs) small non-coding RNAs. Cancer-associated fibroblast–related EV snRNAs and piRNAs are detected in peripheral blood and analyzed using machine learning approaches to enable early diagnosis, risk stratification, and clinically relevant decision-making in colon and pancreatic cancer. Created with BioRender.com Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8948517","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":600690163,"identity":"e6bf246f-6d5a-4b49-b61f-dea040625abb","order_by":0,"name":"Manuel Collado","email":"","orcid":"","institution":"Instituto de Investigaciones Biomédicas Sols-Morreale","correspondingAuthor":false,"prefix":"","firstName":"Manuel","middleName":"","lastName":"Collado","suffix":""},{"id":600690166,"identity":"2718422e-04d3-449d-8ae2-7d70e259b641","order_by":1,"name":"Víctor Medina-Chico","email":"","orcid":"","institution":"Alfonso X El Sabio University","correspondingAuthor":false,"prefix":"","firstName":"Víctor","middleName":"","lastName":"Medina-Chico","suffix":""},{"id":600690167,"identity":"5db519c1-6bf1-46db-a893-2b6e9c28ebde","order_by":2,"name":"María E Castillo","email":"","orcid":"","institution":"University of Alcalá","correspondingAuthor":false,"prefix":"","firstName":"María","middleName":"E","lastName":"Castillo","suffix":""},{"id":600690168,"identity":"155b7255-0c80-44b0-9cff-ed3f6c2b6cb4","order_by":3,"name":"Mercedes Herrera","email":"","orcid":"","institution":"Karolinska Institutet","correspondingAuthor":false,"prefix":"","firstName":"Mercedes","middleName":"","lastName":"Herrera","suffix":""},{"id":600690169,"identity":"cdb22f3c-c99a-46ac-89eb-8abff4906a90","order_by":4,"name":"María Jesús Larriba","email":"","orcid":"","institution":"Instituto de Investigaciones Biomédicas Sols-Morreale","correspondingAuthor":false,"prefix":"","firstName":"María","middleName":"Jesús","lastName":"Larriba","suffix":""},{"id":600690170,"identity":"d771a720-95a7-4522-9360-04e95a2115b1","order_by":5,"name":"Cristina Galindo-Pumariño","email":"","orcid":"","institution":"Instituto Ramón y Cajal de Investigación Sanitaria","correspondingAuthor":false,"prefix":"","firstName":"Cristina","middleName":"","lastName":"Galindo-Pumariño","suffix":""},{"id":600690171,"identity":"32f32045-c304-4389-8acc-7adc174fd846","order_by":6,"name":"Carmen Lisset Flores","email":"","orcid":"","institution":"Instituto de Investigaciones Biomédicas Sols-Morreale","correspondingAuthor":false,"prefix":"","firstName":"Carmen","middleName":"Lisset","lastName":"Flores","suffix":""},{"id":600690172,"identity":"f09bded7-57ac-4c33-b4f4-63e8330c9a7d","order_by":7,"name":"Íñigo Martínez","email":"","orcid":"","institution":"Hospital Universitario Ramón y Cajal","correspondingAuthor":false,"prefix":"","firstName":"Íñigo","middleName":"","lastName":"Martínez","suffix":""},{"id":600690173,"identity":"fe8f2e9f-823d-4c84-ae7a-5dc4a0341a10","order_by":8,"name":"Blanca Morón","email":"","orcid":"","institution":"Hospital Universitario Ramón y Cajal","correspondingAuthor":false,"prefix":"","firstName":"Blanca","middleName":"","lastName":"Morón","suffix":""},{"id":600690174,"identity":"382983cd-ad41-4a2a-8cab-0696433be6c0","order_by":9,"name":"Elisa Conde","email":"","orcid":"","institution":"Instituto Ramón y Cajal de Investigación Sanitaria","correspondingAuthor":false,"prefix":"","firstName":"Elisa","middleName":"","lastName":"Conde","suffix":""},{"id":600690175,"identity":"8690735d-0da4-4cf2-9590-a5d4f756e8ea","order_by":10,"name":"Jorge Villalón","email":"","orcid":"","institution":"Instituto Ramón y Cajal de Investigación Sanitaria","correspondingAuthor":false,"prefix":"","firstName":"Jorge","middleName":"","lastName":"Villalón","suffix":""},{"id":600690176,"identity":"bbf327a3-0c76-4570-90aa-557e4c9311fd","order_by":11,"name":"Laura García-Bermejo","email":"","orcid":"","institution":"Instituto Ramón y Cajal de Investigación Sanitaria","correspondingAuthor":false,"prefix":"","firstName":"Laura","middleName":"","lastName":"García-Bermejo","suffix":""},{"id":600690177,"identity":"eb9e3577-f191-4fe2-a618-e10398aa3ff4","order_by":12,"name":"Félix Hernández-del-Olmo","email":"","orcid":"","institution":"National University of Distance Education","correspondingAuthor":false,"prefix":"","firstName":"Félix","middleName":"","lastName":"Hernández-del-Olmo","suffix":""},{"id":600690178,"identity":"e7936fa1-aae6-4741-a30b-7ca08107d390","order_by":13,"name":"José Manuel González-Sancho","email":"","orcid":"","institution":"Instituto de Investigaciones Biomédicas Sols-Morreale","correspondingAuthor":false,"prefix":"","firstName":"José","middleName":"Manuel","lastName":"González-Sancho","suffix":""},{"id":600690179,"identity":"8c17ed4f-44f8-4057-9608-0dd8511c2f4a","order_by":14,"name":"Carolina de la Pinta","email":"","orcid":"","institution":"Hospital Universitario Ramón y Cajal","correspondingAuthor":false,"prefix":"","firstName":"Carolina","middleName":"de la","lastName":"Pinta","suffix":""},{"id":600690180,"identity":"38af1c20-522b-42ce-aa8d-c164627e1618","order_by":15,"name":"Alfredo Carrato","email":"","orcid":"","institution":"University of Alcalá","correspondingAuthor":false,"prefix":"","firstName":"Alfredo","middleName":"","lastName":"Carrato","suffix":""},{"id":600690181,"identity":"4eaaa7ed-0d1b-4508-894b-c21fe0a77cfa","order_by":16,"name":"Vanesa García-Barberán","email":"","orcid":"","institution":"Hospital Clínico San Carlos","correspondingAuthor":false,"prefix":"","firstName":"Vanesa","middleName":"","lastName":"García-Barberán","suffix":""},{"id":600690182,"identity":"eb487fda-64af-4254-bde2-82af835527c1","order_by":17,"name":"Cristina Peña","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA70lEQVRIie2SMQrCMBRAfxHapeoaScErRAQ37VUsDi6Cjg4OgUBcFA+geAznSKBT8QQOiuDkUBFEQcHvYMfYUTBvC/zHfx8CYLH8JA4HYJ/HEMDNpahMSXIpiMp0mWO6PNfikg62VaAi3t2XrbDkKWd/MyhkE0mi2LHGg7hbm646kfTbhbpvUJiPMYpph5NegxRXqu0ScKkpDBWRohJy0r9WngsVouLdTWGocAzTEW5xaZErR+IWMIWRBGcSpjsyiBs0iN+3RIKalPLEO6TDh27OqDhWTqNWWB3r9dkUloEnfHj/h1yQ7yMWi8Xyp7wAEBBG4kMOH/MAAAAASUVORK5CYII=","orcid":"","institution":"Instituto de Investigaciones Biomédicas Sols-Morreale","correspondingAuthor":true,"prefix":"","firstName":"Cristina","middleName":"","lastName":"Peña","suffix":""}],"badges":[],"createdAt":"2026-02-23 15:09:38","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8948517/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8948517/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":104311259,"identity":"1ccfdd99-cc9d-499d-813f-daffb1b606e8","added_by":"auto","created_at":"2026-03-10 10:57:11","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":644079,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEV sncRNAs robustly discriminate colon cancer patients from healthy individuals.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e(A) Box plots of EV sncRNA expression levels in healthy donors and colon cancer patients, with p-values derived from Student’s t-tests.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e(B) Performance metrics of multiple supervised classification models evaluated using an 80/20 training–validation split and five-fold cross-validation. Area under the ROC curve (AUC), accuracy, and specificity are shown.\u003c/p\u003e\n\u003cp\u003e(C) Receiver operating characteristic (ROC) curve of the optimized Random Forest model distinguishing healthy donors from colon cancer patients.\u003c/p\u003e\n\u003cp\u003e(D) Predicted probabilities assigned by the Random Forest model for each individual sample.\u003cbr\u003e\n(E) Variable importance scores derived from the Random Forest model.\u003c/p\u003e\n\u003cp\u003e(F) Odds ratios obtained from logistic regression analysis for each sncRNA.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/73376a699a7880599b4fde77.png"},{"id":104311315,"identity":"56daffc4-d569-4477-8a9c-dba274fb01d8","added_by":"auto","created_at":"2026-03-10 10:57:24","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":674445,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEV sncRNAs enable detection of benign and premalignant colonic lesions.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e(A) Box plots of EV sncRNA expression levels in healthy donors and individuals with benign or premalignant colonic lesions (hyperplasia and adenoma), with p-values derived from Student’s t-tests.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e(B) Comparative performance of supervised classification models evaluated using five-fold cross-validation, highlighting the discriminative capacity of different algorithms.\u003c/p\u003e\n\u003cp\u003e(C) ROC curve of the optimized Random Forest model applied to the same comparison.\u003cbr\u003e\n(D) Predicted probabilities assigned by the Random Forest model to individual samples.\u003cbr\u003e\n(E) Variable importance scores derived from the Random Forest model.\u003c/p\u003e\n\u003cp\u003e(F) Odds ratios calculated using logistic regression analysis.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/2f253c4e645a947b826ed48d.png"},{"id":104311264,"identity":"b2ce54b7-f92e-49e2-88a2-5ef19add9d10","added_by":"auto","created_at":"2026-03-10 10:57:13","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":647619,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEV sncRNA signature shows diagnostic value in pancreatic ductal adenocarcinoma (PDAC) cancer.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) \u003cem\u003eBox plots of EV sncRNA expression levels in healthy donors and PDAC patients, with p-values derived from Student’s t-tests.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e(B) Performance comparison of supervised classification models evaluated using five-fold cross-validation.\u003c/p\u003e\n\u003cp\u003e(C) ROC curve of the optimized Random Forest model discriminating pancreatic cancer patients from healthy individuals.\u003c/p\u003e\n\u003cp\u003e(D) Predicted probabilities assigned to each sample by the Random Forest model.\u003c/p\u003e\n\u003cp\u003e(E) Variable importance scores from the Random Forest model.\u003c/p\u003e\n\u003cp\u003e(F) Odds ratios derived from logistic regression analysis.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/a6e6c0be5a4ebad23870ac1e.png"},{"id":104311261,"identity":"831c5de8-429f-45ff-8d3e-71569ab1683f","added_by":"auto","created_at":"2026-03-10 10:57:11","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":639543,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eAn EV sncRNA signature identifies gastrointestinal malignancy.\u003c/strong\u003e\u003cbr\u003e\n(A) \u003cem\u003eBox plots of EV sncRNA expression levels in healthy donors and patients with gastrointestinal cancer (colon cancer [CC] and pancreatic ductal adenocarcinoma [PDAC]), with p-values derived from Student’s t-tests.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e(B) Performance metrics of classification models distinguishing healthy individuals from gastrointestinal cancer patients using an 80/20 split and five-fold cross-validation.\u003cbr\u003e\n(C) ROC curve of the optimized Random Forest model for gastrointestinal cancer detection.\u003cbr\u003e\n(D) Predicted probabilities for individual samples classified by the Random Forest model.\u003cbr\u003e\n(E) Variable importance scores derived from the Random Forest model.\u003c/p\u003e\n\u003cp\u003e(F) Odds ratios obtained from logistic regression analysis.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/e4319becad13379fddfebac2.png"},{"id":104311186,"identity":"bf23666e-b400-46ad-9129-096f9b22ace8","added_by":"auto","created_at":"2026-03-10 10:57:05","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":140807,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEV sncRNAs stratify colon cancer patients according to relapse risk.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e(A\u003c/em\u003e \u003cem\u003eBox plots of EV sncRNA expression levels in relapsing and non-relapsing stage I–III colon cancer patients, with p-values derived from Student’s t-tests.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e(B) Performance metrics of supervised classification models predicting three-year relapse using five-fold cross-validation.\u003c/p\u003e\n\u003cp\u003e(C) ROC curve of the optimized Random Forest model for relapse prediction.\u003c/p\u003e\n\u003cp\u003e(D) Ranked predicted relapse probabilities for individual patients.\u003c/p\u003e\n\u003cp\u003e(E) Variable importance scores derived from the Random Forest model.\u003c/p\u003e\n\u003cp\u003e(F) Odds ratios calculated using logistic regression analysis.\u003c/p\u003e\n\u003cp\u003e(G) Kaplan–Meier relapse-free survival analysis of stage I–III colon cancer patients stratified into low- and high-risk groups based on predicted relapse probability. The optimal cut point was determined using a log-rank test.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/41467c5a81a1a516a5cf6f08.png"},{"id":104311187,"identity":"09569010-0dbd-4383-9354-5179af93b4ac","added_by":"auto","created_at":"2026-03-10 10:57:05","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":141707,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eEV sncRNA–based risk modeling identifies colon cancer patients with adverse clinical outcome.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e(A) Performance metrics of classification models predicting overall survival in a TCGA colon cancer cohort based on the sncRNA-regulated target gene expression.\u003c/p\u003e\n\u003cp\u003e(B) ROC curve of the optimized Random Forest model for overall survival prediction.\u003c/p\u003e\n\u003cp\u003e(C) Ranked predicted probabilities for mortality within 36 months.\u003c/p\u003e\n\u003cp\u003e(D) Kaplan–Meier overall survival curves stratifying patients into low- and high-risk groups based on the optimized probability threshold.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/1e3877ad5263b20f01accf02.png"},{"id":106727323,"identity":"f37685a7-52b2-4ebb-b8bc-1f45093922ac","added_by":"auto","created_at":"2026-04-12 18:38:42","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":3699872,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/29e478eb-0e00-4981-a0b5-1f1928df1659.pdf"},{"id":104311307,"identity":"1970b9e4-c4e9-4807-88fb-1afdce7c2316","added_by":"auto","created_at":"2026-03-10 10:57:22","extension":"ods","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":7383,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eSupplementary Table S1. Pairwise correlation analysis of EV sncRNA expression levels in all studied individuals. \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003ePearson correlation coefficients and significance values for U1, RNU1-11P, and piR-36249.\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Table S2. Distribution of EV sncRNA expression levels across all study groups. \u003c/strong\u003e\u003cem\u003eSummary statistics for EV sncRNA expression in healthy donors, patients with benign and premalignant colonic lesions, and colon and pancreatic cancer.\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Table S3. Functional enrichment analysis of CAF-associated EV sncRNA target genes. \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003e20 most significantly enriched Gene Ontology biological processes terms.\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Table S4. List of predicted target genes regulated by CAF-associated EV sncRNAs used to construct the prognostic gene signature.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Table S5. Univariate association of sncRNA target gene expression with overall survival in the TCGA colon cancer cohort. \u003c/strong\u003e\u003cem\u003e\u003cstrong\u003eStudent’s t-test results for genes retained after preprocessing.\u003c/strong\u003e\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Table S6. Variable importance scores for sncRNA target genes derived from the Random Forest model predicting overall survival in the TCGA colon cancer cohort.\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eSupplementary Table S7. Logistic regression odds ratios for sncRNA target genes associated with overall survival in the TCGA colon cancer cohort.\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"ColladoMetalSupplementaryTables.ods","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/b15814a8cb3637252514ff7b.ods"},{"id":104311152,"identity":"06195a83-edaa-408f-a020-cc814d1d1766","added_by":"auto","created_at":"2026-03-10 10:57:00","extension":"png","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":841841,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eGraphical abstract description\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSchematic representation of a non-invasive liquid biopsy strategy based on stroma-associated extracellular vesicles (EVs) small non-coding RNAs. Cancer-associated fibroblast–related EV snRNAs and piRNAs are detected in peripheral blood and analyzed using machine learning approaches to enable early diagnosis, risk stratification, and clinically relevant decision-making in colon and pancreatic cancer. \u003cstrong\u003eCreated with BioRender.com\u003c/strong\u003e\u003c/p\u003e","description":"","filename":"ColladoMelalGraphicalAbstract.png","url":"https://assets-eu.researchsquare.com/files/rs-8948517/v1/faa51e8024ef3ca04ea64bb0.png"}],"financialInterests":"No competing interests reported.","formattedTitle":"Stroma-associated extracellular vesicles snRNAs and piRNAs as non-invasive diagnostic and prognostic biomarkers in colon and pancreatic cancer","fulltext":[{"header":"HIGHLIGHTS","content":"\u003cul\u003e\n \u003cli\u003eCancer-associated fibroblast\u0026ndash;associated extracellular vesicles snRNAs and piRNAs are detectable in peripheral blood by liquid biopsy.\u003c/li\u003e\n \u003cli\u003eA three-sncRNA signature discriminates healthy individuals from patients with benign colonic lesions, colon cancer, and pancreatic ductal adenocarcinoma, supporting broader applicability across gastrointestinal malignancies.\u003c/li\u003e\n \u003cli\u003ePre-operative extracellular vesicles sncRNA levels enable risk stratification and relapse prediction in stage I\u0026ndash;III colon cancer.\u003c/li\u003e\n \u003cli\u003eTarget genes regulated by CAF-associated sncRNAs predict overall survival in a large public colon cancer cohort.\u003c/li\u003e\n\u003c/ul\u003e"},{"header":"INTRODUCTION","content":"\u003cp\u003eGastrointestinal (GI) cancers account for a substantial proportion of cancer incidence and mortality worldwide. In 2022 alone, more than five million new GI cancer cases were diagnosed globally, with close to 3.5\u0026nbsp;million associated deaths. Among these malignancies, colorectal cancer (CRC) ranks as the third most frequently diagnosed cancer and the second leading cause of cancer-related death, representing the most common GI tumor. Approximately two million new CRC cases and one million deaths are reported each year worldwide. Pancreatic ductal adenocarcinoma (PDAC), while less frequent and ranking twelfth in incidence, remains disproportionately lethal and is currently the sixth leading cause of cancer-related mortality. Together, CRC and PDAC account for nearly 2.5\u0026nbsp;million diagnoses and 1.5\u0026nbsp;million deaths annually (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eEarly-stage CRC (stages I\u0026ndash;II) is associated with a favorable prognosis, whereas survival declines sharply in advanced disease (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e). Despite curative-intent surgery, nearly 30% of patients with stage I\u0026ndash;III CRC experience disease recurrence (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e, \u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e). Notably, the risk of recurrence is highest during the first postoperative year and decreases thereafter, with 83% of recurrences occurring within three years of surgery (\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e, \u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e). These observations underscore the clinical importance of diagnosing CRC at early stages, when tumors are more likely to be localized and associated with a lower risk of recurrence and improved long-term outcomes.\u003c/p\u003e \u003cp\u003eIn this context, population-based screening programs play a central role in enabling early detection before the onset of advanced disease. Current screening strategies include guaiac-based fecal occult blood testing (gFOBT), fecal immunochemical testing (FIT), and stool DNA tests, followed by colonoscopy when results are positive (\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e). Although these approaches are relatively inexpensive and easy to implement, their clinical application remains controversial. Their performance is limited in hospitalized patients, individuals with GI bleeding, or those receiving medications such as acetylsalicylic acid, anticoagulants, antiplatelet agents, non-steroidal anti-inflammatory drugs, selective serotonin reuptake inhibitors, or iron supplements (\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e). False-positive results frequently lead to unnecessary invasive procedures, increasing healthcare costs and exposing patients to risks including cardiopulmonary complications related to sedation, bleeding, perforation, and infection (\u003cspan additionalcitationids=\"CR10\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e). Importantly, more than 40% of CRC-related deaths occur in patients older than 75 years, a population in whom invasive diagnostic procedures carry increased risk due to comorbidities and age-related vulnerability (\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e). In this context, no screening strategy has demonstrated clear superiority, and test selection must be individualized (\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e). Screening adherence also remains suboptimal and may benefit from the availability of blood-based biomarkers (\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e).\u003c/p\u003e \u003cp\u003ePDAC is projected to surpass breast, prostate, and colorectal cancers to become the second leading cause of cancer-related death by 2030 (\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e). Pancreatic ductal adenocarcinoma (PDAC) accounts for approximately 90% of pancreatic cancer cases and is characterized by an extremely poor prognosis, with mortality rates closely mirroring incidence (\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e, \u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). Five-year survival currently exceeds only 12% (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e), largely due to the absence of early symptoms, resulting in nearly 90% of tumors being diagnosed at advanced or metastatic stages (\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e). Surgical resection is feasible in only about 20% of patients and improves five-year survival to 15\u0026ndash;25% in this subgroup (\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e). The relatively low incidence of PDAC, the lack of clearly defined high-risk populations, and the absence of non-invasive screening tools restrict screening programs to individuals with hereditary or familial risk. Available screening methods rely on imaging modalities such as endoscopic ultrasound (EUS), magnetic resonance imaging (MRI), magnetic resonance cholangiopancreatography (MRCP), computed tomography (CT), abdominal ultrasound, and positron emission tomography (PET), all of which are costly and limited by suboptimal sensitivity and specificity, preventing their application in the general population (\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e, \u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTumor development and progression depend not only on malignant epithelial cells but also on the surrounding tumor microenvironment (TME), which comprises cellular components, including endothelial cells, pericytes, immune cells, and fibroblasts, and non-cellular elements such as extracellular matrix, growth factors, and extracellular vesicles (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e). Bidirectional communication between tumor cells and stromal components reshapes the TME and supports tumor growth, angiogenesis, and metastatic dissemination (\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e). Fibroblasts, the predominant stromal cell population, undergo phenotypic and functional changes upon interaction with tumor cells, giving rise to cancer-associated fibroblasts (CAFs) (\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e, \u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e, \u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e). Increased stromal content has been consistently associated with adverse outcomes in colon, breast, and esophageal cancers (\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e). Moreover, CAF-related gene signatures and markers have shown prognostic relevance across multiple tumor types, including colorectal, pancreatic, gastric, and lung cancers (\u003cspan additionalcitationids=\"CR29 CR30 CR31\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e). Nevertheless, clinically applicable stroma-derived diagnostic biomarkers remain largely unavailable.\u003c/p\u003e \u003cp\u003eAlthough CRC and PDAC are clinically distinct entities, they share important biological features. Both are epithelial adenocarcinomas arising within the gastrointestinal tract and are characterized by a prominent desmoplastic stroma enriched in cancer-associated fibroblasts. In particular, PDAC is well known for its highly fibrotic and CAF-dense microenvironment (\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e), while increasing evidence highlights the prognostic relevance of stromal content in colon cancer (\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e). This shared stromal architecture provides a biologically plausible rationale to explore whether stroma-derived molecular signals may represent common and systemically detectable biomarkers across these malignancies.\u003c/p\u003e \u003cp\u003eIn this context, extracellular vesicles (EVs) emerge as key mediators of intercellular communication within the TME and as potential carriers of stromal-derived signals into the systemic circulation. These vesicles, measuring approximately 40\u0026ndash;150 nm, transport bioactive cargo, proteins, lipids, RNAs, and DNA fragments that reflects their cellular origin (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e, \u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e). Tumor-derived EVs contribute to CAF activation (\u003cspan additionalcitationids=\"CR38\" citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e), while CAF-derived EVs, particularly through their non-coding RNA (ncRNA) content, enhance tumor cell proliferation, stemness, and chemoresistance (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e). Our group previously demonstrated a distinct ncRNA cargo in CAF-derived versus normal fibroblast (NF)-derived EVs in colon cancer (CC), with enrichment of small nuclear RNAs (snRNAs), PIWI-interacting RNAs (piRNAs), and microRNAs (miRNAs) (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). We further showed that gene expression signatures derived from CAF-related ncRNAs predict prognosis in large cohorts of CC patients (N\u0026thinsp;=\u0026thinsp;1,235) (\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e). These findings suggest that CAF-derived EV ncRNAs may represent clinically informative biomarkers.\u003c/p\u003e \u003cp\u003eBased on this background, the present study was designed to: (i) evaluate whether two snRNAs (U1 and RNU1-11P) and one piRNA (piR-36249), previously identified as enriched in CAF-derived EVs and representing additional classes of small non-coding RNAs, can serve as diagnostic and prognostic biomarkers in colon pathology using peripheral blood samples obtained prior to surgical intervention; (ii) validate their diagnostic performance in an independent PDAC cohort; and (iii) assess the prognostic value of their target genes in a large, publicly available cohort of CC patients. By integrating stromal biology with liquid biopsy\u0026ndash;based modeling strategies, this approach seeks to capture systemically detectable signals of tumor\u0026ndash;stroma interaction at a preoperative stage, when clinically actionable information is often limited and largely dependent on postoperative pathological assessment.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cp\u003ePatient samples and data collection\u003c/p\u003e \u003cp\u003ePeripheral blood samples from CC patients were obtained at Ram\u0026oacute;n y Cajal University Hospital in collaboration with the Nursing Service. Blood samples from sporadic pancreas cancer patients, were collected from the Spanish Familial Pancreatic Cancer Registry (PANGENFAM)(\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e). In addition, the Hospital Ram\u0026oacute;n y Cajal Biobank provided blood samples from healthy donors and from individuals diagnosed with colonic hyperplasia or adenoma, recruited through its established sample collections.\u003c/p\u003e \u003cp\u003eClinicopathological data were retrieved in collaboration with the Medical Oncology, Radiation Oncology, and Pathology Services of Ram\u0026oacute;n y Cajal University Hospital. All participants provided written informed consent prior to sample collection. The study protocol was approved by the local Ethics and Clinical Investigation Committee (Comit\u0026eacute; de \u0026Eacute;tica de la Investigaci\u0026oacute;n con Medicamentos del Hospital Universitario Ram\u0026oacute;n y Cajal)) and conducted in accordance with the Declaration of Helsinki. Blood samples were obtained prior to surgical intervention and subsequently processed and stored by the Biobank.\u003c/p\u003e \u003cp\u003eBlood sample processing\u003c/p\u003e \u003cp\u003ePeripheral blood samples were collected in red Vacutainer SST\u0026trade; II Advance tubes (Becton Dickinson). Samples were processed under standardized conditions to obtain the serum fraction, which was aliquoted and stored at \u0026minus;\u0026thinsp;80\u0026deg;C until further use.\u003c/p\u003e \u003cp\u003eExtracellular vesicles isolation and RNA extraction\u003c/p\u003e \u003cp\u003eEVs were isolated from 250 \u0026micro;L serum aliquots using polymer-based precipitation with the ExoQuick\u0026trade; Exosome Precipitation Solution (System Biosciences), following the manufacturer\u0026rsquo;s instructions. Total RNA was extracted from isolated EVs using the mirVana\u0026trade; miRNA Isolation Kit (Ambion).\u003c/p\u003e \u003cp\u003eTo monitor extraction efficiency, the synthetic RNA spike-in UniSP2 was added to each sample together with MS2 carrier RNA, which improves spike-in stability without interfering with downstream applications. Purified RNA was eluted in nuclease-free water and stored at \u0026minus;\u0026thinsp;80\u0026deg;C until analysis.\u003c/p\u003e \u003cp\u003eRNA reverse transcription\u003c/p\u003e \u003cp\u003eReverse transcription was performed using the miRCURY LNA\u0026trade; RT Kit (Qiagen) according to the manufacturer\u0026rsquo;s protocol. The synthetic \u003cem\u003eCaenorhabditis elegans\u003c/em\u003e miRNA cel-miR-39-3p (Qiagen) was added to each reaction as an internal control for reverse transcription efficiency. Complementary DNA (cDNA) was stored at \u0026minus;\u0026thinsp;20\u0026deg;C until quantitative PCR analysis.\u003c/p\u003e \u003cp\u003eDetection of small non-coding RNAs by qPCR\u003c/p\u003e \u003cp\u003eQuantitative PCR (qPCR) assays were performed to assess the efficiency of RNA extraction and reverse transcription using UniSP2 and cel-miR-39-3p controls, respectively, with miRCURY LNA\u0026trade; miRNA primers (Qiagen). For detection of the small non-coding RNAs (sncRNAs) of interest (piR-36249, U1 and RNU1-11P), custom miRCURY LNA\u0026trade; miRNA primers specific to each sequence were designed (Qiagen).\u003c/p\u003e \u003cp\u003ecDNA samples were diluted 1:11 with nuclease-free water. qPCR reactions were carried out using miRCURY LNA\u0026trade; SYBR Green Master Mix (Qiagen) on a LightCycler\u0026reg; 480 system (Roche). UniSP2 was used for data normalization, as it was added at the same concentration to all samples. Relative expression levels were calculated using the second-derivative method. Missing expression values due to low transcript abundance were imputed using half of the minimum detected value for the corresponding sncRNA.\u003c/p\u003e \u003cp\u003ePredictive modeling and statistical analysis\u003c/p\u003e \u003cp\u003eMultiple supervised machine learning algorithms were evaluated to identify the best-performing predictive models for each clinical endpoint and patient cohort.\u003c/p\u003e \u003cp\u003eThe following datasets were analyzed:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eColonic pathology cohort: A total of 183 individuals, including 40 healthy donors, 14 patients with colonic hyperplasia, 19 with adenomas, and 110 with CC were used to analyze the expression of the three CAF-derived extracellular vesicles sncRNAs in peripheral blood.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePancreatic pathology cohort: A total of 172 individuals, including 40 healthy donors, 34 patients with familial PDAC, and 98 with sporadic PDAC were used to analyze the expression of the three CAF-derived extracellular vesicles sncRNAs in peripheral blood.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003ePublic CC cohort: A publicly available dataset of 597 CC patients from TCGA COADREAD cohort (TCGA; COADREAD Pan-Cancer Atlas, 2018) was analyzed to evaluate overall survival (OS) in relation to the predicted target genes of the studied sncRNAs (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e).\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003eFor each dataset, the performance of multiple classification algorithms was evaluated using a five-fold cross-validation approach. In this framework, the data were repeatedly partitioned into training (80%) and validation (20%) subsets across five iterations, such that each sample was used once for validation and multiple times for training. This strategy provides a robust and unbiased estimate of model performance while reducing the risk of overfitting. Given the presence of class imbalance in several cohorts, the Synthetic Minority Over-sampling Technique (SMOTE) was applied to balance the training data.\u003c/p\u003e \u003cp\u003eAll statistical analyses and machine learning procedures were implemented in Python, using standard scientific computing and machine learning libraries.\u003c/p\u003e \u003cp\u003eFunctional enrichment analyses were performed using the Gene Set Enrichment Analysis (GSEA) platform and the Molecular Signatures Database (MSigDB) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://www.gsea-msigdb.org/gsea/msigdb\u003c/span\u003e\u003cspan address=\"https://www.gsea-msigdb.org/gsea/msigdb\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e), focusing on biological process gene sets.\u003c/p\u003e"},{"header":"RESULTS","content":"\u003cp\u003eClinicopathological characteristics and EV sncRNA profiling of the study cohorts\u003c/p\u003e \u003cp\u003eThe cohort of healthy donors included a total of 40 individuals.\u003c/p\u003e \u003cp\u003eThe CC cohort comprised 110 patients, with a mean age of 74 years (range: 33\u0026ndash;93), including 58 males and 52 females. Tumor location was right-sided in 47 patients (43%) and left-sided in 47 patients (43%). Four tumors were located at the rectosigmoid junction (3%), which were classified according to institutional anatomical and clinical criteria and were included in the colon cancer cohort. Information about tumor location was unavailable for 12 cases (11%). Regarding tumor stage, 22 patients were classified as stage I (20%), 35 as stage II (32%), 37 as stage III (34%), and 12 as stage IV (11%); staging information was missing for four patients (3%). During follow-up, disease relapse was observed in 22 patients (20%), whereas 88 patients (80%) remained relapse-free. For relapse prediction analyses, stage IV patients were excluded, resulting in a final dataset of 97 individuals, including 16 relapses (17%) and 81 non-relapses (83%). In addition, the study included 14 patients with colonic hyperplasia and 19 with adenomas, considered benign and premalignant lesions, respectively. The pancreatic pathology cohort consisted of 132 patients with PDAC.\u003c/p\u003e \u003cp\u003eThe selection of the small non-coding RNAs (sncRNAs) analyzed in this study was based on previous work from our group, which characterized the differential distribution of sncRNAs in cellular and extracellular vesicle (EV) fractions derived from normal fibroblasts (NFs) and CAFs (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). From this dataset, candidate molecules were prioritized according to the magnitude of deregulation between NFs and CAFs and their high read counts, ensuring both biological relevance and technical robustness. Based on these criteria, one piRNA (piR-36249) and two snRNAs (U1 and RNU1-11P) enriched in CAF-derived EVs were selected for evaluation.\u003c/p\u003e \u003cp\u003eCorrelation analysis among the three studied sncRNAs showed significant interdependence between variables (Supplementary Table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). A summary of sncRNA expression levels across all study groups is provided in Supplementary Table S2.\u003c/p\u003e \u003cp\u003eEV sncRNAs discriminate colon cancer from healthy individuals\u003c/p\u003e \u003cp\u003eAs an initial exploratory approach, Student\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-tests were performed to compare healthy donors and CC patients. U1 and RNU1-11P expression levels showed statistically significant differences between groups, whereas piR-36249 did not (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo further evaluate diagnostic performance, multiple supervised classification algorithms were applied. A five-fold cross-validation scheme was used for both parameter optimization and performance evaluation. All tested models demonstrated robust discriminative performance, achieving areas under the receiver operating characteristic curve (AUCs) above 0.70. Detailed performance metrics, including accuracy and specificity, are summarized in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eB.\u003c/p\u003e \u003cp\u003eThe Random Forest model outperformed other classifiers in terms of accuracy and AUC, leading to its selection for subsequent analyses. The average AUC across the five folds was 0.773, as illustrated by the ROC curve (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eC). Predicted probabilities for individual samples are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eD, illustrating clear separation between healthy donors and CC patients. Variable importance estimates derived from the Random Forest model are presented in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eE.\u003c/p\u003e \u003cp\u003eAlthough logistic regression exhibited slightly lower performance compared with Random Forest, its interpretability warranted further consideration. Odds ratios derived from this model are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eF.\u003c/p\u003e \u003cp\u003eDetection of benign and premalignant colonic lesions using EV sncRNAs\u003c/p\u003e \u003cp\u003eGiven the clinical relevance of early lesion detection, we next assessed whether the sncRNA panel could distinguish healthy donors from individuals with benign or premalignant colonic lesions (hyperplasia and adenoma).\u003c/p\u003e \u003cp\u003eStudent\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-test analysis revealed a statistically significant difference for RNU1-11P expression, whereas no significant differences were observed for piR-36249 or U1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eA). Subsequent machine learning analyses indicated that the Support Vector Machine (SVM) model achieved the best predictive performance in this setting, with an AUC of 0.792 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe average AUC across the five folds using the Random Forest model was 0.764 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). Predicted probabilities for individual samples are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eD. Variable importance rankings from the Random Forest model and odds ratios derived from logistic regression are displayed in Figs.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eE and \u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eF, respectively.\u003c/p\u003e \u003cp\u003eEV sncRNAs as diagnostic biomarkers in pancreatic cancer\u003c/p\u003e \u003cp\u003eTo evaluate whether the diagnostic potential of the selected sncRNAs extended beyond colorectal malignancy, their expression was analyzed in EVs isolated from peripheral blood samples of PDAC patients.\u003c/p\u003e \u003cp\u003eStudent\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-tests demonstrated statistically significant differences for all three sncRNAs when comparing PDAC patients with healthy donors (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eA). Following the same analytical workflow established for CC, classification models were evaluated and compared using a five-fold cross-validation approach. All evaluated algorithms achieved AUC values above 0.700 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eRandom Forest again showed the best overall performance. The average AUC across the five was 0.775 (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eC). Sample-level predicted probabilities are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eD, with clear separation between healthy individuals and PDAC patients. Variable importance scores and logistic regression-derived odds ratios are presented in Figs.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eE and \u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eF, respectively.\u003c/p\u003e \u003cp\u003eEV sncRNA signature identifies gastrointestinal cancer\u003c/p\u003e \u003cp\u003eTo explore the ability of the sncRNA panel to detect GI malignancy more broadly, CC and PDAC patients were combined into a single cancer group and compared with healthy donors.\u003c/p\u003e \u003cp\u003eAll three sncRNAs showed significant differential expression between healthy individuals and cancer patients in Student\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-test analyses (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eA). Classification models evaluated via five-fold cross-validation consistently yielded AUC values exceeding 0.700 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eB).\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe average AUC across the five folds was 0.715 (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eC). Predicted probabilities for individual samples are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eD. Variable importance rankings and odds ratios derived from Random Forest and logistic regression models are shown in Figs.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eE and \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eF, respectively.\u003c/p\u003e \u003cp\u003ePrognostic value of EV sncRNAs in colon cancer relapse\u003c/p\u003e \u003cp\u003eWe next assessed whether EV sncRNA expression levels were associated with relapse in CC patients. Student\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-test analysis did not reveal significant associations between relapse status and expression of all three sncRNAs (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). To develop a predictive model for three-year relapse, analyses were restricted to the 97 stage I\u0026ndash;III CC patients included in the relapse cohort. The dataset comprised 81 non-relapsing patients and 16 patients who experienced relapse. Classification algorithms were evaluated using five-fold cross-validation, and performance metrics are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eGiven its overall performance across evaluation metrics, the Random Forest algorithm was selected for further characterization. When assessed by five-fold cross-validation, the model achieved a modest mean AUC of 0.541 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eC), likely reflecting the limited number of patients experiencing relapse in this cohort. Notably, despite the reduced discriminative capacity, the model retained a relatively high specificity (0.690), indicating a consistent ability to correctly identify patients unlikely to relapse. Predicted probabilities were ranked and visualized to facilitate interpretation (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD). Variable importance estimates and odds ratios derived from Random Forest and logistic regression models are shown in Figs.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eE and \u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eF, respectively.\u003c/p\u003e \u003cp\u003eTo further stratify patients, a probability-based thresholding approach was applied. Patients were divided into low- and high-risk groups according to predicted relapse probability, and Kaplan\u0026ndash;Meier survival analyses were performed iteratively to identify the optimal cut point. The most significant separation was obtained at cut point 52 (indicated by a dotted red line in Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eD), yielding a log-rank \u003cem\u003ep\u003c/em\u003e-value of 0.009 (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eG).\u003c/p\u003e \u003cp\u003eThe extracellular vesicles sncRNAs target gene signature predicts overall survival in public TCGA colon cancer cohorts\u003c/p\u003e \u003cp\u003eTo validate the biological and clinical relevance of the studied sncRNAs, we analyzed the expression of their predicted target genes, which were previously identified and functionally characterized by our group in CAF-derived EVs (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e). In that study, a total of 64 target genes regulated by these sncRNAs were defined. Functional enrichment analysis of these genes revealed significant involvement in pathways related to cancer progression, including tyrosine kinase signaling, apoptosis, cell proliferation, and DNA repair (Supplementary Table S3). Based on this enrichment analysis, a prognostic gene signature comprising 43 genes was constructed, corresponding to those included in the ten most significantly represented biological process pathways (Supplementary Table S4).\u003c/p\u003e \u003cp\u003eGene expression and overall survival data were obtained from a TCGA CC cohort of 597 patients. Among these, 514 patients had overall survival exceeding 36 months, while 83 patients died within 36 months of diagnosis. After data preprocessing, 43 uncorrelated variables were retained. Their individual associations with survival were evaluated using Student\u0026rsquo;s \u003cem\u003et\u003c/em\u003e-tests (Supplementary Table S5).\u003c/p\u003e \u003cp\u003eA five-fold cross-validation scheme was used for both parameter optimization and performance evaluation. Random Forest and SVM models achieved AUC values above 0.80, with specificities approaching 1, indicating a very high accuracy in identifying patients who will experience early death (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eA). Although a small subset of patients who died within 36 months were classified as low risk, the near-perfect specificity indicates that patients classified by the model as high risk almost invariably experienced death within this time frame. This characteristic underscores the clinical value of the model for confidently identifying a subgroup of patients with particularly unfavorable prognosis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eUsing the Random Forest model, the mean AUC across the five folds was 0.813 (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eB). Ranked predicted probabilities are shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC, while variable importance scores and logistic regression odds ratios are reported in Supplementary Tables S6 and S7.\u003c/p\u003e \u003cp\u003eUsing the same probability-based stratification approach described above, patients were divided into low- and high-risk groups for mortality. Kaplan\u0026ndash;Meier analysis identified an optimal cut point at position 527 (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eC), yielding a log-rank \u003cem\u003ep\u003c/em\u003e-value\u0026thinsp;\u0026lt;\u0026thinsp;0.001 (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eD).\u003c/p\u003e"},{"header":"DISCUSSION","content":"\u003cp\u003eIn this study, we identify a set of three sncRNAs (two snRNAs, U1 and RNU1-11P, and the piRNA piR-36249) as diagnostic and prognostic non-invasive biomarkers in two highly lethal GI malignancies, CC and PDAC. In addition, the expression of the target genes regulated by these sncRNAs in tumors showed strong prognostic value in a large, independent public CC cohort. Building on our previous work demonstrating that these sncRNAs are deregulated in CAFs and enriched in CAF-derived EVs, this study represents one of the first efforts to translate stroma-associated biomarkers into a non-invasive liquid biopsy setting. By detecting these sncRNAs in serum-derived EVs, our results highlight the clinical potential of tumor stroma\u0026ndash;related signals for diagnosis and risk stratification using peripheral blood samples.\u003c/p\u003e \u003cp\u003eMost circulating biomarkers currently used or under evaluation in GI oncology are derived directly from tumor cells. These include circulating tumor DNA (ctDNA) (\u003cspan citationid=\"CR45\" class=\"CitationRef\"\u003e45\u003c/span\u003e), tumor-derived EVs (\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e), serum protein biomarkers such as carcinoembryonic antigen (\u003cspan citationid=\"CR46\" class=\"CitationRef\"\u003e46\u003c/span\u003e) and CA19-9 (\u003cspan citationid=\"CR47\" class=\"CitationRef\"\u003e47\u003c/span\u003e), as well as circulating miRNAs (\u003cspan citationid=\"CR48\" class=\"CitationRef\"\u003e48\u003c/span\u003e). Although these markers have contributed substantially to clinical management, their performance in early-stage disease remains limited, and their specificity can be compromised in a range of clinical contexts (\u003cspan additionalcitationids=\"CR50\" citationid=\"CR49\" class=\"CitationRef\"\u003e49\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR51\" class=\"CitationRef\"\u003e51\u003c/span\u003e). These limitations highlight the need for complementary biomarkers that capture biological processes not fully reflected by tumor cell\u0026ndash;intrinsic signals.\u003c/p\u003e \u003cp\u003eWithin the tumor microenvironment, accumulating evidence indicates that CAFs actively shape tumor behavior through the transfer of ncRNAs via EVs. In both CC and PDAC, CAF-derived EV ncRNAs have been shown to enhance tumor cell stemness, invasiveness, metastatic potential, and resistance to chemotherapy (\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e, \u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e, \u003cspan additionalcitationids=\"CR53 CR54 CR55\" citationid=\"CR52\" class=\"CitationRef\"\u003e52\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR56\" class=\"CitationRef\"\u003e56\u003c/span\u003e). However, the majority of these studies have focused on miRNAs and long non-coding RNAs, which, despite extensive characterization, have demonstrated limited clinical translation. Challenges related to specificity, reproducibility, and the difficulty of isolating stromal-derived signals from circulation have hindered broader adoption.\u003c/p\u003e \u003cp\u003eBy contrast, other classes of sncRNAs, including snRNAs and piRNAs, remain comparatively understudied in the context of cancer biomarkers. snRNAs are core components of the spliceosome and play a central role in mRNA processing. Disruption of snRNA biogenesis and processing has recently been linked to cancer progression; notably, increased levels of unprocessed snRNAs resulting from Integrator complex dysfunction have been associated with poor prognosis in CRC (\u003cspan citationid=\"CR57\" class=\"CitationRef\"\u003e57\u003c/span\u003e). piRNAs, initially described in germ cells as regulators of transposable element activity, are now recognized as contributors to cancer stem cell maintenance and tumor progression (\u003cspan citationid=\"CR58\" class=\"CitationRef\"\u003e58\u003c/span\u003e). Several studies have suggested that piRNAs may outperform conventional biomarkers in certain oncological settings (\u003cspan citationid=\"CR59\" class=\"CitationRef\"\u003e59\u003c/span\u003e) yet their stromal origin and EV transport have received little attention.\u003c/p\u003e \u003cp\u003eConsistent with these observations, our previous work identified enrichment of specific snRNAs and piRNAs in CAF-derived EVs and linked these molecules to distinct gene expression programs in CC (\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e, \u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e). Here, we move beyond tissue-based analyses and demonstrate that these stroma-associated sncRNAs are not confined to the local tumor microenvironment but can be detected systemically in peripheral blood. Importantly, their ability to discriminate healthy individuals not only from patients with overt malignancy but also from those with benign and premalignant lesions suggests that they capture early stromal activation and tissue remodeling processes. While differences between early lesions and normal tissue are necessarily subtle, the observed discrimination supports their potential utility for disease detection at very early stages, a clinically relevant window in which intervention is more effective and long-term outcomes are improved.\u003c/p\u003e \u003cp\u003eBeyond their diagnostic potential, the three-sncRNA panel also demonstrated clinically relevant prognostic value in CC. The association between sncRNA expression levels and relapse-free survival, together with the risk stratification achieved through probability-based modeling, supports the relevance of these markers beyond disease detection alone. This prognostic dimension was further supported by the analysis of the expression of the genes regulated by the studied sncRNAs, which were used to construct a transcriptional signature and evaluated in a large independent public CC cohort. Although derived from bulk tumor expression data and assessed using overall survival as the clinical endpoint, this gene signature enabled robust patient stratification and provided independent evidence of the biological and clinical relevance of the sncRNA-associated programs. Functional enrichment analyses revealed that these target genes are involved in key pathways related to tumor progression, including tyrosine kinase signaling, apoptosis, proliferation, and DNA repair, supporting a mechanistic link between stromal-derived sncRNAs and malignant behavior. Taken together, these findings support the concept that tumor stroma\u0026ndash;associated biomarkers can provide actionable prognostic information that complements, and in certain clinical contexts may add value beyond, traditional tumor cell\u0026ndash;derived biomarkers.\u003c/p\u003e \u003cp\u003eIn addition to their diagnostic and prognostic associations, the most clinically relevant implication of this work lies in the early identification of a subset of patients with particularly high risk. The high specificity achieved by the proposed models indicates that patients classified as high risk are identified with a high degree of confidence, enabling the recognition of individuals who are likely to experience early relapse or unfavorable outcome. Importantly, all analyses were performed using peripheral blood samples obtained prior to surgical intervention, meaning that this risk information is available before tumor resection and before definitive pathological staging. This preoperative, minimally invasive assessment represents a potential shift in clinical management, particularly in colon cancer, where robust molecular prognostic information is typically derived only from the surgical specimen.\u003c/p\u003e \u003cp\u003eThis capacity for early, pre-surgical risk stratification opens a window for therapeutic intensification or closer surveillance in selected patients, including consideration of more aggressive adjuvant strategies or tailored follow-up schedules. Significantly, rather than replacing established clinical or pathological criteria, this model provides complementary information derived from the tumor stroma. In this context, the integration of stroma-associated EV sncRNAs into clinical workflows could represent a practical and non-invasive tool to support decision-making and improve personalized management of CC patients.\u003c/p\u003e \u003cp\u003eImportantly, the diagnostic potential of the sncRNA signature extended to PDAC, a malignancy for which early detection remains a major unmet clinical need. Despite the biological and clinical differences between colon and pancreatic tumors, the ability of the same sncRNA panel to discriminate PDAC patients from healthy individuals suggests that stroma-associated EV sncRNAs may capture shared stromal activation programs across GI cancers. The combined analysis of CC and PDAC further supports this concept and highlights the possibility of developing pan\u0026ndash;GI cancer screening approaches based on stromal biomarkers.\u003c/p\u003e \u003cp\u003eSeveral limitations of this study should be acknowledged. First, the retrospective, single-center design may introduce selection bias and limits generalizability. Second, the limited number of patients with colonic benign and premalignant lesions, as well as the relatively small number of relapses in the CC cohort, constrains model robustness and warrants validation in larger series. Third, the PDAC cohort lacked longitudinal follow-up data, precluding the assessment of prognostic performance in this setting. In addition, while previous experimental work supports the enrichment of the studied sncRNAs in CAFs and CAF-derived EVs, the precise cellular origin of circulating EVs in peripheral blood cannot be definitively established in this clinical setting. Finally, although the biological relevance of these sncRNAs is supported by their target gene analyses, additional functional studies will be required to further elucidate their mechanistic roles. Future work should therefore include prospective and multicenter validation and experimental approaches to refine both clinical applicability and biological interpretation.\u003c/p\u003e \u003cp\u003eDespite these limitations, our data provide strong evidence that stroma-associated EV snRNAs and piRNAs can be detected in peripheral blood and carry clinically meaningful diagnostic and prognostic information. The consistent associations observed across CC and PDAC support the notion that stromal activation programs are systemically reflected and may be leveraged as accessible biomarkers in GI malignancies. From a biological perspective, targeting the tumor stroma rather than tumor cells alone offers a complementary view of disease behavior. Looking forward, the integration of stromal-derived EV sncRNAs into liquid biopsy strategies holds promise for broader clinical application, including early detection, risk stratification, and personalized patient management. Together, these findings highlight the tumor stroma as a rational and underexplored source of biomarkers with translational relevance in GI cancer.\u003c/p\u003e"},{"header":"CONCLUSIONS","content":"\u003cp\u003eThis study demonstrates that a panel of three sncRNAs, two snRNAs and one piRNA, previously identified as enriched in CAF\u0026ndash;derived EVs can be detected in peripheral blood and used as diagnostic and prognostic biomarkers in CC, with diagnostic applicability also demonstrated in PDAC. By capturing stroma-associated signals through a non-invasive liquid biopsy approach, these biomarkers provide clinically relevant information that complements conventional tumor cell-centered strategies.\u003c/p\u003e \u003cp\u003eIn CC, the sncRNA panel discriminated healthy individuals from patients with benign and premalignant lesions and overt malignancy and showed prognostic value for relapse within three years after surgery. Importantly, these molecular signals were obtained from peripheral blood samples collected prior to surgical intervention, highlighting their potential utility for pre-operative risk assessment and therapeutic decision-making. In addition, validation of the predicted target genes regulated by these sncRNAs in a large public CC cohort confirmed their strong association with overall survival, underscoring the biological relevance of the underlying stromal signaling programs.\u003c/p\u003e \u003cp\u003eImportantly, the diagnostic potential of this sncRNA signature was not restricted to CC. The same panel identified PDAC patients with good diagnostic performance, supporting its applicability across distinct GI malignancies and reinforcing the concept that stromal activation programs may be shared across different tumor types within the GI system.\u003c/p\u003e \u003cp\u003eTogether, these findings support the notion that the tumor stroma is not only a key contributor to cancer progression but also a measurable systemic source of clinically informative biomarkers. CAF\u0026ndash;associated EV snRNAs and piRNAs represent an underexplored class of circulating molecules with significant translational potential. Their integration into clinical workflows could improve early detection, pre-operative stratification, and patient management, particularly in clinical settings where current biomarkers show limited sensitivity or specificity.\u003c/p\u003e \u003cp\u003eFurther prospective and multicenter studies will be required to validate these results and to define the optimal clinical contexts for implementation. Nonetheless, this work provides a strong rationale for the development of stromal-based liquid biopsy strategies and highlights the relevance of stroma-associated EV sncRNAs as accessible and biologically meaningful tools in the management of GI cancers.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003eFUNDING\u003c/h2\u003e\n\u003cp\u003eThis research was supported by PI20/00602 from the Instituto de Salud Carlos III and co-financed by the European Development Regional Fund (FEDER) “A way to achieve Europe” (ERDF); “CIBER de Cáncer” (CB16/12/00273), and Biobank and Biomodels Platform PT20/0045 and PT23/00098 from the Instituto de Salud Carlos III - FEDER “A way to achieve Europe” (ERDF); CNS2023-144882 from Agencia Estatal de Investigación “Plan de Recuperación, Transformación y Resiliencia; Next Generation EU”; PID2022-136729OB-I00 funded by MICIU/AEI/10.13039/501100011033 and FEDER, UE”; and P2022/BMD7212, Comunidad de Madrid. MC was supported by FI21/00132 from Instituto de Salud Carlos III - FEDER “A way to achieve Europe” (ERDF). MJL and JMG-S belong to the Spanish National Research Council (CSIC)'s Cancer Hub.\u003c/p\u003e\n\u003ch2\u003eAUTHOR CONTRIBUTIONS\u003c/h2\u003e\n\u003cp\u003eMC, VMC, MEC, MH, CGP and CP made substantial contribution to conception and design of the work; the acquisition, analysis and interpretation of the data; and draft and revision of the work. MJL, CLF, LGB, FHdO, JMGS, AC and VGB have contributed to the conception and design of the work. IM, BM, EC, JV and CdlP contributed to the acquisition of the data. All authors reviewed the manuscript.\u003c/p\u003e\n\u003ch2\u003eETHICS APPROVAL DECLARATION\u003c/h2\u003e\n\u003cp\u003eAll participants provided written informed consent prior to sample collection. The study protocol was approved by the local Ethics and Clinical Investigation Committee (Comité de Ética de la Investigación con Medicamentos del Hospital Universitario Ramón y Cajal)) and conducted in accordance with the Declaration of Helsinki.\u003c/p\u003e\n\u003ch2\u003eACKNOWLEDGMENTS\u003c/h2\u003e\n\u003cp\u003eThe authors acknowledge the Hospital Universitario Ramón y Cajal Biobank for providing access to biological samples and for its essential contribution to sample processing and management. We also thank the biobank staff for their professionalism and technical support. ChatGPT was used for English text editing. We are grateful to lab members for help and advice throughout this research.\u0026nbsp;\u003c/p\u003e\n\u003ch2\u003eCONFLICTS OF INTEREST\u003c/h2\u003e\n\u003cp\u003eThe authors declare no conflict of interest.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eBray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I et al. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin [Internet]. 2024 May [cited 2026 Jan 23];74(3):229\u0026ndash;63. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/38572751/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/38572751/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSiegel RL, Miller KD, Jemal A, Cancer statistics. 2020. CA Cancer J Clin [Internet]. 2020 Jan [cited 2026 Jan 23];70(1):7\u0026ndash;30. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/31912902/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/31912902/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVan Der Stok EP, Spaander MCW, Gr\u0026uuml;nhagen DJ, Verhoef C, Kuipers EJ. Surveillance after curative treatment for colorectal cancer. Nat Rev Clin Oncol [Internet]. 2017 May 1 [cited 2026 Jan 23];14(5):297\u0026ndash;315. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/27995949/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/27995949/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoute TC, Swartjes H, Greuter MJE, Elferink MAG, van Eekelen R, Vink GR et al. Cumulative Incidence, Risk Factors, and Overall Survival of Disease Recurrence after Curative Resection of Stage II-III Colorectal Cancer: A Population-based Study. Cancer research communications [Internet]. 2024 Feb 29 [cited 2026 Jan 23];4(2):607\u0026ndash;16. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/38363145/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/38363145/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShi Q, Andre T, Grothey A, Yothers G, Hamilton SR, Bot BM et al. Comparison of outcomes after fluorouracil-based adjuvant therapy for stages II and III colon cancer between 1978 to 1995 and 1996 to 2007: evidence of stage migration from the ACCENT database. J Clin Oncol [Internet]. 2013 Oct 10 [cited 2020 Feb 5];31(29):3656\u0026ndash;63. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/pubmed/23980089\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/pubmed/23980089\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eElferink MAG, de Jong KP, Klaase JM, Siemerink EJ, de Wilt JHW. Metachronous metastases from colorectal cancer: a population-based study in North-East Netherlands. Int J Colorectal Dis [Internet]. 2015 Feb 1 [cited 2026 Jan 23];30(2):205\u0026ndash;12. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/25503801/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/25503801/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eM\u0026eacute;ndez G, Rivera-Matos L, Shuja A. Faecal occult blood testing: a review of its use and common misutilisation. BMJ Open Gastroenterol [Internet]. 2025 Jul 17 [cited 2026 Jan 23];12(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/40675631/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/40675631/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFecal Occult Blood. Test - PubMed [Internet]. [cited 2026 Jan 23]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/30725823/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/30725823/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSoin S, Akanbi O, Ahmed A, Kim Y, Pandit S, Wroblewski I et al. Use and abuse of fecal occult blood tests: a community hospital experience. BMC Gastroenterol [Internet]. 2019 Sep 3 [cited 2026 Jan 23];19(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/31481027/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/31481027/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFisher DA, Maple JT, Ben-Menachem T, Cash BD, Decker GA, Early DS et al. Complications of colonoscopy. Gastrointest Endosc [Internet]. 2011 [cited 2026 Jan 23];74(4):745\u0026ndash;52. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/21951473/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/21951473/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCoelho-Prabhu N, Forbes N, Thosani NC, Storm AC, Pawa S, Kohli DR et al. Adverse events associated with EGD and EGD-related techniques. Gastrointest Endosc [Internet]. 2022 Sep 1 [cited 2026 Jan 23];96(3):389\u0026ndash;401.e1. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/35843754/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/35843754/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eColorectal Cancer \u0026mdash; Cancer Stat Facts [Internet]. [cited 2026 Jan 23]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://seer.cancer.gov/statfacts/html/colorect.html\u003c/span\u003e\u003cspan address=\"https://seer.cancer.gov/statfacts/html/colorect.html\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUdaikumar J, Nimmagadda R, Ingawale S, Lella VV, Vijayakumar K, Faye AS et al. Current Perspectives on Colorectal Cancer Screening and Surveillance in the Geriatric Population. Curr Gastroenterol Rep [Internet]. 2025 Dec 1 [cited 2026 Jan 23];27(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/40455318/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/40455318/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShaukat A, Ladabaum U, Kanth P, Lieberman D. AGA Clinical Practice Update on Current Role of Blood Tests for Colorectal Cancer Screening: Commentary. Clinical Gastroenterology and Hepatology [Internet]. 2025 Aug 1 [cited 2026 Jan 23];23(9):1486\u0026ndash;91. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/40267995/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/40267995/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRahib L, Smith BD, Aizenberg R, Rosenzweig AB, Fleshman JM, Matrisian LM, pancreas cancers in the United States. Projecting cancer incidence and deaths to 2030: the unexpected burden of thyroid, liver, and. Cancer Res [Internet]. 2014 Jun 1 [cited 2026 Jan 23];74(11):2913\u0026ndash;21. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/24840647/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/24840647/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWood LD, Hruban RH. Pathology and molecular genetics of pancreatic neoplasms. Cancer J [Internet]. 2012 Nov [cited 2026 Jan 23];18(6):492\u0026ndash;501. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/23187835/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/23187835/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWood LD, Canto MI, Jaffee EM, Simeone DM. Pancreatic Cancer: Pathogenesis, Screening, Diagnosis, and Treatment. Gastroenterology [Internet]. 2022 Aug 1 [cited 2026 Jan 23];163(2):386\u0026ndash;402.e1. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/35398344/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/35398344/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHe J, Ahuja N, Makary MA, Cameron JL, Eckhauser FE, Choti MA et al. 2564 resected periampullary adenocarcinomas at a single institution: Trends over three decades. HPB [Internet]. 2014 [cited 2026 Jan 23];16(1):83\u0026ndash;90. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/23472829/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/23472829/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBallehaninna UK, Chamberlain RS. The clinical utility of serum CA 19\u0026thinsp;\u0026ndash;\u0026thinsp;9 in the diagnosis, prognosis and management of pancreatic adenocarcinoma: An evidence based appraisal. J Gastrointest Oncol [Internet]. 2012 Jun [cited 2019 Nov 26];3(2):105\u0026ndash;19. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/pubmed/22811878\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/pubmed/22811878\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLewis D, Jim\u0026eacute;nez L, Chan KK, Horton S, Wong WWL. A Systematic Review of Cost-Effectiveness Studies on Pancreatic Cancer Screening. Curr Oncol [Internet]. 2025 Apr 1 [cited 2026 Jan 23];32(4). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/40277782/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/40277782/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKobayashi H, Enomoto A, Woods SL, Burt AD, Takahashi M, Worthley DL. Cancer-associated fibroblasts in gastrointestinal cancer [Internet]. Vol. 16, Nature Reviews Gastroenterology and Hepatology. Nature Publishing Group; 2019 [cited 2020 Oct 26]. pp. 282\u0026ndash;95. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/30778141/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/30778141/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalluri R. The biology and function of fibroblasts in cancer [Internet]. Vol. 16, Nature Reviews Cancer. Nature Publishing Group; 2016 [cited 2021 Feb 24]. pp. 582\u0026ndash;98. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/27550820/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/27550820/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSingh AJ, Gray JW. Chemokine signaling in cancer-stroma communications. J Cell Commun Signal [Internet]. 2021 Sep 1 [cited 2026 Jan 23];15(3):361\u0026ndash;81. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/34086259/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/34086259/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang W, Wang J, Liu C, Li Y, Sun C, Wu J et al. Crosstalk and plasticity driving between cancer-associated fibroblasts and tumor microenvironment: significance of breast cancer metastasis. J Transl Med [Internet]. 2023 Dec 1 [cited 2026 Jan 23];21(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/37978384/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/37978384/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMesker WE, Liefers GJ, Junggeburt JMC, Van Pelt GW, Alberici P, Kuppen PJK et al. Presence of a high amount of stroma and downregulation of SMAD4 predict for worse survival for stage I-II colon cancer patients. Cell Oncol [Internet]. 2009 [cited 2026 Jan 23];31(3):169\u0026ndash;78. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/19478385/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/19478385/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDowney CL, Simpkins SA, White J, Holliday DL, Jones JL, Jordan LB et al. The prognostic significance of tumour-stroma ratio in oestrogen receptor-positive breast cancer. Br J Cancer [Internet]. 2014 Apr 1 [cited 2026 Jan 23];110(7):1744\u0026ndash;7. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/24548861/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/24548861/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHe R, Li D, Liu B, Rao J, Meng H, Lin W et al. The prognostic value of tumor-stromal ratio combined with TNM staging system in esophagus squamous cell carcinoma. J Cancer [Internet]. 2021 Jan 1 [cited 2026 Jan 23];12(4):1105\u0026ndash;14. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/33442408/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/33442408/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZheng H, Liu H, Ge Y, Wang X. Integrated single-cell and bulk RNA sequencing analysis identifies a cancer associated fibroblast-related signature for predicting prognosis and therapeutic responses in colorectal cancer. Cancer Cell Int [Internet]. 2021 Dec 1 [cited 2026 Jan 23];21(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/34670584/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/34670584/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang F, Li Z, Xu T, Zhang Q, Ma T, Li S et al. A comprehensive multi-omics analysis identifies a robust scoring system for cancer-associated fibroblasts and intervention targets in colorectal cancer. J Cancer Res Clin Oncol [Internet]. 2024 Mar 1 [cited 2026 Jan 23];150(3). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/38478111/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/38478111/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang J, Chen M, Fang C, Luo P. A cancer-associated fibroblast gene signature predicts prognosis and therapy response in patients with pancreatic cancer. Front Oncol [Internet]. 2022 Nov 18 [cited 2026 Jan 23];12. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/36465388/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/36465388/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eUcaryilmaz Metin C, Ozcan G. Comprehensive bioinformatic analysis reveals a cancer-associated fibroblast gene signature as a poor prognostic factor and potential therapeutic target in gastric cancer. BMC Cancer [Internet]. 2022 Dec 1 [cited 2026 Jan 23];22(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/35739492/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/35739492/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSi Y, Zhao Z, Meng X, Zhao K. RNA-seq and bulk RNA-seq data analysis of cancer-related fibroblasts (CAF) in LUAD to construct a CAF-based risk signature. Sci Rep [Internet]. 2024 Dec 1 [cited 2026 Jan 23];14(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/39369095/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/39369095/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiot S, Balas J, Aubert A, Prigent L, Mercier-Gouy P, Verrier B et al. Stroma Involvement in Pancreatic Ductal Adenocarcinoma: An Overview Focusing on Extracellular Matrix Proteins. Front Immunol [Internet]. 2021 Apr 6 [cited 2026 Feb 18];12. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/33889150/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/33889150/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQian C, Ren L, Zhu S, Chen F, Shen H, Hu X et al. Current perspectives and trends in colorectal cancer and cancer-associated fibroblasts: a review and bibliometric analysis. Front Immunol [Internet]. 2025 [cited 2026 Feb 18];16. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/40677714/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/40677714/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKalluri R, LeBleu VS. The biology, function, and biomedical applications of exosomes. Science [Internet]. 2020 Feb 7 [cited 2026 Jan 23];367(6478). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/32029601/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/32029601/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMathieu M, Martin-Jaular L, Lavieu G, Th\u0026eacute;ry C. Specificities of secretion and uptake of exosomes and other extracellular vesicles for cell-to-cell communication. Nat Cell Biol [Internet]. 2019 Jan 1 [cited 2026 Jan 23];21(1):9\u0026ndash;17. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/30602770/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/30602770/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYoshii S, Hayashi Y, Iijima H, Inoue T, Kimura K, Sakatani A et al. Exosomal microRNAs derived from colon cancer cells promote tumor progression by suppressing fibroblast TP53 expression. Cancer Sci. 2019.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMito I, Takahashi H, Kawabata-Iwakawa R, Horikawa M, Ida S, Tada H et al. Tumor-derived exosomes elicit cancer-associated fibroblasts shaping inflammatory tumor microenvironment in head and neck squamous cell carcinoma. Oral Oncol [Internet]. 2023 Jan 1 [cited 2026 Jan 23];136. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/36462328/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/36462328/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRinguette Goulet C, Bernard G, Tremblay S, Chabaud S, Bolduc S, Pouliot F. Exosomes Induce Fibroblast Differentiation into Cancer-Associated Fibroblasts through TGFβ Signaling. Mol Cancer Res [Internet]. 2018 [cited 2019 Sep 30];16(7):1196\u0026ndash;204. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/pubmed/29636362\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/pubmed/29636362\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRen J, Ding L, Zhang D, Shi G, Xu Q, Shen S et al. Carcinoma-associated fibroblasts promote the stemness and chemoresistance of colorectal cancer by transferring exosomal lncRNA H19. Theranostics [Internet]. 2018 [cited 2019 Sep 30];8(14):3932\u0026ndash;48. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/pubmed/30083271\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/pubmed/30083271\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHu JL, Wang W, Lan XL, Zeng ZC, Liang YS, Yan YR et al. CAFs secreted exosomes promote metastasis and chemotherapy resistance by enhancing cell stemness and epithelial-mesenchymal transition in colorectal cancer. Mol Cancer [Internet]. 2019 May 7 [cited 2020 Nov 11];18(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/31064356/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/31064356/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHerrera M, Llorens C, Rodr\u0026iacute;guez M, Herrera A, Ramos R, Gil B et al. Differential distribution and enrichment of non-coding RNAs in exosomes from normal and Cancer-associated fibroblasts in colorectal cancer. Mol Cancer [Internet]. 2018 Aug 3 [cited 2018 Oct 24];17(1):114. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://molecular-cancer.biomedcentral.com/articles/\u003c/span\u003e\u003cspan address=\"https://molecular-cancer.biomedcentral.com/articles/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.1186/s12943-018-0863-4\u003c/span\u003e\u003cspan address=\"10.1186/s12943-018-0863-4\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHerrera M, Berral-Gonz\u0026aacute;lez A, L\u0026oacute;pez-Cade I, Galindo-Pumari\u0026ntilde;o C, Bueno-Fortes S, Mart\u0026iacute;n-Merino M et al. Cancer-associated fibroblast-derived gene signatures determine prognosis in colon cancer patients [Internet]. Vol. 20, Molecular Cancer. BioMed Central Ltd; 2021 [cited 2021 Jun 15]. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/33926453/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/33926453/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMocci E, Guillen-Ponce C, Earl J, Marquez M, Solera J, Salazar-L\u0026oacute;pez MT et al. PanGen-Fam: Spanish registry of hereditary pancreatic cancer. Eur J Cancer [Internet]. 2015 Sep [cited 2019 Nov 26];51(14):1911\u0026ndash;7. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/pubmed/26212471\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/pubmed/26212471\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWan JCM, Massie C, Garcia-Corbacho J, Mouliere F, Brenton JD, Caldas C et al. Liquid biopsies come of age: towards implementation of circulating tumour DNA. Nat Rev Cancer [Internet]. 2017 Apr 1 [cited 2026 Jan 23];17(4):223\u0026ndash;38. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/28233803/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/28233803/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHall C, Clarke L, Pal A, Buchwald P, Eglinton T, Wakeman C et al. A Review of the Role of Carcinoembryonic Antigen in Clinical Practice. Ann Coloproctol [Internet]. 2019 [cited 2026 Jan 23];35(6):294\u0026ndash;305. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/31937069/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/31937069/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBallehaninna UK, Chamberlain RS. Serum CA 19\u0026thinsp;\u0026ndash;\u0026thinsp;9 as a Biomarker for Pancreatic Cancer-A Comprehensive Review. Indian J Surg Oncol [Internet]. 2011 Jun 1 [cited 2026 Jan 23];2(2):88\u0026ndash;100. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/22693400/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/22693400/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCui M, Wang H, Yao X, Zhang D, Xie Y, Cui R et al. Circulating MicroRNAs in Cancer: Potential and Challenge. Front Genet [Internet]. 2019 Jul 18 [cited 2026 Jan 23];10. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/31379918/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/31379918/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOh HJ, Imam-Aliagan AB, Kim Y, Bin, Kim HJ, Izaguirre IA, Sung CK et al. Clinical applications of circulating biomarkers in non-small cell lung cancer. Front Cell Dev Biol [Internet]. 2024 [cited 2026 Jan 23];12. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/39239557/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/39239557/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDuffy MJ, Crown J. Circulating Tumor DNA as a Biomarker for Monitoring Patients with Solid Cancers: Comparison with Standard Protein Biomarkers. Clin Chem [Internet]. 2022 Nov 1 [cited 2026 Jan 23];68(11):1381\u0026ndash;90. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/35962648/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/35962648/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFiala C, Diamandis EP. Utility of circulating tumor DNA in cancer diagnostics with emphasis on early detection. BMC Med [Internet]. 2018 Oct 2 [cited 2026 Jan 23];16(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/30285732/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/30285732/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang HW, Shi Y, Liu J, Bin, Wang HM, Wang PY, Wu ZJ et al. Cancer-associated fibroblast-derived exosomal microRNA-24-3p enhances colon cancer cell resistance to MTX by down-regulating CDX2/HEPH axis. J Cell Mol Med [Internet]. 2021 Apr 1 [cited 2026 Jan 23];25(8):3699\u0026ndash;713. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/33621425/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/33621425/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao J, Zhang J, Liu J, Zan L. Exosomal transfer of miR-223-3p from carcinoma-associated fibroblasts promotes the malignant properties and chemoresistance of colon cancer cells by targeting NF2/Hippo signaling. Med Oncol [Internet]. 2025 Nov 1 [cited 2026 Jan 23];42(11). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/41026317/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/41026317/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBhome R, Goh RW, Bullock MD, Pillar N, Thirdborough SM, Mellone M et al. Exosomal microRNAs derived from colorectal cancer-associated fibroblasts: role in driving cancer progression. Aging [Internet]. 2017 [cited 2019 Sep 30];9(12):2666\u0026ndash;94. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.ncbi.nlm.nih.gov/pubmed/29283887\u003c/span\u003e\u003cspan address=\"http://www.ncbi.nlm.nih.gov/pubmed/29283887\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eQi R, Bai Y, Li K, Liu N, Xu Y, Dal E et al. Cancer-associated fibroblasts suppress ferroptosis and induce gemcitabine resistance in pancreatic cancer cells by secreting exosome-derived ACSL4-targeting miRNAs. Drug Resist Updat [Internet]. 2023 May 1 [cited 2026 Jan 23];68. Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/37003125/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/37003125/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRichards KE, Zeleniak AE, Fishel ML, Wu J, Littlepage LE, Hill R. Cancer-associated fibroblast exosomes regulate survival and proliferation of pancreatic cancer cells. Oncogene. 2017;36(13):1770\u0026ndash;8.\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEscrich V, Romero-Aranda C, L\u0026oacute;pez R, de Toro M, Metola \u0026Aacute;, Ezcurra B et al. Unprocessed snRNAs Are a Prognostic Biomarker and Correlate with a Poorer Prognosis in Colorectal Cancer. Cancers (Basel) [Internet]. 2024 Jul 1 [cited 2026 Jan 23];16(13). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/39001402/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/39001402/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGarcia-Borja E, Siegl F, Mateu R, Slaby O, Sedo A, Busek P et al. Critical appraisal of the piRNA-PIWI axis in cancer and cancer stem cells. Biomark Res [Internet]. 2024 Dec 1 [cited 2026 Jan 23];12(1). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/38303021/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/38303021/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSohn EJ, Han ME, Park YM, Kim YH, Oh SO. The potential of piR-823 as a diagnostic biomarker in oncology: A systematic review. PLoS One [Internet]. 2023 Dec 1 [cited 2026 Jan 23];18(12). Available from: \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://pubmed.ncbi.nlm.nih.gov/38060527/\u003c/span\u003e\u003cspan address=\"https://pubmed.ncbi.nlm.nih.gov/38060527/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Cancer-associated fibroblasts, Extracellular Vesicles, Small non-coding RNAs, Liquid biopsy, Colon cancer, Pancreatic cancer, Tumor microenvironment, Risk stratification","lastPublishedDoi":"10.21203/rs.3.rs-8948517/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8948517/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cb\u003eBackground:\u003c/b\u003e\u003c/p\u003e \u003cp\u003eColon cancer (CC) and pancreatic ductal adenocarcinoma (PDAC) are among the most lethal gastrointestinal (GI) malignancies, largely due to limitations in early detection and risk stratification. While most circulating biomarkers currently under clinical evaluation are tumor cell\u0026ndash;derived, the tumor stroma, and particularly cancer-associated fibroblasts (CAFs), represents an underexplored source of clinically informative signals.\u003c/p\u003e\u003cp\u003e\u003cb\u003eMethods:\u003c/b\u003e\u003c/p\u003e \u003cp\u003eBased on our prior characterization of CAF-derived extracellular vesicles (EVs) cargo, we evaluated two small nuclear RNAs (snRNAs) and one PIWI-interacting RNA (piRNA) as EV-associated biomarkers in peripheral blood samples obtained prior to surgical intervention. EV RNA was isolated from serum samples of patients with colon pathology (including benign, premalignant, and malignant lesions), PDAC patients, and healthy donors. Small non-coding RNA (sncRNA) expression was quantified by RT\u0026ndash;qPCR. Multiple machine learning models were applied to assess diagnostic and prognostic performance. The prognostic relevance of predicted target genes regulated by these sncRNAs was further evaluated in a large, publicly available CC cohort from The Cancer Genome Atlas.\u003c/p\u003e\u003cp\u003e\u003cb\u003eResults:\u003c/b\u003e\u003c/p\u003e \u003cp\u003eThe sncRNA panel discriminated CC patients from healthy individuals with high diagnostic accuracy (AUC\u0026thinsp;=\u0026thinsp;0.866) and enabled discrimination of benign and premalignant colonic lesions. In PDAC, the same signature achieved robust diagnostic performance (AUC\u0026thinsp;=\u0026thinsp;0.782), and combined analysis of CC and PDAC supported its ability to detect GI malignancy. In stage I\u0026ndash;III CC patients, pre-operative EV sncRNA levels were associated with disease relapse, allowing stratification into low- and high-risk groups (AUC\u0026thinsp;=\u0026thinsp;0.725; log-rank p\u0026thinsp;=\u0026thinsp;0.009). Analysis of 43 predicted sncRNA target genes in a TCGA cohort of 597 CC patients revealed strong prognostic value for overall survival (AUC\u0026thinsp;=\u0026thinsp;0.813; log-rank p\u0026thinsp;\u0026lt;\u0026thinsp;0.001).\u003c/p\u003e\u003cp\u003e\u003cb\u003eConclusions:\u003c/b\u003e\u003c/p\u003e \u003cp\u003eCAF-associated EV snRNAs and piRNAs can be detected in peripheral blood and provide clinically relevant diagnostic and pre-operative prognostic information in CC and PDAC. By capturing stromal-derived signals distinct from tumor cell\u0026ndash;intrinsic biomarkers, these molecules offer a complementary liquid biopsy strategy with potential utility for early detection, pre-surgical risk stratification, and personalized patient management. Together, our findings support the tumor stroma as a systemic source of accessible biomarkers and highlight CAF-associated EV sncRNAs as promising tools for improving clinical decision-making in GI cancers.\u003c/p\u003e","manuscriptTitle":"Stroma-associated extracellular vesicles snRNAs and piRNAs as non-invasive diagnostic and prognostic biomarkers in colon and pancreatic cancer","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-03-10 10:55:45","doi":"10.21203/rs.3.rs-8948517/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"75a304aa-d029-46c6-af24-8c8cef83f72b","owner":[],"postedDate":"March 10th, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2026-04-10T22:53:45+00:00","versionOfRecord":[],"versionCreatedAt":"2026-03-10 10:55:45","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8948517","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8948517","identity":"rs-8948517","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}