Systems biology-driven miR-mRNA integration identifies potential steroid- refractoriness biomarkers in ulcerative colitis and reveals a conserved mechanism across species

Systems biology-driven miR-mRNA integration identifies potential steroid- refractoriness biomarkers in ulcerative colitis and reveals a conserved mechanism across species | 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 Article Systems biology-driven miR-mRNA integration identifies potential steroid- refractoriness biomarkers in ulcerative colitis and reveals a conserved mechanism across species Iris Ginés, Roger Suau, Juan Enrique Naves, Carla Bernal, Laura Clua, and 13 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-7401947/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 and Aims: Approximately 60% of patients with ulcerative colitis (UC) exhibit steroid resistance or dependence, highlighting the need for reliable biomarkers to predict therapeutic response. This study employed a systems biology and machine learning approach to integrate microRNA (miR) and mRNA expression data from both rectal biopsies and plasma samples from UC patients undergoing corticosteroid therapy, aiming to uncover the molecular mechanisms underlying steroid refractoriness. Methods: Whole-transcriptome and miR profiling were performed at baseline and after three days of corticosteroid therapy. Corticosteroid-treated UC patients were classified as responders (R) or non-responders (NR) after seven days of treatment. Mathematical modelling and protein-miR interaction mapping were used to identify mechanistically relevant biomarker candidates. Selected findings were validated in a TNBS-induced colitis mouse model. Results: Key transcriptional co-regulators such as NCOA3, CBP, NCOR1, and NRIP1 were differentially expressed between R and NR, influencing glucocorticoid receptor (GCR) signalling. Multiple miRNAs, including miR-145-5p, miR-10b-5p, and miR-16-5p, were identified as potential biomarkers and regulators of inflammatory and GCR-related pathways. The cross-correlation between plasma and tissue miRs revealed consistent molecular patterns, some of which were also conserved in the murine model, supporting the existence of cross-species steroid response mechanisms. Conclusions: This integrative multi-omic approach provides new insights into molecular steroid-refractoriness in UC and offers a promising framework for developing predictive tools and advancing personalised therapeutic strategies in inflammatory bowel disease. Health sciences/Biomarkers Biological sciences/Computational biology and bioinformatics Health sciences/Diseases Health sciences/Gastroenterology Biological sciences/Immunology Ulcerative colitis TNBS-colitis Biomarkers Steroid-refractoriness miR Machine Learning Systems Biology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Ulcerative colitis (UC) is a chronic inflammatory disease of the colon characterized by periods of relapse and remission, significantly impacting patients' quality of life (1). While corticosteroids remain a cornerstone in the management of UC flares, approximately 60% of patients exhibit steroid resistance or dependence, requiring alternative therapeutic strategies (2). Identifying reliable biomarkers to predict corticosteroid response is crucial for optimising patient management, minimising unnecessary steroid exposure, and preventing complications such as osteoporosis, infections, and metabolic disturbances (3). Glucocorticoids exert their effects primarily through the glucocorticoid receptor (GCR), a ligand-activated transcription factor that regulates inflammation by modulating gene expression. However, alterations in GCR signalling, including receptor isoform imbalances and impaired nuclear translocation, have been associated with corticosteroid resistance in UC (4). Additionally, transcriptional co-regulators such as NCOA3, CBP, and NCOR1 play a pivotal role in fine-tuning GCR activity, influencing inflammatory pathways and treatment response (5). Beyond transcriptional regulation, non-coding RNAs, particularly microRNAs (miRs), have emerged as key post-transcriptional regulators of GCR function, NF-κB signalling, and cytokine expression, further shaping corticosteroid responsiveness (6). In line with these findings, our research group (7)applied a systems biology approach to investigate the mechanisms underlying steroid refractoriness in patients with UC. By integrating transcriptomic data, the study identified 18 proteins associated with corticosteroid response, focusing on ANP32E. This chaperone protein, which is involved in histone exchange and GCR-induced transcription, was expressed differentially by responders (R) and non-responders (NR). These findings suggested that alterations in chromatin remodelling, NF-κB signalling, and angiogenesis contribute to steroid failure in UC. However, this study did not assess post-transcriptional regulatory mechanisms, such as the miR-mediated modulation of these pathways, nor did it validate findings using complementary in vivo models. Building upon these insights, our study employed a systems biology and machine learning approach to integrate mRNA and miR expression profiles from both rectal biopsies and, for the first time in our research, from plasma samples of UC patients undergoing corticosteroid therapy. Unlike our previous integrative analyses, which were limited to mucosal tissue, this study incorporated circulating microRNAs as a novel layer of molecular information. We aimed to explore the plasma microRNAs profiles associated with corticosteroid refractoriness, integrating these with rectal and in vivo data to generate testable mechanistic hypotheses and guide future biomarker development. By analysing samples collected at baseline and after three days of corticosteroid treatment, we identified disparities in miR-mRNA interactions according to corticosteroid effect, as well as novel miR profiles with high predictive potential for treatment failure. Additionally, we described key molecular findings in a TNBS-induced colitis mouse model supporting the existence of cross-species conserved mechanisms. Collectively, this integrative strategy provides new insights into molecular steroid-refractoriness in UC, and may generate clinically useful tools which can help to guide subsequent clinical studies to improve the management of this chronic inflammatory bowel disease. Importantly, this approach aimed not only to identify predictive markers, but also to mechanistically contextualize them within the glucocorticoid response network, enhancing their translational relevance. Materials and Methods Patients and study design The present study complements the previous identification of a steroid-refractoriness-linked molecular mechanism in UC (7) by integrating intestinal mRNA and miR profiles. In brief, rectal biopsies and peripheral blood were collected from 24 UC patients prior to treatment (baseline), and three days after the initiation of corticosteroid therapy—either 0·75–1 mg/kg/day of oral prednisone for moderate disease flares, or 1 mg/kg/day of intravenous prednisolone in cases of severe activity. Samples from 10 healthy controls were also included for comparison. Disease activity was assessed clinically at day seven of glucocorticoid treatment, classifying the patients as steroid-responsive (responders, R) if they had mild activity or inactive disease according to the severity criteria of the Montreal classification, or steroid-refractory (non-responders, NR) if moderate or severe clinical activity was still present or if rescue therapy was needed) (8). Whole-genome mRNA expression analysis was performed on rectal biopsies collected before and after treatment using oligonucleotide microarrays (HT-12 v4, Illumina, USA). In parallel, the expression of approximately 3,600 known miRs was profiled using next-generation sequencing (TruSeq small RNA, Illumina, USA). Technical details for both assays are as previously described (7). The raw data for both mRNA microarrays and miR sequencing were deposited in NCBI’s Gene Expression Omnibus (GEO), assigning to them the GEO Series accession number GSE114527 and GSE114591, respectively in compliance with MIAME standards (https://www.ncbi.nlm.nih.gov/geo/info/MIAME. html). Microarray data were normalized with the Neqc method and differential expression was analysed with Linear Models for Microarray data (LIMMA) (9,10), considering them significant if False Discovery Rate (FDR) ≤0·01. Gene information was one-to-one mapped to proteins for its introduction into mathematical models, including differential FCs and significant protein or cluster correlations between the control and intervention cohorts (11). Circulating miR sequencing The previous study described sequencing methods for miRs measurement in rectal biopsies (11). In the present study, small RNAs (<1000nt) were isolated from 300µl of plasma using a miRCURY RNA isolation kit (EXIQON, Cat. No.300112), following the manufacturer’s instructions. Plasma samples were collected and stored at -80⁰C until all samples had been collected, then the plasmas were thawed and used at the same time. MiR libraries (135-160bp) were prepared starting from total RNA extracted from 200 ul of plasma using the TruSeq small RNA library preparation kit (Illumina, USA) with 15 cycles. Individual libraries were quantified by DNA-1000 Bioanalyzer assay (Agilent). Equal mass library pool was size selected automatically using PippinPrep 3% agarose cassettes with PippinPrep software v.6·0. QIAquick column purifications were performed prior to and after size selection. The size selected pool was quantified by qPCR with Kappa Library Quantification Kit (Roche). Sequencing was performed on a HiSeq2500 system (Illumina Inc.) at VHIO using adapted protocols for TruSeq SR cluster kit v3 (Part#15023335)) following cBot™ (Part#15006165) and HiSeq2500 System (Part#15035786) guidelines. Raw data was analysed by Real Time Analysis software v. 1·17·21·3 and Sequence Analysis Viewer software v. 1·8·20 (Illumina Inc.). Raw data from this sequencing were deposited in NCBI: GEO Series accession number GSE122618. miR data analysis and mapping After the quality control and normalization procedures, the number of reads of a particular miR sequence was counted, discarding those without sufficient signal, as previously described (8,12,13). The counts were normalized through the weighted trimmed mean of M-values (12), and differential expression was calculated by limma’s voom methods (12,13). To make the plasmatic miR profiles and those obtained from the rectal biopsies comparable, the criteria used for the compilation and mapping in Lorén V et al. was recalculated, considering fold change (FC) between R and NR differentially expressed and functionally significant if FDR<0·05, or classifier p-value<0·05, or non-adjusted p-value1·5. Selected miRs were mapped to protein by means of MirTarBase (14) and MiRDB Database (15,16). Each miR-protein pair received a score based on the evidence rates(11) (7). When a protein was targeted by more than one miR, a list of reliable biological activity was composed. Those protein effects that were inconsistent with this list received a negative score, and those that agreed received a positive one. This score also rated the concordance with the experimental data and miR-protein correlations. Finally, scores were summarized in a matrix that integrates both proteins and miRs. When the number of proteins per miR was too large, only up to 40 proteins with the best scores were taken. Additionally, those proteins with any contradictory information (e.g. duplicated targets) were removed, and the resulting list was stored as protein restrictions to be introduced into the models. A schematic workflow had previously been described by Lorén et al (7). Mathematical Modelling Using artificial intelligence and pattern recognition techniques, mathematical models that could suggest mechanistic hypotheses following biological environments were generated. This process was carried out by transforming biological information into mathematical models, as previously described by Lorén et al. (11). In summary, information about glucocorticoids and UC was compiled in a hand-curated database (Biological Effectors Database, Anaxomics, Barcelona, Spain), matching biological processes (including adverse events, indications, diseases, and pathways) to their molecular effectors. These molecular relationships were updated from public sources (KEGG, REACTOME, INTACT, BIOGRID, MINT, PubMed and DrugBank). Thus, a biological network focused on glucocorticoid targets and UC effectors was defined. The biological map was transformed into a mathematical model, mimicking current knowledge and predicting new data (Therapeutic Performance Mapping System, Anaxomics). This was achieved by training the biological network with the “Truth Table”, a collection of known stimulus-response relationships that act as mathematical restrictions. Once these rules or "truths" were established, transcriptome information was added to refine the model (mRNA and miR that meet the selection requirements described above). A mathematical model challenge with glucocorticoid traced a molecular response network that gives a relative value to each interaction between proteins (node). With this computational Systems Biology-based approach, Lorén et al. characterized the mechanism of action (MoA) related to steroid-refractoriness in UC (supplementary material & methods). miR biomarker identification Both biopsy and plasma miR profiles were statistically analysed to select those that possessed greater potential as biomarkers. These calculations were carried out with all the sequenced miRs or with those miRs most closely related to the MoA, both at baseline and on day three. The probability that each miR studied was related to the MoA was calculated according to its validated target proteins on R vs. NR model. The expression sample by sample of each miR was used to identify the miRs that better classified the samples in R and NR (explained in detail in the previous section “miR data analysis and mapping”). These miRs were then combined with each other or with clinical variables (e.g. C-Reactive Protein (CRP), smoking, delay in GC treatment, previous oral or topical treatment with mesalazine) in order to identify the combinations that best classified R and NR samples. Different statistical approaches were used to identify potential sets of biomarkers from miR data (Table 1). The following parameters were used in order to select the best miRs classifiers, alone or in combination: a) Accuracy (proportion of samples correctly classified); b) Generalization Capability (by using a Leave-One-Out strategy, the probability of correctly predicting any independent samples) ), c) Sensitivity (= True Positives / Positives ), d) Specificity (= True Negatives / Negatives ), d) Precision (= True Positives / Positive calls ), e) Negative predictive value (NPV) (= True Negatives / Negative calls ) (7). Biopsy-plasma correlation The miRs with a greater probabilityof affecting the MoA of corticosteroid in the treatment of UC were identified among miRs differentially expressed by R and NR in plasma samples. Then, the targets of these miRs were filtered to find correlations related to the process under study: only the targets present in the model of corticosteroid treatment of UC were considered. These MoA related plasma miRs (pmiRs) and related targets were compared with different sources of information from biopsy (Figure 1): Plasma miRs vs. biopsy differential miRs Plasma miRs MoA related targets vs. biopsy differential miRs targets. Plasma miRs MoA related targets vs. proteins translated from differential mRNAs. Plasma miRs MoA related targets vs. key proteins obtained from the MoA analysis in the previous phases of the project. The correlation described in point 4 was also assessed in an indirect setting, i.e. evaluating if there is any interaction in Anaxomics databases relating both groups of proteins (distance 1, D1). This protocol was performed both at baseline and at day three of treatment. The In silico functional analysis of selected miRNAs included target prediction using miRTarBase, DIANA, and miRDB, followed by functional pathway enrichment using Gene Ontology (GO), KEGG, and STRING databases. A selection score was developed to prioritize mRNA targets based on corticosteroid response relevance, interspecies conservation, and evidence from human datasets. This integrated analysis allowed us to identify the most functionally relevant miRNA-mRNA interactions in the context of glucocorticoid sensitivity and inflammation resolution in colitis. Validation of candidate miR biomarkers in a TNBS-induced colitis mouse model Ten to twelve-week-old C57BL/6J mice (male and female) were purchased from Jackson Laboratory (Charles River; Spain) and were housed in the Germans Trias i Pujol Research Institute (IGTP) conventional facility. The studies were conducted in compliance with animal research guidelines. All the experiments were approved by the Institutional Animal Care and Use Committee of the IGTP and authorized by the Catalan Government (no. 9933). Colitis was induced by intrarectal instillation of 3 mg TNBS in 100 µL 50% ethanol (17). Mice were randomized into four groups (n=10/group; sex-balanced): (i) healthy controls; (ii) TNBS colitis (endpoint day 3); (iii) TNBS + vehicle (PBS, subcutaneous, days 4–7); (iv) TNBS + dexamethasone (1.5 mg/kg s.c., days 4–7). Dexamethasone dosing was based on prior dose-response screenings. Inflammatory response was monitored longitudinally using bioluminescence imaging (BLI), as previously described by Suau et al.Suau et al . (17). Images were acquired at multiple time points and normalized to each animal’s pre-induction baseline signal. Inflammation levels were expressed as FC relative to this baseline. To assess treatment effect, we also calculated the average FC between colitis onset (day 0) and treatment initiation (day 4) and compared it to values during the treatment phase (days 4–7). To objectively evaluate treatment response, a composite BLI-based score (range 0–4) was established, based on dynamic changes in FC after corticosteroid administration: if FC decreased (i.e., negative FC) on two separate days during days 4–7 (+1 point); if the average FC over the entire treatment period was negative (+1 point); if FC was negative at day 7 (+2 point). Based on the total score: a) Mice scoring 0 or 1 were classified as non-responders (NR); b) mice scoring 2 to 4 were classified as responders (R). After euthanasia (endpoint day 7), colon samples were cpllected, preserved in RNAlater, and stored at −80°C. Total RNA was extracted (miRNeasy Mini kit/QIAcube system, QIAGEN, Germany) from homogenized (gentleMACS™; Miltenyi Biotech, Germany) 25 mg colon tissue, and integrity was confirmed (RIN ≥6.5, Agilent Bioanalyzer, Nano kit and SmallRNA kit, Agilent Technologies, USA). For mRNA quantification, 1 µg of RNA was retrotranscribed (PrimeScript RT kit, Takara, Japan), and qRT-PCR was performed (TaqMan™ probes for Creb1, Nr3c1, Smad7, Vegfa; ThermoFisher Scientific, Spain) using a LightCycler480 (Roche Diagnostics, Switzerland). Expression was normalized to Gapdh and B2m. For miR quantification, reverse transcription and qRT-PCR were performed using the TaqMan™ Advanced miRNA cDNA Synthesis Kit and specific probes (mmu-miR-224-5p, -10b-5p, -218-5p, -145a-5p, -183-5p, -16-5p). Data were normalized to mmu-miR-30e-5p and mmu-miR-103-3p. Data from qRT-PCR were again calculated using the 2 −ΔΔCt method (18). Standard statistics All quantitative data were checked for normality using a Shapiro-Wilk test. We also applied the Mann–Whitney U test in pairwise comparisons between different animal groups at the same time point. The comparisons of the qRT-PCR data between treatment groups and sexes were performed with the Kruskal–Wallis test with a post hoc Dunn’s test analysis or a Mann–Whitney U test. Comparisons with p ≤ 0·05 were statistically significant. All these analyses were performed with R software version 4·2·0. Results Transcriptomic and microRNA profiling distinguishes corticosteroid responders from non-responders in UC patients This study included 24 patients with UC, categorised at day seven as R (n=13) or NR (n=11) to corticosteroid treatment. To investigate the mechanisms underlying response to corticosteroids, mRNA and miR expression levels were analysed at baseline (day 0) and on day three of treatment. Two types of potential biomarkers were identified: protein biomarkers derived from mRNA and miR data. Systems biology approaches were employed to minimize false discoveries and enhance the robustness of biomarker identification. Protein-coding transcript biomarkers linked to corticosteroid responsiveness Candidate biomarker transcripts coding for protein were identified by integrating mRNA transcriptome data with models built using both mRNA and miR information. Separate analyses were conducted for baseline and day three samples, comparing R and NR groups. A bibliographic review was performed to assess the relevance of identified proteins to UC pathology or corticosteroid treatment. At baseline, 18 protein biomarker combinations were identified as the best candidates to differentiate between R and NR patients. Similarly, 15 combinations were identified at day three. These combinations were evaluated for accuracy and generalization capability using Therapeutic Performance Mapping System strategies (19), which yielded the most reliable candidates. Table 2 summarizes the top-performing combinations of three protein biomarkers for each time point. This systems biology analysis identified proteins linked to both UC and corticosteroid treatment at basal and day three levels. These proteins not only serve as biomarkers but also provide mechanistic insights into the response to corticosteroids. In this study, four key co-activators and co-repressors affecting GCR transcriptional activity were identified: NCOA3, CBP, NCOR1, and NRIP1. These proteins influence transcriptional activity related to corticosteroid response and inflammatory pathways. These findings may help stratify patients at baseline, allowing for the early identification of individuals at risk of steroid refractoriness and guiding personalised treatment strategies. MicroRNAs modulating steroid response: identification and biomarker potential To explore the post-transcriptional regulation mechanisms involved in corticosteroid response, we analyzed miRs from both plasma and rectal biopsies. Our goal was to identify miRs that not only differentiate R from NR, but that are also mechanistically integrated into the corticosteroid MoA. miRs influence gene expression pathways linked to GCR signalling and inflammation. Plasma and biopsy samples were analyzed at baseline and day three. Two complementary approaches were applied: (1) using all sequenced miRs and (2) focusing on MoA-related miRs. The best miR combinations were selected based on classification accuracy, generalization capability, and minimal descriptor count (Table 3). Mechanistic insights into miRNA-mediated regulation of corticosteroid pathways A mechanistic analysis of the best miR biomarker combinations was performed to provide further biological justification. The identified miRs were analysed for their interactions with key molecular pathways involved in corticosteroid response. Figures 2 and 3 illustrate the relationship between miR targets and proteins within the previously reported MoA at baseline and day three, respectively. At baseline, the identified plasma miRs, such as hsa-miR-10a-5p and hsa-miR-324-5p, influenced GCR function by targeting chaperones, such as HS71A/B and co-activators essential for GCR activity, like NCOAs (Figure 2) . These miRs also regulated NF-κB, TNF-α and VEGF pathways, which are involved in inflammation and corticosteroid resistance. Specific targets included ETS1 and P53, key regulators of transcriptional and apoptotic pathways. At day three , miR-20a-5p and miR-181a were identified as key regulators of GCR activity, while hsa-miR-433 was linked to PI3K modulation through its interaction with the adaptor protein GRB2 (Figure 3) . These findings highlight the dynamic nature of miR regulation in the response to corticosteroids. The regulatory roles of these miRs over GCR-related pathways suggest that they are not only passive markers but could actively modulate corticosteroid responsiveness in UC. Circulating–tissue miRNA concordance supports systemic biomarker validity Statistically significant miRs from plasma samples (adjusted p-value < 0·05) were correlated with miRs, mRNA, and model information from biopsy samples at baseline and day three. At baseline(Figure 4), five plasma miRs (hsa-miR-194-5p, hsa-miR-145-5p, hsa-miR-216a-5p, miR-224-5p and hsa-miR-487a-3p) were identified as potentially affecting corticosteroid response. These miRs were related to biopsy findings through common protein targets and relationships with key proteins from biopsy models, such as VEGFA. Specifically, this protein was shown to be targeted by the plasma hsa-miR-145-5p and biopsy hsa-miR-16-5p (Figure 4A and B). This cross-validation between tissue and circulating miRs reinforces their potential as minimally invasive biomarkers for monitoring or predicting treatment outcomes. At day three,hsa-miR-214-3p was found to be significantly differential and was correlated mechanistically with rectal biopsy miRs and key proteins involved in the response to corticosteroids (Figure 5). Briefly, hsa-miR-214-3p shared mRNA targets with hsa-miR-625-3p, hsa-miR-29a-3p, miR-423-3p and hsa-miR-10b-5p. Many of these mRNA targets were at the same time related to the MoA identified proteins, especially P53, EZH2 and CTNB1. Validation in TNBS-colitis model reveals conserved corticosteroid-regulatory miRNAs Bioluminescence and macroscopic indices Based on the bioluminescence FC relative to baseline, mice were classified as R or NR, following a methodology used in a previous study (17). The BLI results showed that the R were able to reduce their intestinal inflammation from the sixth day of treatment (Figure 6A). Consistently, the colons of R mice to corticosteroids were significantly lighter and longer compared to non-treated mice, reflecting reduced oedema and ulceration (Figure 6B). miRNA profiling Six miRNAs previously linked to human UC were conserved and quantifiable in the TNBS model (Figure 6C). miR‑10b‑5p, miR‑145a‑5p and miR‑16‑5p were markedly down‑regulated in R versus NR mice (p < 0·05), suggesting their association with steroid efficacy. miR‑224‑5p and miR‑183‑5p were up‑regulated at day seven irrespective of response, indicating a broader association with inflammation. Target‑gene expression Creb1 (a shared target of miR‑10b‑5p/16‑5p/218‑5p) was up‑regulated in all TNBS groups (Figure 6D), echoing its pro‑inflammatory role in human UC. Smad7 , potentially under negative post-transcriptional regulation by miR-16-5p (Figure 6E), was significantly increased in R mice, aligning with the resolution of TGF-β signalling. In contrast, Nr3c1 (Gcr) expression decreased in NR mice, reflecting the glucocorticoid resistance observed in patients. Vegfa expression remained unchanged, suggesting species-specific regulatory differences miRNA–mRNA correlations Finally, integrative correlation analyses across all experimental groups (Figure 6D) revealed consistent inverse relationships between specific miRNAs and key proteins involved in corticosteroid response. Notably, miR-183-5p showed a significant negative correlation with Gcr (ρ = –0·47, p = 0·01), while miR-218-5p inversely correlated with Creb1 (ρ = –0·38, p = 0·037). A similar trend was observed for miR-10b-5p, which also negatively correlated with Creb1(ρ = –0·30, p = 0·098), and miR-16-5p, which inversely correlated with Smad7 (ρ = –0·32, p = 0·084) and Vegfa(ρ = –0·35, p = 0·055). These relationships were characterized by a conserved pattern in which low miRNA expression levels aligned with elevated expression of their putative targets (Figure 6C & 6D), suggesting a functional post-transcriptional regulatory mechanism. Such conserved inverse correlations reinforce the hypothesis that miR-10b-5p, miR-218-5p, miR-183-5p, and miR-16-5p actively participate in modulating corticosteroid-related pathways. These findings provide mechanistic support for the functional relevance of the selected miRNA panel and its potential utility as a regulatory signature in corticosteroid responsiveness. Discussion In this study, by integrating protein biomarkers with miR biomarkers, we provide a comprehensive molecular profile of corticosteroid response in UC human and mouse model contexts, offering potential predictive and mechanistic insights. The identified protein biomarkers play critical roles in GCR signalling and transcriptional regulation. NCOR1 and NRIP1 function as transcriptional corepressors, modulating GCR activity and inflammatory responses (20,21), potentially contributing to corticosteroid resistance. In contrast, NCOA3 and CBP act as co-activators, enhancing GCR-mediated transcriptional activity and influencing inflammatory and apoptotic pathways (22). These findings highlight the complex interplay between co-repressors and co-activators according to corticosteroid sensitivity in patients with UC. This mechanistic positioning reinforces the value of these molecules not merely as biomarkers but as active effectors within the corticosteroid response cascade. This study provides insight into the molecular status of UC patients before and after starting corticosteroid treatment, elucidating key molecules involved in corticosteroid response. Some of the proteins identified at baseline, such as GCR, IL32, and ATG5, reflect a pre-existing molecular state and glucocorticoid sensitivity. GCR is the primary receptor mediating corticosteroid effects, representing baseline glucocorticoid sensitivity (22). IL32, which is linked to chronic inflammation (23), may indicate an elevated inflammatory burden predictive of treatment resistance. ATG5, a regulator of autophagy (24), suggests a potential role in cellular stress responses influencing treatment outcomes. These proteins offer the potential for the early stratification of patients into likely R and NR before corticosteroid administration, enabling personalized treatment strategies. Proteins uniquely regulated at day three, such as IKKB and NCOA2, illustrate dynamic molecular changes induced by corticosteroid exposure. IKKB, an activator of the NF-κB pathway (25), suggests ongoing inflammatory signalling despite corticosteroid action. NCOA2, a steroid hormone receptor coactivator (22), indicates adaptive transcriptional responses to glucocorticoid therapy. Additionally, proteins consistently present at both time points (basal and day three), including CBP, NFKB1, and CREB1, emphasize persistent transcriptional regulation and inflammatory control mechanisms that may be critical to corticosteroid efficacy. These biomarkers also intersect with key signalling pathways such as NF-κB and p53, which are crucial for inflammatory regulation and apoptosis (26). The interplay between these pathways underscores the complexity of the mechanisms of response to corticosteroids. A previous study by our research group analysing the molecular mechanisms underlying steroid failure in UC identified NF-κB as a key player, in which responders showed increased GCR sensitivity and downregulation of its downstream effectors, particularly ETS1, RELA, and VEGFA, while CASP8 and P53 were markedly upregulated in R compared to NR (7). Here, NCOA3, CBP, NCOR1, and NRIP1 were identified as the best potential biomarkers due to their combined influence on GCR transcriptional activity. NCOA3 and CBP function as co-activators, whereas NCOR1 and NRIP1 act as co-repressors (20–22). Notably, CBP also modulates NF-κB activity via TF65 (27,28) and ETS1 (29) and promotes pro-angiogenic VEGFA and VGFR1 expression (30). This highlights the dual regulatory role of CBP, balancing both anti-inflammatory and pro-inflammatory pathways, which may be critical in determining corticosteroid efficacy. In addition, CBP also affects P53 transcriptional activity, however, depending on the cellular location, the effects are opposed: in the cytoplasm, CBP causes the degradation of P53, while it activates it in the nucleus (31). It is important to note that the co-activators/co-repressors, in addition to affecting transcription factors (TFs), are also affected by the interaction with these TFs, at least impeding their association with other TFs (as could be happening in the interactions CBP-GCR and CBP-other TFs). Hence, they should be taken into account, in particular in the case of CBP, which regulates several TFs that are involved in the MoA. The identification of miR biomarkers in both mucosal rectal biopsies and plasma samples provides valuable molecular signatures for distinguishing corticosteroid R from NR. However, the clinical utility of these biomarkers depends on understanding their systemic versus local expression patterns. By examining correlations between miRs in plasma and mucosal rectal biopsies, as well as by linking plasma miRs to key proteins identified from biopsy data, insights can be gained into how intestinal tissue pathology is reflected in circulating biomarkers. This integrative approach enhances the potential for non-invasive diagnostic tools while uncovering mechanistic relationships critical to corticosteroid resistance. Importantly, two miRs show a significantly differential expression between R and NR at baseline: miR-224-5p, which has been found to be related to the process of interest, and miR-487a-3p. Whereas the latter appears to increase in both mucosal and plasmatic samples in the NR cohort, the former appears to be higher in the rectal mucosa of NR and lower in the blood of the same cohort. Reports suggest that some cancer cells can extrude miRs via exosomes to maintain their oncogenesis (32), thus leading to increased levels of the miR in serum (33) while reducing it in the tissue. In addition, miR-224 has been proven to be upregulated in colorectal cancer tissue and has been related to its development through promoting proliferation (34). The mechanistic links of these miRs with key transcriptional pathways support their candidacy as functional predictors rather than surrogate markers. The identification of miR biomarkers, particularly miR-214-3p and miR-10b-5p , strengthens the robustness of our findings. These miRs were found to correlate strongly with rectal biopsy-derived data, reinforcing their potential as non-invasive diagnostic markers . Additionally, both miRs modulate P53 signalling and inflammatory mediators , providing a mechanistic justification for their role in corticosteroid response. The correlation between plasma and rectal biopsy miRs highlights the systemic nature of corticosteroid response, suggesting that plasma-derived miRs could serve as surrogate markers for intestinal inflammation . Notably, miR-224-5p showed opposite expression patterns in plasma and biopsy samples, a phenomenon observed in cancer studies where exosomal miR export affects tissue miR levels (35). Interestingly, our findings highlight VEGF as a key protein identified in the biopsy analysis. Focusing on this protein, which is known to be overexpressed in UC and to play a role in its development (36), it was further examined in relation to miR regulation. At baseline, we observed that, in R patients, plasma hsa-miR-145-5p changes, targeting VEGF, potentially contributing to the modulation of its expression. Notably, this regulatory interaction was no longer evident by day three following corticosteroid treatment, suggesting a possible mechanism through which R patients may achieve clinical improvement. The loss of this interaction post-treatment could reflect a normalisation of VEGF expression levels or a shift in regulatory dynamics as inflammation subsides. The TNBS colitis model recapitulated key aspects of human corticosteroid resistance, validating miR-145a-5p as a conserved biomarker of poor therapeutic response. This miR downregulation in both mR and mNR mice underscores its role in sustaining inflammation, potentially through Creb1 -mediated pathways. The selective suppression of miR-10b-5p and miR-16-5p in mR aligns with their putative roles in enhancing Gcr sensitivity and Vegfa signalling, respectively (see section 5 of results). However, the absence of miR-224-5p differential expression in mice contrasts with human findings, possibly due to anatomical differences or compensatory mechanisms in acute murine colitis. Similarly, the lack of Vegfa changes highlights the limitations in translating angiogenic pathways across species, urging caution in extrapolating mechanistic insights. The negative correlation between miR-10b-5p and miR-218-5p with Creb1 suggests a potential post-transcriptional regulatory mechanism in which these miRs may contribute to the fine-tuning of Creb1 expression in response to corticosteroid treatment. Creb1 plays a key role in inflammatory signalling and Gcr activation, making these miRs relevant candidates for modulating corticosteroid sensitivity (27). The significant correlation of miR-218-5p (p=0.037) reinforces this idea, suggesting that its downregulation could be associated with increased Creb1 expression, potentially influencing treatment resistance. Similarly, the negative correlation of miR-183-5p with GCR highlights its potential role in glucocorticoid signalling regulation. Given that Gcr is the primary mediator of corticosteroid action, an increase in miR-183-5p levels could contribute to a reduced Gcr expression, impairing corticosteroid response. This aligns with previous findings where GCR dysfunction has been linked to steroid resistance in UC patients (37). Moreover, miR-16-5p's negative correlation with Smad7 and Vegfa further supports its involvement in Tgf-β and angiogenesis pathways. Smad7 is a well-known inhibitor of Tgf-β signalling (38). Its negative correlation with miR-16-5p suggests a potential mechanism in which higher miR-16-5p levels could suppress Smad7, thereby promoting TGF-β activity and tissue repair. Likewise, Vegfa , a key angiogenic factor (39), also showed a negative correlation with miR-16-5p, indicating a possible role in vascular remodelling and mucosal healing. Despite the suggesting evidence provided by this study, some limitations should be noted. Although the plasma miRs identified in this study showed a high discriminatory power in the discovery cohort, we were unable to replicate these findings in an independent validation set due to the presence of hemolysis in most of those samples. However, it is worth noting that none of the samples included in this study showed clear signs of hemolysis. This underscores a technical limitation in circulating miR analysis which may confound quantification and obscure biological signals. Future validation studies should ensure rigorous preanalytical controls to avoid such confounders. Notably, mucosal-derived signals, and particularly biopsy miRs, appeared more robust and were not subject to such limitations, strengthening their value as mechanistically and technically reliable predictors. Importantly, several of the circulating miRs identified, such as miR-145-5p and miR-10b-5p, were found to be mechanistically linked to rectal tissue miRs, shared common targets, and showed consistent behaviours in the TNBS-induced colitis murine model. This cross-validation across compartments and species reinforces their biological plausibility, suggesting that these circulating miRNAs could be part of a conserved corticosteroid-response network in ulcerative colitis. Therefore, plasma-derived miRs from the discovery cohort are justified within the framework of an integrative systems biology approach, where convergence of evidence, rather than replication alone, builds mechanistic confidence. Future studies in non-hemolyzed plasma samples are necessary to assess the robustness and translational potential of these candidate biomarkers. Overall, the integration of multi-layered data anchored in the MoA of corticosteroids highlights a subset of molecular signals that are mechanistically grounded, thereby offering a robust platform for future biomarker validation and precision medicine strategies in UC. Declarations Conflict of Interest The authors have no conflicts of interest to declare. Data availability Gene profiling and miR sequencing data from the biopsy samples were deposited in NCBI’s GEO and are accessible through the GEO Series accession number GSE114527 and GSE114591. Data from the miR sequencing from plasma samples were also deposited in NCBI’s GEO and are accessible through the GEO Series accession number GSE122618. The other datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request. Funding This study was supported by public funding from the Instituto de Salud Carlos III (ISCIII, Ministry of Health, Government of Spain) through the following projects: PI22/01498, PI18/00892, and PI16/01937, all co-funded by the European Union (European Regional Development Fund – ERDF, “A way to make Europe”). Additional support was provided by ACCIÓ – Generalitat de Catalunya (project VALUNI16-1-0001). The authors also acknowledge economic and institutional support from CIBER (ISCIII), IISPV and the IGTP. Acknowledgements We acknowledge the patients and the IGTP BioBank for its collaboration in sample collection and processing. We thank the VHIO Cancer Genomics Group for their excellent support with NGS services of small RNA sequencing of plasma samples. We would also like to acknowledge the participation of Mariona Llaves and Eva Jou on the miR and mRNA analysis of mice samples and Ana Garcia for her advice on animal procedures. We thank Mireia Coma from Anaxomics Biotech S.L. for the implementation of the mathematical model in our analysis and the relation between the plasma and biopsy miR. Finally, we also wish to thank Eduard Cabré for his participation in the study as clinician through patient recruitment and sample gathering. Author contributions I.G. and R.S. conducted the main experiments. J.E.N. was responsible for patient recruitment, inclusion, and sample collection. V.L., I.G., and R.S. prepared samples for sequencing. A.M.A. performed the biopsy microarray and sequencing, along with the corresponding differential expression analyses. R.P., L.S., M.L.B., G.R., and J.F.SH. carried out plasma sequencing and its subsequent differential expression analysis. C.S. implemented the mathematical model and integrated plasma miRNA data with biopsy findings. I.G., R.S., L.C., and C.B. conducted the animal experiments, including sample collection and subsequent in silico and qRT-PCR analyses. J.M., C.Ser., L.S., E.D., and M.M. conceptualized and secured public competitive funding for the project. E.D., M.M., and L.S. provided critical revision of the manuscript. I.G., R.S., and J.M. interpreted the data and wrote the first draft of the manuscript, with R.S. contributing significantly to all aspects of the work. J.M. coordinated the overall project. J.M. and C.Ser. are guarantors of this work. References Kaplan GG, Ng SC. Understanding and Preventing the Global Increase of Inflammatory Bowel Disease. Gastroenterology. 2017;152(2). Llaó J, Naves JE, Ruiz-Cerulla A, Marín L, Mañosa M, Rodríguez-Alonso L, et al. Intravenous corticosteroids in moderately active ulcerative colitis refractory to oral corticosteroids. J Crohns Colitis. 2014;8(11). Salice M, Rizzello F, Calabrese C, Calandrini L, Gionchetti P. A current overview of corticosteroid use in active ulcerative colitis. Vol. 13, Expert Review of Gastroenterology and Hepatology. 2019. Ramamoorthy S, Cidlowski JA. Ligand-induced repression of the glucocorticoid receptor gene is mediated by an NCoR1 repression complex formed by long-range chromatin interactions with intragenic glucocorticoid response elements. Mol Cell Biol [Internet]. 2013 May 1 [cited 2025 Mar 10];33(9):1711–22. Available from: https://pubmed.ncbi.nlm.nih.gov/23428870/ Van Moortel L, Verhee A, Thommis J, Houtman R, Melchers D, Delhaye L, et al. Selective Modulation of the Human Glucocorticoid Receptor Compromises GR Chromatin Occupancy and Recruitment of p300/CBP and the Mediator Complex. Mol Cell Proteomics [Internet]. 2024 Mar 1 [cited 2025 Mar 10];23(3). Available from: https://pubmed.ncbi.nlm.nih.gov/38387774/ MacLellan SA, Lawson J, Baik J, Guillaud M, Poh CFY, Garnis C. Differential expression of miRNAs in the serum of patients with high-risk oral lesions. Cancer Med [Internet]. 2012 [cited 2025 Mar 10];1(2):268–74. Available from: https://pubmed.ncbi.nlm.nih.gov/23342275/ Lorén V, Garcia-Jaraquemada A, Naves JE, Carmona X, Mañosa M, Aransay AM, et al. ANP32E, a Protein Involved in Steroid-Refractoriness in Ulcerative Colitis, Identified by a Systems Biology Approach. J Crohns Colitis [Internet]. 2019 Mar 26 [cited 2025 Jan 5];13(3):351–61. Available from: https://pubmed.ncbi.nlm.nih.gov/30329026/ Montreal Classification for Inflammatory Bowel Disease (IBD) [Internet]. [cited 2025 May 14]. Available from: https://www.mdcalc.com/calc/10237/montreal-classification-inflammatory-bowel-disease-ibd Shi W, Oshlack A, Smyth GK. Optimizing the noise versus bias trade-off for Illumina whole genome expression BeadChips. Nucleic Acids Res [Internet]. 2010 Dec [cited 2025 May 14];38(22). Available from: https://pubmed.ncbi.nlm.nih.gov/20929874/ Smyth GK, Michaud J, Scott HS. Use of within-array replicate spots for assessing differential expression in microarray experiments. Bioinformatics [Internet]. 2005 May 1 [cited 2025 May 14];21(9):2067–75. Available from: https://pubmed.ncbi.nlm.nih.gov/15657102/ Lorén V, Manyé J, Fuentes MC, Cabré E, Ojanguren I, Espadaler J. Comparative Effect of the I3.1 Probiotic Formula in Two Animal Models of Colitis. Probiotics Antimicrob Proteins [Internet]. 2017 Mar 1 [cited 2025 May 14];9(1):71–80. Available from: https://pubmed.ncbi.nlm.nih.gov/27832441/ Robinson MD, Oshlack A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol [Internet]. 2010 Mar 2 [cited 2025 May 14];11(3). Available from: https://pubmed.ncbi.nlm.nih.gov/20196867/ Ritchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res [Internet]. 2015 Jan 6 [cited 2025 May 14];43(7):e47. Available from: https://pubmed.ncbi.nlm.nih.gov/25605792/ Hsu S Da, Tseng YT, Shrestha S, Lin YL, Khaleel A, Chou CH, et al. MiRTarBase update 2014: An information resource for experimentally validated miRNA-target interactions. Nucleic Acids Res [Internet]. 2014 Jan 1 [cited 2025 May 14];42(D1). Available from: https://pubmed.ncbi.nlm.nih.gov/24304892/ Wang X, El Naqa IM. Prediction of both conserved and nonconserved microRNA targets in animals. Bioinformatics [Internet]. 2008 [cited 2025 May 14];24(3):325–32. Available from: https://pubmed.ncbi.nlm.nih.gov/18048393/ Wang X. miRDB: A microRNA target prediction and functional annotation database with a wiki interface. RNA [Internet]. 2008 Jun [cited 2025 May 14];14(6):1012–7. Available from: https://pubmed.ncbi.nlm.nih.gov/18426918/ Suau R, Garcia A, Bernal C, Llaves M, Schiering K, Jou-Ollé E, et al. Response Variability to Drug Testing in Two Models of Chemically Induced Colitis. Int J Mol Sci [Internet]. 2023 Apr 1 [cited 2025 May 14];24(7). Available from: https://pubmed.ncbi.nlm.nih.gov/37047397/ Livak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2-ΔΔCT method. Methods [Internet]. 2001 [cited 2025 May 23];25(4):402–8. Available from: https://pubmed.ncbi.nlm.nih.gov/11846609/ Martín-Segura A, Casadomé-Perales Á, Fazzari P, Mas JM, Artigas L, Valls R, et al. Aging increases hippocampal DUSP2 by a membrane cholesterol loss-mediated RTK/p38MAPK activation mechanism. Front Neurol [Internet]. 2019 Jun 25 [cited 2025 Jul 2];10(JUN):465444. Available from: www.frontiersin.org Ramamoorthy S, Cidlowski JA. Ligand-induced repression of the glucocorticoid receptor gene is mediated by an NCoR1 repression complex formed by long-range chromatin interactions with intragenic glucocorticoid response elements. Mol Cell Biol [Internet]. 2013 May 1 [cited 2025 Jan 6];33(9):1711–22. Available from: https://pubmed.ncbi.nlm.nih.gov/23428870/ Monczor F, Chatzopoulou A, Zappia CD, Houtman R, Meijer OC, Fitzsimons CP. A model of glucocorticoid receptor interaction with coregulators predicts transcriptional regulation of target genes. Front Pharmacol. 2019;10(MAR). Van Moortel L, Verhee A, Thommis J, Houtman R, Melchers D, Delhaye L, et al. Selective Modulation of the Human Glucocorticoid Receptor Compromises GR Chromatin Occupancy and Recruitment of p300/ CBP and the Mediator Complex. Molecular and Cellular Proteomics. 2024;23(3). Xin T, Chen M, Duan L, Xu Y, Gao P. Interleukin-32: Its role in asthma and potential as a therapeutic agent. Vol. 19, Respiratory Research. 2018. Changotra H, Kaur S, Yadav SS, Gupta GL, Parkash J, Duseja A. ATG5: A central autophagy regulator implicated in various human diseases. Vol. 40, Cell Biochemistry and Function. 2022. Freitas RHCN, Fraga CAM. NF-κB-IKKβ Pathway as a Target for Drug Development: Realities, Challenges and Perspectives. Curr Drug Targets. 2018;19(16). Qian K, Yuan L, Wang S, Kuang Y, Jin Q, Long D, et al. Inhibitor of apoptosis-stimulating p53 protein protects against inflammatory bowel disease in mice models by inhibiting the nuclear factor kappa B signaling. Clin Exp Immunol. 2021;205(2). McKay LI, Cidlowski JA. CBP (CREB Binding Protein) Integrates NF-κB (Nuclear Factor-κB) and Glucocorticoid Receptor Physical Interactions and Antagonism. Molecular Endocrinology. 2000;14(8). Peng Q, Hua Y, Xu H, Chen X, Xu H, Wang L, et al. The NCOA1-CBP-NF-κB transcriptional complex induces inflammation response and triggers endotoxin-induced myocardial dysfunction. Exp Cell Res. 2022;415(2). Ang DA, Carter JM, Deka K, Tan JHL, Zhou J, Chen Q, et al. Aberrant non-canonical NF-κB signalling reprograms the epigenome landscape to drive oncogenic transcriptomes in multiple myeloma. Nature Communications 2024 15:1 [Internet]. 2024 Mar 21 [cited 2025 Feb 24];15(1):1–20. Available from: https://www.nature.com/articles/s41467-024-46728-4 Li L, Song Q, Zhou J, Ji Q. Controllers of histone methylation-modifying enzymes in gastrointestinal cancers. Vol. 174, Biomedicine and Pharmacotherapy. 2024. Shi D, Pop MS, Kulikov R, Love IM, Kung A, Grossman SR. CBP and p300 are cytoplasmic E4 polyubiquitin ligases for p53. Proc Natl Acad Sci U S A [Internet]. 2009 Sep 22 [cited 2025 Jan 5];106(38):16275–80. Available from: https://pubmed.ncbi.nlm.nih.gov/19805293/ Ohshima K, Inoue K, Fujiwara A, Hatakeyama K, Kanto K, Watanabe Y, et al. Let-7 microRNA family is selectively secreted into the extracellular environment via exosomes in a metastatic gastric cancer cell line. PLoS One [Internet]. 2010 [cited 2025 Jan 5];5(10). Available from: https://pubmed.ncbi.nlm.nih.gov/20949044/ MacLellan SA, Lawson J, Baik J, Guillaud M, Poh CFY, Garnis C. Differential expression of miRNAs in the serum of patients with high-risk oral lesions. Cancer Med [Internet]. 2012 [cited 2025 Jan 5];1(2):268–74. Available from: https://pubmed.ncbi.nlm.nih.gov/23342275/ Zhang X, Zhang X, Liu C, Jia N, Li X, Xiao J. MiR‑224 promotes colorectal cancer cells proliferation via downregulation of P21WAF1/CIP1. Mol Med Rep [Internet]. 2014 Mar [cited 2025 Jan 5];9(3):941–6. Available from: https://pubmed.ncbi.nlm.nih.gov/24430932/ Ghosh S, Bose M, Ray A, Bhattacharyya SN. Polysome arrest restricts miRNA turnover by preventing exosomal export of miRNA in growth-retarded mammalian cells. Mol Biol Cell. 2015;26(6). Zdravkovic ND, Jovanovic IP, Radosavljevic GD, Arsenijevic AN, Zdravkovic ND, Mitrovic SL, et al. Potential dual immunomodulatory role of VEGF in ulcerative colitis and colorectal carcinoma. Int J Med Sci [Internet]. 2014 Jul 2 [cited 2025 May 26];11(9):936–47. Available from: https://pubmed.ncbi.nlm.nih.gov/25076849/ Huang H, Wang W. Molecular mechanisms of glucocorticoid resistance. Eur J Clin Invest [Internet]. 2023 Feb 1 [cited 2025 Mar 3];53(2). Available from: https://pubmed.ncbi.nlm.nih.gov/36346177/ Hu Y, He J, He L, Xu B, Wang Q. Expression and function of Smad7 in autoimmune and inflammatory diseases. J Mol Med [Internet]. 2021 Sep 1 [cited 2025 Mar 3];99(9):1209–20. Available from: https://link.springer.com/article/10.1007/s00109-021-02083-1 Ahmad A, Nawaz MI. Molecular mechanism of VEGF and its role in pathological angiogenesis. J Cell Biochem [Internet]. 2022 Dec 1 [cited 2025 Mar 3];123(12):1938–65. Available from: https://onlinelibrary.wiley.com/doi/full/10.1002/jcb.30344 Kenji Kira LAR. The Feature Selection Problem: Traditional Methods and a New Algorithm. [cited 2025 May 21]; Available from: https://aaai.org/papers/00129-aaai92-020-the-feature-selection-problem-traditional-methods-and-a-new-algorithm/ Pattern Recognition - Sergios Theodoridis, Konstantinos Koutroumbas - Google Llibres [Internet]. [cited 2025 May 21]. Available from: https://books.google.es/books?hl=ca&lr=&id=gAGRCmp8Sp8C&oi=fnd&pg=PP1&dq=Theodoridis,+S.+and+Koutroumbas,+K.+(1999)+Pattern+Recognition,+Academic+Press,+pp.+341%E2%80%93342&ots=pKR6YFPdZM&sig=sBwDek5UZ9R-CtCos9ZPs-3hExI&redir_esc=y#v=onepage&q&f=false Fay MP, Proschan MA. Wilcoxon-Mann-Whitney or t-test? On assumptions for hypothesis tests and multiple interpretations of decision rules. Stat Surv [Internet]. 2010 [cited 2025 May 21];4:1. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC2857732/ Hanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology [Internet]. 1983 Sep 1 [cited 2025 May 21];148(3):839–43. Available from: /doi/pdf/10.1148/radiology.148.3.6878708 Tables Table 1. List of different approaches used to identify potential sets of biomarkers from miR sequencing data. Approaches Reference RELIEF feature selection method Kira, Kenji and Rendell, Larry (1992). The Feature Selection Problem: Traditional Methods and a New Algorithm . AAAI-92 Proceedings (40) ENTROPY+CORRELATION feature selection method (a lso known as Kullback-Liebler distance or divergence) Theodoridis, S. and Koutroumbas, K. (1999) Pattern Recognition , Academic Press, pp. 341–342 (41) WILCOXON feature selection method Fay, Michael P.; Proschan, Michael A. (2010). "Wilcoxon–Mann–Whitney or t-test? On assumptions for hypothesis tests and multiple interpretations of decision rules" . Statistics Surveys 4: 1–39 (42) SIMPLE REGRESSION feature selection method Anaxomics method. Stands for a feed forward selection method. Initially both feature combinations are evaluated using a logistic regression. Then, the feature sets are grown (with logistic regression), checking all the combinations obtained when one feature is added to the pool of previously identified best combinations. ROC feature selection method Hanley, James A.; McNeil, Barbara J. (1983). A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology 148 (3): 839–843. (43) ROC; receiver operating characteristic. Table 2. Summary of the best potential protein-coding transcript biomarker combination for each time point. Upregulated and downregulated proteins in R (vs NR) are represented by up green and down orange arrows, respectively. Day UniProt Unique identifier name UniProt code Generalization capability Accuracy Baseline NCOR1 Q75376 82.35% 100% NCOA3 Q9Y6Q9 CBP Q92793 Day 3 NRIP1 P48552 80.95% 100% NCOR1 ⬇ Q75376 CBP Q92793 NR, non-responders; R, responders. Table 3. Proteins identified in basal and/or day three analysis related to both UC and CS. Upregulated and downregulated proteins in R (vs NR) are represented by up green and down orange arrows, respectively. Protein name Uniprot code Baseline Day 3 Shared or unique Functional role References CBP Q92793 ✔ ✔ Shared Coactivator of GCR and NF-κB; modulates transcription and angiogenesis PMID: 23154639, PMID: 23160045 CEBPB P17676 ✔ ⬆ ✔ Shared Regulates inflammatory gene expression PMID: 19958093, PMID: 25203139 HDAC1 Q13547 ✔ ⬆ ✔ Shared Chromatin remodeling, transcription regulation PMID: 22517765, PMID: 25203139 CREB1 P16220 ✔ ✔ Shared Stress and inflammatory response regulation PMID: 23154639, PMID: 18762182 NFKB1 P19838 ✔ ✔ Shared Central transcription factor regulating immune and inflammatory pathways PMID: 23322997, PMID: 25133403, PMID: 20407229 ETS1 P14921 ✔ ✔ Shared Transcription factor influencing cell proliferation and differentiation PMID: 25926267, PMID: 12145332 IKKB O14920 ✘ ✔ Unique to Day 3 Activator of NF-κB signaling; regulates inflammation PMID: 15753535, PMID: 24347992 CALR P27797 ✘ ✔ ⬆ Unique to Day 3 Calcium-binding chaperone involved in protein folding PMID: 16995349, PMID: 9201693 FLNA P21333 ✘ ✔ Unique to Day 3 Cytoskeletal protein regulating cell structure and signaling PMID: 22198550, PMID: 16639003 E2F4 Q16254 ✘ ✔ Unique to Day 3 Transcription factor controlling cell cycle progression PMID: 21247883, PMID: 10867026 NCOA2 Q15596 ✘ ✔ Unique to Day 3 Coactivator of GCR; enhances transcriptional activity PMID: 19198856, PMID: 15207724 PGFRB P09619 ✘ ✔ Unique to Day 3 Cell proliferation and angiogenesis PMID: 21047522, PMID: 24228109 BCR P11274 ✘ ✔ Unique to Day 3 Signaling in cell growth and immune response PMID: 23825635, PMID: 23264597 NDUS1 P28331 ✘ ✔ Unique to Day 3 Role in mitochondrial respiratory chain PMID: 20440543, PMID: 23670350 LIF P15018 ✘ ✔ Unique to Day 3 Cytokine involved in inflammation PMID: 9889404, PMID: 15985451 AT2A2 P16615 ✔ ✘ Unique to Basal Calcium signaling regulation PMID: 17131044, PMID: 19726057 KLF5 Q13887 ✔ ✘ Unique to Basal Regulates cell proliferation and inflammation PMID: 17262812, PMID: 14634001 IL32 P24001 ✔ ⬇ ✘ ⬇ Unique to Basal Cytokine involved in inflammation PMID: 17590175, PMID: 24791863, PMID: 22646473 GCR P04150 ✔ ✘ Unique to Basal Glucocorticoid receptor regulating CS action PMID: 19646928, PMID: 24084075 AP2A1 O95782 ✔ ✘ Unique to Basal Vesicular transport regulation PMID: 20376207, PMID: 9421462 STA5A P42229 ✔ ✘ Unique to Basal Signal transduction and transcription PMID: 22019623, PMID: 16804404 PO2F1 P14859 ✔ ✘ Unique to Basal DNA binding and transcriptional regulation PMID: 22479607, PMID: 12807698, PMID: 9722596 CHIP Q9UNE7 ✔ ✘ Unique to Basal Protein quality control, ubiquitination PMID: 25258038, PMID: 20661446 ATG5 Q9H1Y0 ✔ ✘ Unique to Basal Autophagy regulation PMID: 25642769, PMID: 24897381 ANGP2 O15123 ✔ ✘ Unique to Basal Angiogenesis PMID: 25759532, PMID: 18037159 CS, corticosteroid; GCR, glucocorticoid receptor; NR, non-responders; R, responders. Table 4. Summary of the best potential miRs biomarker combination per visit and sample. Day/ Sample All miRs miRs related to MoA Baseline/ Plasma Combination 1 Feature Selection Method: ENTROPY + CORRELATION Generalization capability: 78.95% Accuracy: 100% Selected features: hsa-miR-218-5p hsa-miR-6754-5p hsa-miR-4767 hsa-miR-939-3p hsa-miR-548s hsa-miR-34c-5p Combination 2 Feature Selection Method: SIMPLE REGRESSION Generalization capability: 73.68% Accuracy: 100% Selected features: hsa-miR-224-5p hsa-miR-324-5p hsa-miR-10a-5p hsa-let-7d-5p Day 3/ Plasma Combination 1 Feature Selection Method: SIMPLE REGRESSION Generalization capability: 70.00% Accuracy: 100% Selected features: hsa-miR-181A-5p hsa-miR-548at-5p Combination 2 Feature Selection Method: SIMPLE REGRESSION Generalization capability: 70.00% Accuracy: 100% Selected features: hsa-miR-181a-5p hsa-miR-433-3p hsa-miR-20a-5p Baseline/ Biopsy Combination 1 Feature Selection Method: ROC Generalization capability: 93.33% Accuracy: 100% Selected features: hsa-miR-769-5p hsa-miR-6516-3p hsa-miR-625-5p Combination 2 Feature Selection Method: ROC Generalization capability: 93.33% Accuracy: 100% Selected features: hsa-miR-10B-5p hsa-miR-449a hsa-miR-494-3p hsa-miR-224-5p hsa-miR-339-5p Day 3/ Biopsy Combination 2 Feature Selection Method: RELIEF Generalization capability: 100% Accuracy: 100% Selected features: hsa-miR-140-5p hsa-miR-431-3p hsa-miR-25-3p hsa-miR-151a-3p hsa-miR-7706 hsa-miR-299-5p hsa-miR-362-5p Combination 2 Feature Selection Method: RELIEF Generalization capability: 100% Accuracy: 100% Selected features: hsa-miR-140-5p hsa-miR-25-3p hsa-miR-130b-5p miR, microRNA; MoA, mechanism of action; ROC, receiver operating characteristic. Additional Declarations No competing interests reported. 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-7401947","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":514670884,"identity":"545ce281-8917-4337-b183-53e89937c8f8","order_by":0,"name":"Iris Ginés","email":"","orcid":"","institution":"Hospital Universitari de Tarragona Joan XXIII. Institut d´Investigació Sanitària Pere Virgili, Universitat Rovira i Virgili","correspondingAuthor":false,"prefix":"","firstName":"Iris","middleName":"","lastName":"Ginés","suffix":""},{"id":514670885,"identity":"bb78a1f5-93bc-4b4c-b5e4-0ce9ecbc9cec","order_by":1,"name":"Roger Suau","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAArElEQVRIiWNgGAWjYNCCCgmStZyRYOABMxKI1cHYxkCCFvP+M4afeedZ2O2XbmD88PMHEVpkbuQYS/Nuk0jukTnALNlDjC0SEjwGYC08EglsDDxEaeE/Y/ybdw5EC+MforQw5JhJ8zZI2IG0MBNni0RameWcYxIJPHcONkvLpBHlsMObb7ypqbNnn9188OMbGyK0wEBigwRjAwnqgcCegfQ0MwpGwSgYBSMFAAC6hSstYoX1XwAAAABJRU5ErkJggg==","orcid":"","institution":"Germans Trias i Pujol Research Institute (IGTP)","correspondingAuthor":true,"prefix":"","firstName":"Roger","middleName":"","lastName":"Suau","suffix":""},{"id":514670890,"identity":"b0e36fdc-9d93-4626-8f0c-3a124552d117","order_by":2,"name":"Juan Enrique Naves","email":"","orcid":"","institution":"Hospital del Mar","correspondingAuthor":false,"prefix":"","firstName":"Juan","middleName":"Enrique","lastName":"Naves","suffix":""},{"id":514670892,"identity":"71f66f1c-1338-4771-a3a0-61f6664ba287","order_by":3,"name":"Carla Bernal","email":"","orcid":"","institution":"Germans Trias i Pujol Research Institute (IGTP)","correspondingAuthor":false,"prefix":"","firstName":"Carla","middleName":"","lastName":"Bernal","suffix":""},{"id":514670894,"identity":"9b7b5bac-b928-4f0b-b680-7d5219d3ad14","order_by":4,"name":"Laura Clua","email":"","orcid":"","institution":"Germans Trias i Pujol Research Institute (IGTP)","correspondingAuthor":false,"prefix":"","firstName":"Laura","middleName":"","lastName":"Clua","suffix":""},{"id":514670897,"identity":"f26ac62c-4979-499c-a309-d442cb973018","order_by":5,"name":"Violeta Lorén","email":"","orcid":"","institution":"Germans Trias i Pujol Research Institute (IGTP)","correspondingAuthor":false,"prefix":"","firstName":"Violeta","middleName":"","lastName":"Lorén","suffix":""},{"id":514670898,"identity":"5cf2c2db-6cd2-4fa7-8b7c-2e38cc5a8be0","order_by":6,"name":"Raquel Pluvinet","email":"","orcid":"","institution":"IGTP","correspondingAuthor":false,"prefix":"","firstName":"Raquel","middleName":"","lastName":"Pluvinet","suffix":""},{"id":514670899,"identity":"8709787c-39b9-4f4f-bf87-10ab4506f3ac","order_by":7,"name":"Gabriel Rech","email":"","orcid":"","institution":"IGTP","correspondingAuthor":false,"prefix":"","firstName":"Gabriel","middleName":"","lastName":"Rech","suffix":""},{"id":514670900,"identity":"767e323d-dfe7-4d7c-a703-fd152aa37b0d","order_by":8,"name":"Marta López Balastegui","email":"","orcid":"","institution":"IGTP","correspondingAuthor":false,"prefix":"","firstName":"Marta","middleName":"López","lastName":"Balastegui","suffix":""},{"id":514670901,"identity":"353ccfc6-6385-43ce-b31b-060ef8e2930c","order_by":9,"name":"José Francisco Sánchez Herrero","email":"","orcid":"","institution":"IGTP","correspondingAuthor":false,"prefix":"","firstName":"José","middleName":"Francisco Sánchez","lastName":"Herrero","suffix":""},{"id":514670902,"identity":"22a35e8c-1238-4676-a6d6-61f094d49ecd","order_by":10,"name":"Cristina Segú","email":"","orcid":"","institution":"Anaxomics Biotech, S.L","correspondingAuthor":false,"prefix":"","firstName":"Cristina","middleName":"","lastName":"Segú","suffix":""},{"id":514670907,"identity":"5eb97dfd-a373-4ec2-bd31-adf85c99a3a2","order_by":11,"name":"Ana Maria Aransay","email":"","orcid":"","institution":"CIC bioGUNE","correspondingAuthor":false,"prefix":"","firstName":"Ana","middleName":"Maria","lastName":"Aransay","suffix":""},{"id":514670909,"identity":"ee4b944d-d3da-40ec-bc00-d2b4b84a0a22","order_by":12,"name":"Marcus Buschbeck","email":"","orcid":"","institution":"Program of Applied Epigenetics, Josep Carreras Leukaemia Research Institute (IJC)","correspondingAuthor":false,"prefix":"","firstName":"Marcus","middleName":"","lastName":"Buschbeck","suffix":""},{"id":514670910,"identity":"7a896725-a47a-498d-871a-c529efa90a8d","order_by":13,"name":"Míriam Mañosa","email":"","orcid":"","institution":"Germans Trias i Pujol Research Institute (IGTP)","correspondingAuthor":false,"prefix":"","firstName":"Míriam","middleName":"","lastName":"Mañosa","suffix":""},{"id":514670911,"identity":"64da83ba-cb6e-4644-b86d-5b3fdf7b638c","order_by":14,"name":"Lauro Sumoy","email":"","orcid":"","institution":"IGTP","correspondingAuthor":false,"prefix":"","firstName":"Lauro","middleName":"","lastName":"Sumoy","suffix":""},{"id":514670912,"identity":"6f671094-ad2f-4b3e-89b8-f1e535352afe","order_by":15,"name":"Carolina Serena","email":"","orcid":"","institution":"Hospital Universitari de Tarragona Joan XXIII. Institut d´Investigació Sanitària Pere Virgili, Universitat Rovira i Virgili","correspondingAuthor":false,"prefix":"","firstName":"Carolina","middleName":"","lastName":"Serena","suffix":""},{"id":514670913,"identity":"3e63172d-73f3-44e4-90c2-eec5f285f6cc","order_by":16,"name":"Eugeni Domènech","email":"","orcid":"","institution":"Germans Trias i Pujol Research Institute (IGTP)","correspondingAuthor":false,"prefix":"","firstName":"Eugeni","middleName":"","lastName":"Domènech","suffix":""},{"id":514670914,"identity":"c5e7791f-793f-4b33-8fa7-1c73d183bace","order_by":17,"name":"Josep Manyé","email":"","orcid":"","institution":"Germans Trias i Pujol Research Institute (IGTP)","correspondingAuthor":false,"prefix":"","firstName":"Josep","middleName":"","lastName":"Manyé","suffix":""}],"badges":[],"createdAt":"2025-08-18 17:23:11","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-7401947/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-7401947/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":91489019,"identity":"860a5a9a-f1dd-4b0a-b47a-c6c6f29e068f","added_by":"auto","created_at":"2025-09-17 05:11:39","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":7965143,"visible":true,"origin":"","legend":"\u003cp\u003eSchematic representation of the biopsy‐plasma assotiation. A non‐adjusted p‐value \u0026lt; 0·05 was used to define differential mRNA. miRs are considered differential following the same parameters as in restriction extraction (adjusted p‐value \u0026lt; 0·05, or classifier p‐value \u0026lt; 0·05, or non‐adjusted p‐value \u0026lt; 0·01). MoA, mechanism of action; NR, non-responder; R, responder.\u003c/p\u003e","description":"","filename":"Figure1.png","url":"https://assets-eu.researchsquare.com/files/rs-7401947/v1/d856a2d222b14d30e2001dc8.png"},{"id":91489018,"identity":"f838c034-6cbe-46f4-bd69-8a2524863c5a","added_by":"auto","created_at":"2025-09-17 05:11:39","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":2009992,"visible":true,"origin":"","legend":"\u003cp\u003eRepresentation of the mechanistic relationship between the potential miR biomarkers identified from data at basal visit and the proteins in the MoA. Plot was made using Graphviz 12·2·1. miR; microRNA.\u003c/p\u003e","description":"","filename":"Figure2.png","url":"https://assets-eu.researchsquare.com/files/rs-7401947/v1/a6f7b2f9ed8f6238e69092c3.png"},{"id":91489017,"identity":"84c5c1d9-79e8-4ae9-8f8a-5a3097652231","added_by":"auto","created_at":"2025-09-17 05:11:39","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":2357855,"visible":true,"origin":"","legend":"\u003cp\u003eRepresentation of the mechanistic relationship between the potential miR biomarkers identified from plasma at day three visit and the proteins in the MoA. Plot was made using Graphviz 12·2·1. miR; microRNA.\u003c/p\u003e","description":"","filename":"Figure3.png","url":"https://assets-eu.researchsquare.com/files/rs-7401947/v1/cae164254dcbbe9d2290b802.png"},{"id":91489021,"identity":"31d42f76-8512-4bb2-a502-6eea194ea1d0","added_by":"auto","created_at":"2025-09-17 05:11:40","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":32109669,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation between plasma miRs and A) biopsy miRs and miRs at the basal visit and B) the key proteins identified from the models created from biopsy data in the basal visit comparison. Note: “P” and “B” at the end of the miR names mean they are from plasma or biopsy data, respectively. miR, microRNA; MoA, mechanism of action; NR, non-responder; R, responder.\u003c/p\u003e","description":"","filename":"Figure4.png","url":"https://assets-eu.researchsquare.com/files/rs-7401947/v1/d3965d3332e37fdb8d958760.png"},{"id":91489020,"identity":"b4e89b0f-9417-4086-b11b-c60961c1137b","added_by":"auto","created_at":"2025-09-17 05:11:39","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":11911779,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelations between plasma miR and biopsy miRs, miRs and key proteins identified from models created from biopsy data in the day three visit comparison. Note: “P” and “B” at the end of the miR names mean they are from plasma or biopsy data, respectively. miR, microRNA; MoA, mechanism of action; NR, non-responder; R, responder.\u003c/p\u003e","description":"","filename":"Figure5.png","url":"https://assets-eu.researchsquare.com/files/rs-7401947/v1/86ccea4fe90cbb44d5bbe996.png"},{"id":91489023,"identity":"00ff002f-4272-4409-9991-4f239b6c7059","added_by":"auto","created_at":"2025-09-17 05:11:41","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":28362777,"visible":true,"origin":"","legend":"\u003cp\u003eValidation of miR biomarkers in the TNBS colitis murine model of response to dexamethasone. (A) Inflammation measured through BLI (y-axis) depending on the response to treatment by experimental day (x-axis); \u003csup\u003e#\u003c/sup\u003ep \u0026lt; 0·05 \u003cem\u003evs.\u003c/em\u003e baseline \u0026amp; **p \u0026lt; 0·05 \u003cem\u003evs\u003c/em\u003e. others. (B) Colon weight/length ratio (y-axis) by treatment condition (color) and day of the experiment (x-axis) (*p = 0·05, non-treated \u003cem\u003evs.\u003c/em\u003e responder). (C) miR expression expressed as FC and normalized by the housekeeping gene expression of miR-103-3p and miR-30e-5p (y-axis). The x-axis represents the experimental day and the color represents the treatment condition. Data is given as FC and normalized using the Gapdh and B2m housekeeping genes. The miR is depicted as boxplots. Box represents IQR, while the medium line corresponds to the median. Whiskers represent the maximum and minimum value. (D) mRNA expression of the targets of the aforementioned miRs. Boxplots are built as mentioned above. [*p £ 0·05, **p£ 0·01, ***p£ 0·005]. (E) Scatter plots with a correlation analysis of the miR (x-axis) and its mRNA targets’ expression (y-axis). Data is represented as points. R\u003csup\u003e2\u003c/sup\u003e is given as the statistic of the Spearman correlation and P-value as significance. A fitted line is given as a representation of the y = mx+n formula. BLI, bioluminescence intensity; FC, fold change; IQR, interquartile range; R, rho.\u003c/p\u003e","description":"","filename":"Figure6.png","url":"https://assets-eu.researchsquare.com/files/rs-7401947/v1/ede12dafa91ab5cbda0f5723.png"},{"id":94112888,"identity":"c262049e-bee8-4c6f-8431-af290bfd5315","added_by":"auto","created_at":"2025-10-22 13:47:36","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":69914264,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-7401947/v1/e5a971b4-4c61-42ce-acd6-46f805c09c19.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Systems biology-driven miR-mRNA integration identifies potential steroid- refractoriness biomarkers in ulcerative colitis and reveals a conserved mechanism across species","fulltext":[{"header":"Introduction","content":"\u003cp\u003eUlcerative colitis (UC) is a chronic inflammatory disease of the colon characterized by periods of relapse and remission, significantly impacting patients' quality of life (1). While corticosteroids remain a cornerstone in the management of UC flares, approximately 60% of patients exhibit steroid resistance or dependence, requiring alternative therapeutic strategies (2). Identifying reliable biomarkers to predict corticosteroid response is crucial for optimising patient management, minimising unnecessary steroid exposure, and preventing complications such as osteoporosis, infections, and metabolic disturbances (3).\u003c/p\u003e\n\u003cp\u003eGlucocorticoids exert their effects primarily through the glucocorticoid receptor (GCR), a ligand-activated transcription factor that regulates inflammation by modulating gene expression. However, alterations in GCR signalling, including receptor isoform imbalances and impaired nuclear translocation, have been associated with corticosteroid resistance in UC\u0026nbsp;(4). Additionally, transcriptional co-regulators such as NCOA3, CBP, and NCOR1 play a pivotal role in fine-tuning GCR activity, influencing inflammatory pathways and treatment response\u0026nbsp;(5). Beyond transcriptional regulation, non-coding RNAs, particularly microRNAs (miRs), have emerged as key post-transcriptional regulators of GCR function, NF-κB signalling, and cytokine expression, further shaping corticosteroid responsiveness\u0026nbsp;(6).\u003c/p\u003e\n\u003cp\u003eIn line with these findings, our research group (7)applied a systems biology approach to investigate the mechanisms underlying steroid refractoriness in patients with UC. By integrating transcriptomic data, the study identified 18 proteins associated with corticosteroid response, focusing on ANP32E. This chaperone protein, which is involved in histone exchange and GCR-induced transcription, was expressed differentially by responders (R) and non-responders (NR). These findings suggested that alterations in chromatin remodelling, NF-κB signalling, and angiogenesis contribute to steroid failure in UC. However, this study did not assess post-transcriptional regulatory mechanisms, such as the miR-mediated modulation of these pathways, nor did it validate findings using complementary \u003cem\u003ein vivo\u003c/em\u003e models.\u003c/p\u003e\n\u003cp\u003eBuilding upon these insights, our study employed a systems biology and machine learning approach to integrate mRNA and miR expression profiles from both rectal biopsies and, for the first time in our research, \u0026nbsp;from plasma samples of UC patients undergoing corticosteroid therapy. Unlike our previous integrative analyses, which were limited to mucosal tissue, this study incorporated circulating microRNAs as a novel layer of molecular information. We aimed to explore the plasma microRNAs profiles associated with corticosteroid refractoriness, integrating these with rectal and in vivo data to generate testable mechanistic hypotheses and guide future biomarker development. By analysing samples collected at baseline and after three days of corticosteroid treatment, we identified disparities in miR-mRNA interactions according to corticosteroid effect, as well as novel miR profiles with high predictive potential for treatment failure. Additionally, we described key molecular findings in a TNBS-induced colitis mouse model supporting the existence of cross-species conserved mechanisms. Collectively, this integrative strategy provides new insights into molecular steroid-refractoriness in UC, and may generate clinically useful tools which can help to guide subsequent clinical studies to improve the management of this chronic inflammatory bowel disease. Importantly, this approach aimed not only to identify predictive markers, but also to mechanistically contextualize them within the glucocorticoid response network, enhancing their translational relevance.\u003c/p\u003e"},{"header":"Materials and Methods","content":"\u003cp\u003e\u003cstrong\u003ePatients and study design\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe present study complements the previous identification of a steroid-refractoriness-linked molecular mechanism in UC (7) by integrating intestinal mRNA and miR profiles. In brief, rectal biopsies and peripheral blood were collected from 24 UC patients\u0026nbsp;prior to treatment (baseline), and\u0026nbsp;three\u0026nbsp;days\u0026nbsp;after\u0026nbsp;the\u0026nbsp;initiation of corticosteroid therapy\u0026mdash;either 0\u0026middot;75\u0026ndash;1 mg/kg/day of oral prednisone for moderate disease flares, or 1 mg/kg/day of intravenous prednisolone in cases of severe activity.\u0026nbsp;Samples from 10 healthy controls were also included for comparison.\u0026nbsp;Disease activity was assessed clinically at day\u0026nbsp;seven\u0026nbsp;of glucocorticoid treatment, classifying the patients as steroid-responsive (responders, R) if they had\u0026nbsp;mild activity or inactive disease according to the severity criteria of the Montreal classification, or steroid-refractory (non-responders, NR) if moderate or severe clinical activity was still present or if rescue therapy was needed)\u0026nbsp;(8).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eWhole-genome mRNA expression analysis was performed on rectal biopsies collected before and after treatment using oligonucleotide microarrays (HT-12 v4, Illumina, USA). In parallel, the expression of approximately 3,600 known miRs was profiled using next-generation sequencing (TruSeq small RNA, Illumina, USA). Technical details for both assays are as previously described (7). The raw data for both mRNA microarrays and miR sequencing were deposited in NCBI\u0026rsquo;s Gene Expression Omnibus (GEO), assigning to them the GEO Series accession number GSE114527 and GSE114591, respectively in compliance with MIAME standards (https://www.ncbi.nlm.nih.gov/geo/info/MIAME. html).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eMicroarray data were normalized with the Neqc method and differential expression was analysed with Linear Models for Microarray data (LIMMA) (9,10), considering them significant if False Discovery Rate (FDR) \u0026le;0\u0026middot;01. Gene information was one-to-one mapped to proteins for its introduction into mathematical models, including differential FCs and significant protein or cluster correlations between the control and intervention cohorts (11).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCirculating miR sequencing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe previous study described sequencing methods for miRs measurement in rectal biopsies (11). In the present study, small RNAs (\u0026lt;1000nt) were isolated from 300\u0026micro;l of plasma using a miRCURY RNA isolation kit (EXIQON, Cat. No.300112), following the manufacturer\u0026rsquo;s instructions. Plasma samples were collected and stored at -80⁰C until all samples had been collected, then the plasmas were thawed and used at the same time. MiR libraries (135-160bp) were prepared starting from total RNA extracted from 200 ul of plasma using the TruSeq small RNA library preparation kit (Illumina, USA) with 15 cycles. Individual libraries were quantified by DNA-1000 Bioanalyzer assay (Agilent). Equal mass library pool was size selected automatically using PippinPrep 3% agarose cassettes with PippinPrep software v.6\u0026middot;0. QIAquick column purifications were performed prior to and after size selection. The size selected pool was quantified by qPCR with Kappa Library Quantification Kit (Roche). Sequencing was performed on a HiSeq2500 system (Illumina Inc.) at VHIO using adapted protocols for TruSeq SR cluster kit v3 (Part#15023335)) following cBot\u0026trade; (Part#15006165) and HiSeq2500 System (Part#15035786) guidelines. Raw data was analysed by Real Time Analysis software v. 1\u0026middot;17\u0026middot;21\u0026middot;3 and Sequence Analysis Viewer software v. 1\u0026middot;8\u0026middot;20 (Illumina Inc.). Raw data from this sequencing were deposited in NCBI: GEO Series accession number GSE122618.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003emiR data analysis and mapping\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAfter the quality control and normalization procedures, the number of reads of a particular miR sequence was counted, discarding those without sufficient signal, as previously described (8,12,13). The counts were normalized through the weighted trimmed mean of M-values (12), and differential expression was calculated by limma\u0026rsquo;s voom methods (12,13). To make the plasmatic miR profiles and those obtained from the rectal biopsies comparable, the criteria used for the compilation and mapping in Lor\u0026eacute;n V \u003cem\u003eet al.\u003c/em\u003e was recalculated, considering fold change (FC) between R and NR differentially expressed and functionally significant if FDR\u0026lt;0\u0026middot;05, or classifier p-value\u0026lt;0\u0026middot;05, or non-adjusted p-value\u0026lt;0\u0026middot;01, and |FC|\u0026gt;1\u0026middot;5.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eSelected miRs were mapped to protein by means of MirTarBase (14) and MiRDB Database (15,16). Each miR-protein pair received a score based on the evidence rates(11) (7). When a protein was targeted by more than one miR, a list of reliable biological activity was composed. Those protein effects that were inconsistent with this list received a negative score, and those that agreed received a positive one. This score also rated the concordance with the experimental data and miR-protein correlations. Finally, scores were summarized in a matrix that integrates both proteins and miRs. \u0026nbsp;When the number of proteins per miR was too large, only up to 40 proteins with the best scores were taken. Additionally, those proteins with any contradictory information (e.g. duplicated targets) were removed, and the resulting list was stored as protein restrictions to be introduced into the models. A schematic workflow\u0026nbsp;had previously been described by Lor\u0026eacute;n et al (7).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMathematical Modelling\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eUsing artificial intelligence and pattern recognition techniques, mathematical models that could suggest mechanistic hypotheses following biological environments were generated. This process was carried out by transforming biological information into mathematical models, as previously described by Lor\u0026eacute;n \u003cem\u003eet al.\u003c/em\u003e\u0026nbsp; \u0026nbsp;(11). In summary, information about glucocorticoids and UC was compiled in a hand-curated database (Biological Effectors Database, Anaxomics, Barcelona, Spain), matching biological processes (including adverse events, indications, diseases, and pathways) to their molecular effectors. These molecular relationships were updated from public sources (KEGG, REACTOME, INTACT, BIOGRID, MINT, PubMed and DrugBank). Thus, a biological network focused on glucocorticoid targets and UC effectors was defined. The biological map was transformed into a mathematical model, mimicking current knowledge and predicting new data (Therapeutic Performance Mapping System, Anaxomics). This was achieved by training the biological network with the \u0026ldquo;Truth Table\u0026rdquo;, a collection of known stimulus-response relationships that act as mathematical restrictions. Once these rules or \u0026quot;truths\u0026quot; were established, transcriptome information was added to refine the model (mRNA and miR that meet the selection requirements described above). A mathematical model challenge with glucocorticoid traced a molecular response network that gives a relative value to each interaction between proteins (node). With this computational Systems Biology-based approach, Lor\u0026eacute;n \u003cem\u003eet al.\u0026nbsp;\u003c/em\u003echaracterized the mechanism of action (MoA) related to steroid-refractoriness in UC (supplementary material \u0026amp; methods).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003emiR biomarker identification\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eBoth biopsy and plasma miR profiles were statistically analysed to select those that possessed greater potential as biomarkers. These calculations were carried out with all the sequenced miRs or with those miRs most closely related to the MoA, both at baseline and on day three. The probability that each miR studied was related to the MoA was calculated according to its validated target proteins on R \u003cem\u003evs.\u003c/em\u003e NR model. The expression sample by sample of each miR was used to identify the miRs that better classified the samples in R and NR (explained in detail in the previous section \u0026ldquo;miR data analysis and mapping\u0026rdquo;). These miRs were then combined with each other or with clinical variables (e.g. C-Reactive Protein (CRP), smoking, delay in GC treatment, previous oral or topical treatment with mesalazine) in order to identify the \u0026nbsp;combinations that best classified R and NR samples. Different statistical approaches were used to identify potential sets of biomarkers from miR data (Table 1). The following parameters were used in order to select the best miRs classifiers, alone or in combination: a) Accuracy (proportion of samples correctly classified); b) Generalization Capability (by using a Leave-One-Out strategy, the probability of correctly predicting \u0026nbsp;any independent samples) ), c) Sensitivity (= \u003cem\u003eTrue Positives\u003c/em\u003e/\u003cem\u003ePositives\u003c/em\u003e), d) Specificity (= \u003cem\u003eTrue Negatives\u003c/em\u003e/\u003cem\u003eNegatives\u003c/em\u003e), d) Precision (= \u003cem\u003eTrue Positives\u003c/em\u003e/\u003cem\u003ePositive calls\u003c/em\u003e), e) Negative predictive value (NPV) (= \u003cem\u003eTrue Negatives\u003c/em\u003e/\u003cem\u003eNegative calls\u003c/em\u003e) (7).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eBiopsy-plasma correlation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe miRs with a greater probabilityof affecting the MoA of\u0026nbsp;corticosteroid\u0026nbsp;in the treatment of UC were identified among miRs differentially expressed\u0026nbsp;by\u0026nbsp;R and NR in plasma samples. Then, the targets of these miRs were filtered to find correlations related to the process under study: only the targets present in the model of\u0026nbsp;corticosteroid\u0026nbsp;treatment of UC were considered. These MoA related plasma miRs (pmiRs) and related targets were compared with different sources of information from biopsy (Figure 1):\u003c/p\u003e\n\u003col\u003e\n \u003cli\u003ePlasma miRs vs. biopsy differential miRs\u0026nbsp;\u003c/li\u003e\n \u003cli\u003ePlasma miRs MoA related targets vs. biopsy differential miRs targets.\u003c/li\u003e\n \u003cli\u003ePlasma miRs MoA related targets vs. proteins translated from differential mRNAs.\u003c/li\u003e\n \u003cli\u003ePlasma miRs MoA related targets vs. key proteins obtained from the MoA analysis in the previous phases of the project.\u003c/li\u003e\n\u003c/ol\u003e\n\u003cp\u003eThe correlation described in point 4 was also assessed in an indirect setting, i.e. evaluating if there is any interaction in Anaxomics databases relating both groups of proteins (distance 1, D1). This protocol was performed both at baseline and at day three of treatment.\u003c/p\u003e\n\u003cp\u003eThe\u003cem\u003eIn silico\u003c/em\u003e functional analysis of selected miRNAs included target prediction using miRTarBase, DIANA, and miRDB, followed by functional pathway enrichment using Gene Ontology (GO), KEGG, and STRING databases. A selection score was developed to prioritize mRNA targets based on corticosteroid response relevance, interspecies conservation, and evidence from human datasets. This integrated analysis allowed us to identify the most functionally relevant miRNA-mRNA interactions in the context of glucocorticoid sensitivity and inflammation resolution in colitis.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eValidation of candidate miR biomarkers in a TNBS-induced colitis mouse model\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTen to twelve-week-old C57BL/6J mice (male and female) were purchased from Jackson Laboratory (Charles River; Spain) and were housed in the Germans Trias i Pujol Research Institute (IGTP) conventional facility. The studies were conducted in compliance with animal research guidelines. All the experiments were approved by the Institutional Animal Care and Use Committee of the IGTP and authorized by the Catalan Government (no. 9933).\u003c/p\u003e\n\u003cp\u003eColitis was induced by intrarectal instillation of 3 mg TNBS in 100 \u0026micro;L 50% ethanol\u0026nbsp;(17). Mice were randomized into four groups (n=10/group; sex-balanced): (i) healthy controls; (ii) TNBS colitis (endpoint day 3); (iii) TNBS + vehicle (PBS, subcutaneous, days 4\u0026ndash;7); (iv) TNBS + dexamethasone (1.5 mg/kg s.c., days 4\u0026ndash;7). Dexamethasone dosing was based on prior dose-response screenings.\u003c/p\u003e\n\u003cp\u003eInflammatory response was monitored longitudinally using bioluminescence imaging (BLI), as previously described by Suau et al.Suau \u003cem\u003eet al\u003c/em\u003e. (17). Images were acquired at multiple time points and normalized to each animal\u0026rsquo;s pre-induction baseline signal. Inflammation levels were expressed as FC relative to this baseline. To assess treatment effect, we also calculated the average FC between colitis onset (day 0) and treatment initiation (day 4) and compared it to values during the treatment phase (days 4\u0026ndash;7). To objectively evaluate treatment response, a composite BLI-based score (range 0\u0026ndash;4) was established, based on dynamic changes in FC after corticosteroid administration: if FC decreased (i.e., negative FC) on\u0026nbsp;two separate days\u0026nbsp;during days 4\u0026ndash;7 (+1 point); if the\u0026nbsp;average FC\u0026nbsp;over the entire treatment period was negative (+1 point); if FC was\u0026nbsp;negative at day 7 (+2 point). Based on the total score: a) Mice scoring\u0026nbsp;0 or 1\u0026nbsp;were classified as\u0026nbsp;non-responders (NR); b) mice scoring\u0026nbsp;2 to 4\u0026nbsp;were classified as\u0026nbsp;responders (R).\u003c/p\u003e\n\u003cp\u003eAfter euthanasia (endpoint day 7), colon samples were cpllected, preserved in RNAlater, and stored at \u0026minus;80\u0026deg;C. Total RNA was extracted (miRNeasy Mini kit/QIAcube system, QIAGEN, Germany) from homogenized (gentleMACS\u0026trade;; Miltenyi Biotech, Germany) 25 mg colon tissue, and integrity was confirmed (RIN \u0026ge;6.5, Agilent Bioanalyzer, Nano kit and SmallRNA kit, Agilent Technologies, USA).\u003c/p\u003e\n\u003cp\u003eFor mRNA quantification, 1 \u0026micro;g of RNA was retrotranscribed (PrimeScript RT kit, Takara, Japan), and qRT-PCR was performed (TaqMan\u0026trade; probes for\u0026nbsp;Creb1,\u0026nbsp;Nr3c1,\u0026nbsp;Smad7,\u0026nbsp;Vegfa; ThermoFisher Scientific, Spain) using a LightCycler480 (Roche Diagnostics, Switzerland). Expression was normalized to\u0026nbsp;Gapdh\u0026nbsp;and\u0026nbsp;B2m. For miR quantification, reverse transcription and qRT-PCR were performed using the TaqMan\u0026trade; Advanced miRNA cDNA Synthesis Kit and specific probes (mmu-miR-224-5p, -10b-5p, -218-5p, -145a-5p, -183-5p, -16-5p). Data were normalized to mmu-miR-30e-5p and mmu-miR-103-3p. Data from qRT-PCR were again calculated using the 2\u003csup\u003e\u0026minus;\u0026Delta;\u0026Delta;Ct\u003c/sup\u003e method (18).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStandard statistics\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll quantitative data were checked for normality using a Shapiro-Wilk test. We also applied the Mann\u0026ndash;Whitney U test in pairwise comparisons between different animal groups at the same time point. The comparisons of the qRT-PCR data between treatment groups and sexes were performed with the Kruskal\u0026ndash;Wallis test with a post hoc Dunn\u0026rsquo;s test analysis or a Mann\u0026ndash;Whitney U test. Comparisons with p \u0026le; 0\u0026middot;05 were statistically significant. All these analyses were performed with R software version 4\u0026middot;2\u0026middot;0.\u003cstrong\u003e\u003cbr\u003e\u003c/strong\u003e\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eTranscriptomic and microRNA profiling distinguishes corticosteroid responders from non-responders in UC patients\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThis study included 24 patients with UC, categorised at day seven as R (n=13) or NR (n=11) to corticosteroid treatment. To investigate the mechanisms underlying response to corticosteroids, mRNA and miR expression levels were analysed at baseline (day 0) and on day three of treatment. Two types of potential biomarkers were identified: protein biomarkers derived from mRNA and miR data. Systems biology approaches were employed to minimize false discoveries and enhance the robustness of biomarker identification.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eProtein-coding transcript biomarkers linked to corticosteroid responsiveness\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eCandidate biomarker transcripts coding for protein were identified by integrating mRNA transcriptome data with models built using both mRNA and miR information. Separate analyses were conducted for baseline and day three samples, comparing R and NR groups. A bibliographic review was performed to assess the relevance of identified proteins to UC pathology or corticosteroid treatment.\u003c/p\u003e\n\u003cp\u003eAt baseline, 18 protein biomarker combinations were identified as the best candidates to differentiate between R and NR patients. Similarly, 15 combinations were identified at day three. These combinations were evaluated for accuracy and generalization capability using Therapeutic Performance Mapping System strategies (19), which yielded the most reliable candidates. Table 2 summarizes the top-performing combinations of three protein biomarkers for each time point. This systems biology analysis identified proteins linked to both UC and corticosteroid treatment at basal and day three levels. These proteins not only serve as biomarkers but also provide mechanistic insights into the response to corticosteroids.\u003c/p\u003e\n\u003cp\u003eIn this study, four key co-activators and co-repressors affecting GCR transcriptional activity were identified: NCOA3, CBP, NCOR1, and NRIP1. These proteins influence transcriptional activity related to corticosteroid response and inflammatory pathways.\u003c/p\u003e\n\u003cp\u003eThese findings may help stratify patients at baseline, allowing for the early identification of individuals at risk of steroid refractoriness and guiding personalised treatment strategies.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMicroRNAs modulating steroid response: identification and biomarker potential\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eTo explore the post-transcriptional regulation mechanisms involved in corticosteroid response, we analyzed miRs from both plasma and rectal biopsies. Our goal was to identify miRs that not only differentiate R from NR, but that are also mechanistically integrated into the corticosteroid MoA. miRs influence gene expression pathways linked to GCR signalling and inflammation. Plasma and biopsy samples were analyzed at baseline and day three. Two complementary approaches were applied: (1) using all sequenced miRs and (2) focusing on MoA-related miRs. The best miR combinations were selected based on classification accuracy, generalization capability, and minimal descriptor count (Table 3).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMechanistic insights into miRNA-mediated regulation of corticosteroid pathways\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA mechanistic analysis of the best miR biomarker combinations was performed to provide further biological justification. The identified miRs were analysed for their interactions with key molecular pathways involved in corticosteroid response. Figures 2 and 3 illustrate the relationship between miR targets and proteins within the previously reported MoA at baseline and day three, respectively.\u003c/p\u003e\n\u003cp\u003eAt \u003cstrong\u003ebaseline,\u0026nbsp;\u003c/strong\u003ethe identified plasma miRs, such as hsa-miR-10a-5p and hsa-miR-324-5p, influenced GCR function by targeting chaperones, such as HS71A/B and co-activators essential for GCR activity, like NCOAs \u003cstrong\u003e(Figure 2)\u003c/strong\u003e. These miRs also regulated NF-κB, TNF-α and VEGF pathways, which are involved in inflammation and corticosteroid resistance. Specific targets included ETS1 and P53, key regulators of transcriptional and apoptotic pathways.\u003c/p\u003e\n\u003cp\u003eAt \u003cstrong\u003eday\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003ethree\u003c/strong\u003e\u003cstrong\u003e,\u0026nbsp;\u003c/strong\u003emiR-20a-5p and miR-181a were identified as key regulators of GCR activity, while hsa-miR-433 was linked to PI3K modulation through its interaction with the adaptor protein GRB2 \u003cstrong\u003e(Figure 3)\u003c/strong\u003e. These findings highlight the dynamic nature of miR regulation in the response to corticosteroids.\u003c/p\u003e\n\u003cp\u003eThe regulatory roles of these miRs over GCR-related pathways suggest that they are not only passive markers but could actively modulate corticosteroid responsiveness in UC.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCirculating–tissue miRNA concordance supports systemic biomarker validity\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eStatistically significant miRs from plasma samples (adjusted p-value \u0026lt; 0·05) were correlated with miRs, mRNA, and model information from biopsy samples at baseline and day three. At baseline(Figure 4), five plasma miRs (hsa-miR-194-5p, hsa-miR-145-5p, hsa-miR-216a-5p, miR-224-5p and hsa-miR-487a-3p) were identified as potentially affecting corticosteroid response. These miRs were related to biopsy findings through common protein targets and relationships with key proteins from biopsy models, such as VEGFA. Specifically, this protein was shown to be targeted by the plasma hsa-miR-145-5p \u0026nbsp;and biopsy hsa-miR-16-5p (Figure 4A and B). \u0026nbsp;This cross-validation between tissue and circulating miRs reinforces their potential as minimally invasive biomarkers for monitoring or predicting treatment outcomes.\u003c/p\u003e\n\u003cp\u003eAt day three,hsa-miR-214-3p was found to be significantly differential and was correlated mechanistically with rectal biopsy miRs and key proteins involved in the response to corticosteroids (Figure 5). Briefly, hsa-miR-214-3p shared mRNA targets with hsa-miR-625-3p, hsa-miR-29a-3p, miR-423-3p and hsa-miR-10b-5p. Many of these mRNA targets were at the same time related to the MoA identified proteins, especially P53, EZH2 and CTNB1.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eValidation in TNBS-colitis model reveals conserved corticosteroid-regulatory miRNAs\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cu\u003eBioluminescence and macroscopic indices\u003c/u\u003e\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eBased on the bioluminescence FC relative to baseline, mice were classified as R or NR, following a methodology \u0026nbsp;used in a previous study (17). The BLI results showed that the R were able to reduce their intestinal inflammation from the sixth day of treatment (Figure 6A). \u0026nbsp;Consistently, the colons of R mice to corticosteroids were significantly lighter and longer compared to non-treated mice, reflecting reduced oedema and ulceration (Figure 6B).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cu\u003emiRNA profiling\u0026nbsp;\u003c/u\u003e\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eSix miRNAs previously linked to human UC were conserved and quantifiable in the TNBS model (Figure 6C). miR‑10b‑5p, miR‑145a‑5p and miR‑16‑5p were markedly down‑regulated in R versus NR mice (p \u0026lt; 0·05), suggesting their association with steroid efficacy. miR‑224‑5p and miR‑183‑5p were up‑regulated at day seven irrespective of response, indicating a broader association with inflammation.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cem\u003e\u003cu\u003eTarget‑gene expression\u003c/u\u003e\u003c/em\u003e\u003c/p\u003e\n\u003ch3\u003e\u003cem\u003eCreb1\u003c/em\u003e (a shared target of miR‑10b‑5p/16‑5p/218‑5p) was up‑regulated in all TNBS groups (Figure 6D), echoing its pro‑inflammatory role in human UC.\u0026nbsp;\u003cem\u003eSmad7\u003c/em\u003e, potentially under negative post-transcriptional regulation by miR-16-5p (Figure 6E), was significantly increased in R mice, aligning with the resolution of TGF-β signalling. In contrast, \u003cem\u003eNr3c1\u003c/em\u003e (Gcr) expression decreased in NR mice, reflecting the glucocorticoid resistance observed in patients. \u003cem\u003eVegfa\u003c/em\u003e expression remained unchanged, suggesting species-specific regulatory differences\u003c/h3\u003e\n\u003cp\u003e\u003cem\u003e\u003cu\u003emiRNA–mRNA correlations\u0026nbsp;\u003c/u\u003e\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFinally, integrative correlation analyses across all experimental groups (Figure 6D) revealed consistent inverse relationships between specific miRNAs and key proteins involved in corticosteroid response. Notably, miR-183-5p showed a significant negative correlation with Gcr (ρ = –0·47, \u003cem\u003ep\u003c/em\u003e = 0·01), while miR-218-5p inversely correlated with Creb1 (ρ = –0·38, \u003cem\u003ep\u003c/em\u003e = 0·037). A similar trend was observed for miR-10b-5p, which also negatively correlated with Creb1(ρ = –0·30, \u003cem\u003ep\u003c/em\u003e = 0·098), and miR-16-5p, which inversely correlated with Smad7 (ρ = –0·32, \u003cem\u003ep\u003c/em\u003e = 0·084) and Vegfa(ρ = –0·35, \u003cem\u003ep\u003c/em\u003e = 0·055).\u003c/p\u003e\n\u003cp\u003eThese relationships were characterized by a conserved pattern in which low miRNA expression levels aligned with elevated expression of their putative targets (Figure 6C \u0026amp; 6D), suggesting a functional post-transcriptional regulatory mechanism. Such conserved inverse correlations reinforce the hypothesis that miR-10b-5p, miR-218-5p, miR-183-5p, and miR-16-5p actively participate in modulating corticosteroid-related pathways. These findings provide mechanistic support for the functional relevance of the selected miRNA panel and its potential utility as a regulatory signature in corticosteroid responsiveness.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, by integrating protein biomarkers with miR biomarkers, we provide a comprehensive molecular profile of corticosteroid response in UC human and mouse model contexts, offering potential predictive and mechanistic insights. The identified protein biomarkers play critical roles in GCR signalling and transcriptional regulation. NCOR1 and NRIP1 function as transcriptional corepressors, modulating GCR activity and inflammatory responses \u0026nbsp;(20,21), potentially contributing to corticosteroid resistance. In contrast, NCOA3 and CBP act as co-activators, enhancing GCR-mediated transcriptional activity and influencing inflammatory and apoptotic pathways (22). These findings highlight the complex interplay between co-repressors and co-activators according to corticosteroid sensitivity in patients with UC. This mechanistic positioning reinforces the value of these molecules not merely as biomarkers but as active effectors within the corticosteroid response cascade.\u003c/p\u003e\n\u003cp\u003eThis study provides insight into the molecular status of UC patients before and after starting corticosteroid treatment, elucidating key molecules involved in corticosteroid response. Some of the proteins identified at baseline, such as GCR, IL32, and ATG5, reflect a pre-existing molecular state and glucocorticoid sensitivity.\u0026nbsp;\u003cstrong\u003eGCR\u003c/strong\u003e is the primary receptor mediating\u0026nbsp;corticosteroid\u0026nbsp;effects, representing baseline glucocorticoid sensitivity (22).\u0026nbsp;IL32,\u0026nbsp;which is\u0026nbsp;linked to chronic inflammation\u0026nbsp;(23), may indicate an elevated inflammatory burden predictive of treatment resistance. ATG5, a regulator of autophagy\u0026nbsp;(24), suggests a potential role in cellular stress responses influencing treatment outcomes. These proteins offer the potential for\u0026nbsp;the\u0026nbsp;early stratification of patients into likely R and NR before corticosteroid administration, enabling personalized treatment strategies. Proteins uniquely regulated at day\u0026nbsp;three, such as IKKB and NCOA2, illustrate dynamic molecular changes induced by corticosteroid exposure. IKKB, an activator of the NF-κB pathway (25), suggests ongoing inflammatory signalling despite corticosteroid action. NCOA2, a steroid hormone receptor coactivator (22), indicates adaptive transcriptional responses to glucocorticoid therapy. Additionally, proteins consistently present at both time points (basal and day\u0026nbsp;three), including CBP, NFKB1, and CREB1, emphasize persistent transcriptional regulation and inflammatory control mechanisms that may be critical to corticosteroid efficacy. These biomarkers also intersect with key signalling pathways such as NF-κB and p53, which are crucial for inflammatory regulation and apoptosis\u0026nbsp;(26). The interplay between these pathways underscores the complexity of the mechanisms of response to corticosteroids. A previous study by our research group analysing the molecular mechanisms underlying steroid failure in UC identified NF-κB as a key player, in which responders showed increased GCR sensitivity and downregulation of its downstream effectors, particularly ETS1, RELA, and VEGFA, while CASP8 and P53 were markedly upregulated in R compared to NR\u0026nbsp;(7).\u003c/p\u003e\n\u003cp\u003eHere, NCOA3, CBP, NCOR1, and NRIP1 were identified as the best potential biomarkers due to their combined influence on GCR transcriptional activity. NCOA3 and CBP function as co-activators, whereas NCOR1 and NRIP1 act as co-repressors (20–22). Notably, CBP also modulates NF-κB activity via TF65 (27,28) and ETS1\u0026nbsp;(29)\u0026nbsp;and promotes pro-angiogenic VEGFA and VGFR1 expression\u0026nbsp;(30).\u0026nbsp;This highlights the dual regulatory role of CBP, balancing both anti-inflammatory and pro-inflammatory pathways, which may be critical in determining\u0026nbsp;corticosteroid\u0026nbsp;efficacy.\u0026nbsp;In addition, CBP also affects P53 transcriptional activity, however, depending on the cellular location, the effects are opposed: in the cytoplasm, CBP causes the degradation of P53, while \u0026nbsp;it activates it in the nucleus (31). It is important to note that the co-activators/co-repressors, in addition to affecting transcription factors (TFs), are also affected by the interaction with these TFs, at least impeding their association with other TFs (as could be happening in the interactions CBP-GCR and CBP-other TFs). Hence, they should be taken into account, in particular in the case of CBP, which regulates several TFs that are involved in the MoA.\u003c/p\u003e\n\u003cp\u003eThe identification of miR biomarkers in both mucosal rectal biopsies and plasma samples provides valuable molecular signatures for distinguishing\u0026nbsp;corticosteroid\u0026nbsp;R from NR. However, the clinical utility of these biomarkers depends on understanding their systemic versus local expression patterns. By examining correlations between miRs in plasma and mucosal rectal biopsies, as well as by linking plasma miRs to key proteins identified from biopsy data, insights can be gained into how intestinal tissue pathology is reflected in circulating biomarkers. This integrative approach enhances the potential for non-invasive diagnostic tools while uncovering mechanistic relationships critical to\u0026nbsp;corticosteroid\u0026nbsp;resistance.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eImportantly, two miRs show a significantly differential expression between R and NR at baseline: miR-224-5p, which has been found to be related to the process of interest, and miR-487a-3p. Whereas the latter appears to increase in both mucosal and plasmatic samples in the NR cohort, the former appears to be higher in the rectal mucosa of NR and lower in the blood of the same cohort. Reports suggest that some cancer cells can extrude miRs via exosomes to maintain their oncogenesis (32), thus leading to increased levels of the miR in serum (33) while reducing it in the tissue. In addition, miR-224 has been proven to be upregulated in colorectal cancer tissue and has been related to its development through promoting\u0026nbsp;proliferation\u0026nbsp;(34). The mechanistic links of these miRs with key transcriptional pathways support their candidacy as functional predictors rather than surrogate markers.\u003c/p\u003e\n\u003cp\u003eThe identification of miR biomarkers, particularly \u003cstrong\u003emiR-214-3p and miR-10b-5p\u003c/strong\u003e\u003cstrong\u003e,\u0026nbsp;\u003c/strong\u003estrengthens the robustness of our findings. These miRs were found to correlate strongly with rectal biopsy-derived data, reinforcing their potential as \u003cstrong\u003enon-invasive diagnostic markers\u003c/strong\u003e. Additionally, both miRs modulate \u003cstrong\u003eP53 signalling and inflammatory mediators\u003c/strong\u003e, providing a mechanistic justification for their role in corticosteroid response.\u003c/p\u003e\n\u003cp\u003eThe correlation between plasma and rectal biopsy miRs highlights the systemic nature of corticosteroid response, suggesting that plasma-derived miRs could serve as \u003cstrong\u003esurrogate markers for intestinal inflammation\u003c/strong\u003e. Notably, miR-224-5p showed opposite expression patterns in plasma and biopsy samples, a phenomenon observed in cancer studies where exosomal miR export affects tissue miR levels (35).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eInterestingly, our findings highlight VEGF as a key protein identified in the biopsy analysis.\u0026nbsp;Focusing on this protein, which is known to be overexpressed in UC and to play a role in its development (36), it was further examined in relation to miR regulation. At baseline, we observed that, in R patients, plasma hsa-miR-145-5p changes, targeting VEGF, potentially contributing to the modulation of its expression. Notably, this regulatory interaction was no longer evident by day three following corticosteroid treatment, suggesting a possible mechanism through which R patients may achieve clinical improvement. The loss of this interaction post-treatment could reflect a normalisation of VEGF expression levels or a shift in regulatory dynamics as inflammation subsides.\u003c/p\u003e\n\u003cp\u003eThe TNBS colitis model recapitulated key aspects of human corticosteroid resistance, validating miR-145a-5p as a conserved biomarker of poor therapeutic response. This miR downregulation in both mR and mNR mice underscores its role in sustaining inflammation, potentially through \u003cem\u003eCreb1\u003c/em\u003e-mediated pathways. The selective suppression of miR-10b-5p and miR-16-5p in mR aligns with their putative roles in enhancing \u003cem\u003eGcr\u003c/em\u003e sensitivity and \u003cem\u003eVegfa\u003c/em\u003e signalling, respectively (see section 5 of results). However, the absence of miR-224-5p differential expression in mice contrasts with human findings, possibly due to anatomical differences or compensatory mechanisms in acute murine colitis. Similarly, the lack of \u003cem\u003eVegfa\u003c/em\u003e changes highlights the limitations in translating angiogenic pathways across species, urging caution in extrapolating mechanistic insights.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe negative correlation between miR-10b-5p and miR-218-5p with \u003cem\u003eCreb1\u003c/em\u003e suggests a potential post-transcriptional regulatory mechanism in which these miRs may contribute to the fine-tuning of \u003cem\u003eCreb1\u003c/em\u003e expression in response to corticosteroid treatment. \u003cem\u003eCreb1\u003c/em\u003e plays a key role in inflammatory signalling and \u003cem\u003eGcr\u003c/em\u003e activation, making these miRs relevant candidates for modulating corticosteroid sensitivity (27). The significant correlation of miR-218-5p (p=0.037) reinforces this idea, suggesting that its downregulation could be associated with increased \u003cem\u003eCreb1\u003c/em\u003e expression, potentially influencing treatment resistance.\u003c/p\u003e\n\u003cp\u003eSimilarly, the negative correlation of miR-183-5p with GCR highlights its potential role in glucocorticoid signalling regulation. Given that \u003cem\u003eGcr\u003c/em\u003e is the primary mediator of corticosteroid action, an increase in miR-183-5p levels could contribute to a reduced \u003cem\u003eGcr\u003c/em\u003e expression, impairing corticosteroid response. This aligns with previous findings where GCR dysfunction has been linked to steroid resistance in UC patients (37).\u003c/p\u003e\n\u003cp\u003eMoreover, miR-16-5p's negative correlation with \u003cem\u003eSmad7\u003c/em\u003e and \u003cem\u003eVegfa\u003c/em\u003e further supports its involvement in \u003cem\u003eTgf-β\u003c/em\u003e and angiogenesis pathways. \u003cem\u003eSmad7\u003c/em\u003e is a well-known inhibitor of \u003cem\u003eTgf-β\u003c/em\u003e signalling (38).\u0026nbsp;Its negative correlation with miR-16-5p suggests a potential mechanism in which higher miR-16-5p levels could suppress Smad7, thereby promoting TGF-β activity and tissue repair. Likewise, \u003cem\u003eVegfa\u003c/em\u003e, a key angiogenic factor (39), also showed a negative correlation with miR-16-5p, indicating a possible role in vascular remodelling and mucosal healing.\u003c/p\u003e\n\u003cp\u003eDespite the suggesting evidence provided by this study, some limitations should be noted. Although the plasma miRs identified in this study showed a high discriminatory power in the discovery cohort, we were unable to replicate these findings in an independent validation set due to the presence of hemolysis in most of those samples.\u0026nbsp;However, it is worth noting that none of the samples included in this study showed clear signs of hemolysis.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;This underscores a technical limitation in circulating miR analysis which may confound quantification and obscure biological signals. Future validation studies should ensure rigorous preanalytical controls to avoid such confounders. Notably, mucosal-derived signals, and particularly biopsy miRs, appeared more robust and were not subject to such limitations, strengthening their value as mechanistically and technically reliable predictors.\u003c/p\u003e\n\u003cp\u003eImportantly, several of the circulating miRs identified, such as miR-145-5p and miR-10b-5p, were found to be mechanistically linked to rectal tissue miRs, shared common targets, and showed consistent behaviours in the TNBS-induced colitis murine model. This cross-validation across compartments and species reinforces their biological plausibility, suggesting that these circulating miRNAs could be part of a conserved corticosteroid-response network in ulcerative colitis. Therefore, plasma-derived miRs from the discovery cohort are justified within the framework of an integrative systems biology approach, where convergence of evidence, rather than replication alone, builds mechanistic confidence. Future studies in non-hemolyzed plasma samples are necessary to assess the robustness and translational potential of these candidate biomarkers.\u003c/p\u003e\n\u003cp\u003eOverall, the integration of multi-layered data anchored in the MoA of corticosteroids highlights a subset of molecular signals that are mechanistically grounded, thereby offering a robust platform for future biomarker validation and precision medicine strategies in UC.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eConflict of Interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors have no conflicts of interest to declare.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGene profiling and miR sequencing data from the biopsy samples were deposited in NCBI\u0026rsquo;s GEO and are accessible through the GEO Series accession number GSE114527 and GSE114591. Data from the miR sequencing from plasma samples were also deposited in NCBI\u0026rsquo;s GEO and are accessible through the GEO Series accession number GSE122618. The other datasets generated and/or analyzed during the current study are available from the corresponding author on reasonable request.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eFunding\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;This study was supported by public funding from the Instituto de Salud Carlos III (ISCIII, Ministry of Health, Government of Spain) through the following projects: PI22/01498, PI18/00892, and PI16/01937, all co-funded by the European Union (European Regional Development Fund \u0026ndash; ERDF, \u0026ldquo;A way to make Europe\u0026rdquo;). Additional support was provided by ACCI\u0026Oacute; \u0026ndash; Generalitat de Catalunya (project VALUNI16-1-0001). The authors also acknowledge economic and institutional support from CIBER (ISCIII), IISPV and the IGTP.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe acknowledge the patients and the IGTP BioBank for its collaboration in sample collection and processing. We thank the VHIO Cancer Genomics Group for their excellent support with NGS services of small RNA sequencing of plasma samples. We would also like to acknowledge the participation of Mariona Llaves and Eva Jou on the miR and mRNA analysis of mice samples and Ana Garcia for her advice on animal procedures. We thank Mireia Coma from Anaxomics Biotech S.L. for the implementation of the mathematical model in our analysis and the relation between the plasma and biopsy miR. Finally, we also wish to thank Eduard Cabr\u0026eacute; for his participation in the study as clinician through patient recruitment and sample gathering.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eI.G. and R.S. conducted the main experiments. J.E.N. was responsible for patient recruitment, inclusion, and sample collection. V.L., I.G., and R.S. prepared samples for sequencing. A.M.A. performed the biopsy microarray and sequencing, along with the corresponding differential expression analyses. R.P., L.S., M.L.B., G.R., and J.F.SH. carried out plasma sequencing and its subsequent differential expression analysis. C.S. implemented the mathematical model and integrated plasma miRNA data with biopsy findings. I.G., R.S., L.C., and C.B. conducted the animal experiments, including sample collection and subsequent in silico and qRT-PCR analyses. J.M., C.Ser., L.S., E.D., and M.M. conceptualized and secured public competitive funding for the project. E.D., M.M., and L.S. provided critical revision of the manuscript. I.G., R.S., and J.M. interpreted the data and wrote the first draft of the manuscript, with R.S. contributing significantly to all aspects of the work. J.M. coordinated the overall project. J.M. and C.Ser. are guarantors of this work.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eKaplan GG, Ng SC. Understanding and Preventing the Global Increase of Inflammatory Bowel Disease. Gastroenterology. 2017;152(2).\u003c/li\u003e\n\u003cli\u003eLla\u0026oacute; J, Naves JE, Ruiz-Cerulla A, Mar\u0026iacute;n L, Ma\u0026ntilde;osa M, Rodr\u0026iacute;guez-Alonso L, et al. Intravenous corticosteroids in moderately active ulcerative colitis refractory to oral corticosteroids. J Crohns Colitis. 2014;8(11).\u003c/li\u003e\n\u003cli\u003eSalice M, Rizzello F, Calabrese C, Calandrini L, Gionchetti P. A current overview of corticosteroid use in active ulcerative colitis. Vol. 13, Expert Review of Gastroenterology and Hepatology. 2019.\u003c/li\u003e\n\u003cli\u003eRamamoorthy S, Cidlowski JA. Ligand-induced repression of the glucocorticoid receptor gene is mediated by an NCoR1 repression complex formed by long-range chromatin interactions with intragenic glucocorticoid response elements. Mol Cell Biol [Internet]. 2013 May 1 [cited 2025 Mar 10];33(9):1711\u0026ndash;22. Available from: https://pubmed.ncbi.nlm.nih.gov/23428870/\u003c/li\u003e\n\u003cli\u003eVan Moortel L, Verhee A, Thommis J, Houtman R, Melchers D, Delhaye L, et al. Selective Modulation of the Human Glucocorticoid Receptor Compromises GR Chromatin Occupancy and Recruitment of p300/CBP and the Mediator Complex. Mol Cell Proteomics [Internet]. 2024 Mar 1 [cited 2025 Mar 10];23(3). Available from: https://pubmed.ncbi.nlm.nih.gov/38387774/\u003c/li\u003e\n\u003cli\u003eMacLellan SA, Lawson J, Baik J, Guillaud M, Poh CFY, Garnis C. Differential expression of miRNAs in the serum of patients with high-risk oral lesions. Cancer Med [Internet]. 2012 [cited 2025 Mar 10];1(2):268\u0026ndash;74. Available from: https://pubmed.ncbi.nlm.nih.gov/23342275/\u003c/li\u003e\n\u003cli\u003eLor\u0026eacute;n V, Garcia-Jaraquemada A, Naves JE, Carmona X, Ma\u0026ntilde;osa M, Aransay AM, et al. ANP32E, a Protein Involved in Steroid-Refractoriness in Ulcerative Colitis, Identified by a Systems Biology Approach. J Crohns Colitis [Internet]. 2019 Mar 26 [cited 2025 Jan 5];13(3):351\u0026ndash;61. Available from: https://pubmed.ncbi.nlm.nih.gov/30329026/\u003c/li\u003e\n\u003cli\u003eMontreal Classification for Inflammatory Bowel Disease (IBD) [Internet]. [cited 2025 May 14]. Available from: https://www.mdcalc.com/calc/10237/montreal-classification-inflammatory-bowel-disease-ibd\u003c/li\u003e\n\u003cli\u003eShi W, Oshlack A, Smyth GK. Optimizing the noise versus bias trade-off for Illumina whole genome expression BeadChips. Nucleic Acids Res [Internet]. 2010 Dec [cited 2025 May 14];38(22). Available from: https://pubmed.ncbi.nlm.nih.gov/20929874/\u003c/li\u003e\n\u003cli\u003eSmyth GK, Michaud J, Scott HS. Use of within-array replicate spots for assessing differential expression in microarray experiments. Bioinformatics [Internet]. 2005 May 1 [cited 2025 May 14];21(9):2067\u0026ndash;75. Available from: https://pubmed.ncbi.nlm.nih.gov/15657102/\u003c/li\u003e\n\u003cli\u003eLor\u0026eacute;n V, Many\u0026eacute; J, Fuentes MC, Cabr\u0026eacute; E, Ojanguren I, Espadaler J. Comparative Effect of the I3.1 Probiotic Formula in Two Animal Models of Colitis. Probiotics Antimicrob Proteins [Internet]. 2017 Mar 1 [cited 2025 May 14];9(1):71\u0026ndash;80. Available from: https://pubmed.ncbi.nlm.nih.gov/27832441/\u003c/li\u003e\n\u003cli\u003eRobinson MD, Oshlack A. A scaling normalization method for differential expression analysis of RNA-seq data. Genome Biol [Internet]. 2010 Mar 2 [cited 2025 May 14];11(3). Available from: https://pubmed.ncbi.nlm.nih.gov/20196867/\u003c/li\u003e\n\u003cli\u003eRitchie ME, Phipson B, Wu D, Hu Y, Law CW, Shi W, et al. Limma powers differential expression analyses for RNA-sequencing and microarray studies. Nucleic Acids Res [Internet]. 2015 Jan 6 [cited 2025 May 14];43(7):e47. Available from: https://pubmed.ncbi.nlm.nih.gov/25605792/\u003c/li\u003e\n\u003cli\u003eHsu S Da, Tseng YT, Shrestha S, Lin YL, Khaleel A, Chou CH, et al. MiRTarBase update 2014: An information resource for experimentally validated miRNA-target interactions. Nucleic Acids Res [Internet]. 2014 Jan 1 [cited 2025 May 14];42(D1). Available from: https://pubmed.ncbi.nlm.nih.gov/24304892/\u003c/li\u003e\n\u003cli\u003eWang X, El Naqa IM. Prediction of both conserved and nonconserved microRNA targets in animals. Bioinformatics [Internet]. 2008 [cited 2025 May 14];24(3):325\u0026ndash;32. Available from: https://pubmed.ncbi.nlm.nih.gov/18048393/\u003c/li\u003e\n\u003cli\u003eWang X. miRDB: A microRNA target prediction and functional annotation database with a wiki interface. RNA [Internet]. 2008 Jun [cited 2025 May 14];14(6):1012\u0026ndash;7. Available from: https://pubmed.ncbi.nlm.nih.gov/18426918/\u003c/li\u003e\n\u003cli\u003eSuau R, Garcia A, Bernal C, Llaves M, Schiering K, Jou-Oll\u0026eacute; E, et al. Response Variability to Drug Testing in Two Models of Chemically Induced Colitis. Int J Mol Sci [Internet]. 2023 Apr 1 [cited 2025 May 14];24(7). Available from: https://pubmed.ncbi.nlm.nih.gov/37047397/\u003c/li\u003e\n\u003cli\u003eLivak KJ, Schmittgen TD. Analysis of relative gene expression data using real-time quantitative PCR and the 2-\u0026Delta;\u0026Delta;CT method. Methods [Internet]. 2001 [cited 2025 May 23];25(4):402\u0026ndash;8. Available from: https://pubmed.ncbi.nlm.nih.gov/11846609/\u003c/li\u003e\n\u003cli\u003eMart\u0026iacute;n-Segura A, Casadom\u0026eacute;-Perales \u0026Aacute;, Fazzari P, Mas JM, Artigas L, Valls R, et al. Aging increases hippocampal DUSP2 by a membrane cholesterol loss-mediated RTK/p38MAPK activation mechanism. Front Neurol [Internet]. 2019 Jun 25 [cited 2025 Jul 2];10(JUN):465444. Available from: www.frontiersin.org\u003c/li\u003e\n\u003cli\u003eRamamoorthy S, Cidlowski JA. Ligand-induced repression of the glucocorticoid receptor gene is mediated by an NCoR1 repression complex formed by long-range chromatin interactions with intragenic glucocorticoid response elements. Mol Cell Biol [Internet]. 2013 May 1 [cited 2025 Jan 6];33(9):1711\u0026ndash;22. Available from: https://pubmed.ncbi.nlm.nih.gov/23428870/\u003c/li\u003e\n\u003cli\u003eMonczor F, Chatzopoulou A, Zappia CD, Houtman R, Meijer OC, Fitzsimons CP. A model of glucocorticoid receptor interaction with coregulators predicts transcriptional regulation of target genes. Front Pharmacol. 2019;10(MAR).\u003c/li\u003e\n\u003cli\u003eVan Moortel L, Verhee A, Thommis J, Houtman R, Melchers D, Delhaye L, et al. Selective Modulation of the Human Glucocorticoid Receptor Compromises GR Chromatin Occupancy and Recruitment of p300/ CBP and the Mediator Complex. Molecular and Cellular Proteomics. 2024;23(3).\u003c/li\u003e\n\u003cli\u003eXin T, Chen M, Duan L, Xu Y, Gao P. Interleukin-32: Its role in asthma and potential as a therapeutic agent. Vol. 19, Respiratory Research. 2018.\u003c/li\u003e\n\u003cli\u003eChangotra H, Kaur S, Yadav SS, Gupta GL, Parkash J, Duseja A. ATG5: A central autophagy regulator implicated in various human diseases. Vol. 40, Cell Biochemistry and Function. 2022.\u003c/li\u003e\n\u003cli\u003eFreitas RHCN, Fraga CAM. NF-\u0026kappa;B-IKK\u0026beta; Pathway as a Target for Drug Development: Realities, Challenges and Perspectives. Curr Drug Targets. 2018;19(16).\u003c/li\u003e\n\u003cli\u003eQian K, Yuan L, Wang S, Kuang Y, Jin Q, Long D, et al. Inhibitor of apoptosis-stimulating p53 protein protects against inflammatory bowel disease in mice models by inhibiting the nuclear factor kappa B signaling. Clin Exp Immunol. 2021;205(2).\u003c/li\u003e\n\u003cli\u003eMcKay LI, Cidlowski JA. CBP (CREB Binding Protein) Integrates NF-\u0026kappa;B (Nuclear Factor-\u0026kappa;B) and Glucocorticoid Receptor Physical Interactions and Antagonism. Molecular Endocrinology. 2000;14(8).\u003c/li\u003e\n\u003cli\u003ePeng Q, Hua Y, Xu H, Chen X, Xu H, Wang L, et al. The NCOA1-CBP-NF-\u0026kappa;B transcriptional complex induces inflammation response and triggers endotoxin-induced myocardial dysfunction. Exp Cell Res. 2022;415(2).\u003c/li\u003e\n\u003cli\u003eAng DA, Carter JM, Deka K, Tan JHL, Zhou J, Chen Q, et al. Aberrant non-canonical NF-\u0026kappa;B signalling reprograms the epigenome landscape to drive oncogenic transcriptomes in multiple myeloma. Nature Communications 2024 15:1 [Internet]. 2024 Mar 21 [cited 2025 Feb 24];15(1):1\u0026ndash;20. Available from: https://www.nature.com/articles/s41467-024-46728-4\u003c/li\u003e\n\u003cli\u003eLi L, Song Q, Zhou J, Ji Q. Controllers of histone methylation-modifying enzymes in gastrointestinal cancers. Vol. 174, Biomedicine and Pharmacotherapy. 2024.\u003c/li\u003e\n\u003cli\u003eShi D, Pop MS, Kulikov R, Love IM, Kung A, Grossman SR. CBP and p300 are cytoplasmic E4 polyubiquitin ligases for p53. Proc Natl Acad Sci U S A [Internet]. 2009 Sep 22 [cited 2025 Jan 5];106(38):16275\u0026ndash;80. Available from: https://pubmed.ncbi.nlm.nih.gov/19805293/\u003c/li\u003e\n\u003cli\u003eOhshima K, Inoue K, Fujiwara A, Hatakeyama K, Kanto K, Watanabe Y, et al. Let-7 microRNA family is selectively secreted into the extracellular environment via exosomes in a metastatic gastric cancer cell line. PLoS One [Internet]. 2010 [cited 2025 Jan 5];5(10). Available from: https://pubmed.ncbi.nlm.nih.gov/20949044/\u003c/li\u003e\n\u003cli\u003eMacLellan SA, Lawson J, Baik J, Guillaud M, Poh CFY, Garnis C. Differential expression of miRNAs in the serum of patients with high-risk oral lesions. Cancer Med [Internet]. 2012 [cited 2025 Jan 5];1(2):268\u0026ndash;74. Available from: https://pubmed.ncbi.nlm.nih.gov/23342275/\u003c/li\u003e\n\u003cli\u003eZhang X, Zhang X, Liu C, Jia N, Li X, Xiao J. MiR‑224 promotes colorectal cancer cells proliferation via downregulation of P21WAF1/CIP1. Mol Med Rep [Internet]. 2014 Mar [cited 2025 Jan 5];9(3):941\u0026ndash;6. Available from: https://pubmed.ncbi.nlm.nih.gov/24430932/\u003c/li\u003e\n\u003cli\u003eGhosh S, Bose M, Ray A, Bhattacharyya SN. Polysome arrest restricts miRNA turnover by preventing exosomal export of miRNA in growth-retarded mammalian cells. Mol Biol Cell. 2015;26(6).\u003c/li\u003e\n\u003cli\u003eZdravkovic ND, Jovanovic IP, Radosavljevic GD, Arsenijevic AN, Zdravkovic ND, Mitrovic SL, et al. Potential dual immunomodulatory role of VEGF in ulcerative colitis and colorectal carcinoma. Int J Med Sci [Internet]. 2014 Jul 2 [cited 2025 May 26];11(9):936\u0026ndash;47. Available from: https://pubmed.ncbi.nlm.nih.gov/25076849/\u003c/li\u003e\n\u003cli\u003eHuang H, Wang W. Molecular mechanisms of glucocorticoid resistance. Eur J Clin Invest [Internet]. 2023 Feb 1 [cited 2025 Mar 3];53(2). Available from: https://pubmed.ncbi.nlm.nih.gov/36346177/\u003c/li\u003e\n\u003cli\u003eHu Y, He J, He L, Xu B, Wang Q. Expression and function of Smad7 in autoimmune and inflammatory diseases. J Mol Med [Internet]. 2021 Sep 1 [cited 2025 Mar 3];99(9):1209\u0026ndash;20. Available from: https://link.springer.com/article/10.1007/s00109-021-02083-1\u003c/li\u003e\n\u003cli\u003eAhmad A, Nawaz MI. Molecular mechanism of VEGF and its role in pathological angiogenesis. J Cell Biochem [Internet]. 2022 Dec 1 [cited 2025 Mar 3];123(12):1938\u0026ndash;65. Available from: https://onlinelibrary.wiley.com/doi/full/10.1002/jcb.30344\u003c/li\u003e\n\u003cli\u003eKenji Kira LAR. The Feature Selection Problem: Traditional Methods and a New Algorithm. [cited 2025 May 21]; Available from: https://aaai.org/papers/00129-aaai92-020-the-feature-selection-problem-traditional-methods-and-a-new-algorithm/\u003c/li\u003e\n\u003cli\u003ePattern Recognition - Sergios Theodoridis, Konstantinos Koutroumbas - Google Llibres [Internet]. [cited 2025 May 21]. Available from: https://books.google.es/books?hl=ca\u0026amp;lr=\u0026amp;id=gAGRCmp8Sp8C\u0026amp;oi=fnd\u0026amp;pg=PP1\u0026amp;dq=Theodoridis,+S.+and+Koutroumbas,+K.+(1999)+Pattern+Recognition,+Academic+Press,+pp.+341%E2%80%93342\u0026amp;ots=pKR6YFPdZM\u0026amp;sig=sBwDek5UZ9R-CtCos9ZPs-3hExI\u0026amp;redir_esc=y#v=onepage\u0026amp;q\u0026amp;f=false\u003c/li\u003e\n\u003cli\u003eFay MP, Proschan MA. Wilcoxon-Mann-Whitney or t-test? On assumptions for hypothesis tests and multiple interpretations of decision rules. Stat Surv [Internet]. 2010 [cited 2025 May 21];4:1. Available from: https://pmc.ncbi.nlm.nih.gov/articles/PMC2857732/\u003c/li\u003e\n\u003cli\u003eHanley JA, McNeil BJ. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology [Internet]. 1983 Sep 1 [cited 2025 May 21];148(3):839\u0026ndash;43. Available from: /doi/pdf/10.1148/radiology.148.3.6878708\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Tables","content":"\u003cp\u003e\u003cstrong\u003eTable 1.\u003c/strong\u003e List of different approaches used to identify potential sets of biomarkers from miR sequencing data.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 236px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eApproaches\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 330px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eReference\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 236px;\"\u003e\n \u003cp\u003eRELIEF feature selection method\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 330px;\"\u003e\n \u003cp\u003eKira, Kenji and Rendell, Larry (1992). \u003cem\u003eThe Feature Selection Problem: Traditional Methods and a New Algorithm\u003c/em\u003e. AAAI-92 Proceedings (40)\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 236px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eENTROPY+CORRELATION feature selection method (a\u003c/strong\u003elso known as Kullback-Liebler distance or divergence)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 330px;\"\u003e\n \u003cp\u003eTheodoridis, S. and Koutroumbas, K. (1999) \u003cem\u003ePattern Recognition\u003c/em\u003e, Academic Press, pp. 341\u0026ndash;342 (41)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 236px;\"\u003e\n \u003cp\u003eWILCOXON feature selection method\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 330px;\"\u003e\n \u003cp\u003eFay, Michael P.; Proschan, Michael A. (2010). \u003cem\u003e\u0026quot;Wilcoxon\u0026ndash;Mann\u0026ndash;Whitney or t-test? On assumptions for hypothesis tests and multiple interpretations of decision rules\u0026quot;\u003c/em\u003e. Statistics Surveys 4: 1\u0026ndash;39 \u0026nbsp;(42)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 236px;\"\u003e\n \u003cp\u003eSIMPLE REGRESSION feature selection method\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 330px;\"\u003e\n \u003cp\u003e\u003cem\u003eAnaxomics method.\u003c/em\u003e Stands for a feed forward selection method. Initially both feature combinations are evaluated using a logistic regression. Then, the feature sets are grown (with logistic regression), checking all the combinations obtained when one feature is added to the pool of previously identified best combinations.\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd style=\"width: 236px;\"\u003e\n \u003cp\u003eROC feature selection method\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 330px;\"\u003e\n \u003cp\u003eHanley, James A.; McNeil, Barbara J. (1983). \u003cem\u003eA method of comparing the areas under receiver operating characteristic curves derived from the same cases.\u003c/em\u003e Radiology 148 (3): 839\u0026ndash;843. (43)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eROC; receiver operating characteristic.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 2.\u003c/strong\u003e Summary of the best potential protein-coding transcript biomarker combination for each time point. Upregulated and downregulated proteins in R (vs NR) are represented by up green and down orange arrows, respectively.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eDay\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eUniProt Unique\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eidentifier name\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eUniProt code\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eGeneralization capability\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eAccuracy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eBaseline\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNCOR1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eQ75376\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e82.35%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e100%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNCOA3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eQ9Y6Q9\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCBP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eQ92793\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e\u003cstrong\u003eDay 3\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNRIP1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eP48552\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e80.95%\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\"\u003e\n \u003cp\u003e100%\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eNCOR1\u0026nbsp;⬇\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eQ75376\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eCBP\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\"\u003e\n \u003cp\u003eQ92793\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eNR, non-responders; R, responders.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 3.\u003c/strong\u003e Proteins identified in basal and/or day three analysis related to both UC and CS. Upregulated and downregulated proteins in R (vs NR) are represented by up green and down orange arrows, respectively.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"633\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eProtein name\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eUniprot code\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eBaseline\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eDay 3\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eShared or unique\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eFunctional role\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eReferences\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eCBP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQ92793\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eShared\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCoactivator of GCR and NF-\u0026kappa;B; modulates transcription and angiogenesis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 23154639, PMID: 23160045\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eCEBPB\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP17676\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔ ⬆\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eShared\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRegulates inflammatory gene expression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 19958093, PMID: 25203139\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eHDAC1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQ13547\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔ ⬆\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eShared\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eChromatin remodeling, transcription regulation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 22517765, PMID: 25203139\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eCREB1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP16220\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eShared\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eStress and inflammatory response regulation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 23154639, PMID: 18762182\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eNFKB1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP19838\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eShared\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCentral transcription factor regulating immune and inflammatory pathways\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 23322997, PMID: 25133403, PMID: 20407229\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eETS1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP14921\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eShared\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTranscription factor influencing cell proliferation and differentiation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 25926267, PMID: 12145332\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eIKKB\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eO14920\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eActivator of NF-\u0026kappa;B signaling; regulates inflammation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 15753535, PMID: 24347992\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eCALR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP27797\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔ ⬆\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCalcium-binding chaperone involved in protein folding\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 16995349, PMID: 9201693\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eFLNA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP21333\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCytoskeletal protein regulating cell structure and signaling\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 22198550, PMID: 16639003\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eE2F4\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQ16254\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eTranscription factor controlling cell cycle progression\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 21247883, PMID: 10867026\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eNCOA2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQ15596\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCoactivator of GCR; enhances transcriptional activity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 19198856, PMID: 15207724\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003ePGFRB\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP09619\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCell proliferation and angiogenesis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 21047522, PMID: 24228109\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eBCR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP11274\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSignaling in cell growth and immune response\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 23825635, PMID: 23264597\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eNDUS1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP28331\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRole in mitochondrial respiratory chain\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 20440543, PMID: 23670350\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eLIF\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP15018\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Day 3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCytokine involved in inflammation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 9889404, PMID: 15985451\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAT2A2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP16615\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCalcium signaling regulation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 17131044, PMID: 19726057\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eKLF5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQ13887\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eRegulates cell proliferation and inflammation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 17262812, PMID: 14634001\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eIL32\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP24001\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔ ⬇\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘ ⬇\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eCytokine involved in inflammation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 17590175, PMID: 24791863, PMID: 22646473\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eGCR\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP04150\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eGlucocorticoid receptor regulating CS action\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 19646928, PMID: 24084075\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eAP2A1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eO95782\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eVesicular transport regulation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 20376207, PMID: 9421462\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eSTA5A\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP42229\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eSignal transduction and transcription\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 22019623, PMID: 16804404\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003ePO2F1\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eP14859\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eDNA binding and transcriptional regulation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 22479607, PMID: 12807698, PMID: 9722596\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eCHIP\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQ9UNE7\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eProtein quality control, ubiquitination\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 25258038, PMID: 20661446\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eATG5\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eQ9H1Y0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAutophagy regulation\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 25642769, PMID: 24897381\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd\u003e\n \u003cp\u003e\u003cstrong\u003eANGP2\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eO15123\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✔\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003e✘\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eUnique to Basal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003eAngiogenesis\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd\u003e\n \u003cp\u003ePMID: 25759532, PMID: 18037159\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eCS, corticosteroid; GCR, glucocorticoid receptor; NR, non-responders; R, responders.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable 4.\u003c/strong\u003e Summary of the best potential miRs biomarker combination per visit and sample.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"576\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDay/\u003c/strong\u003e\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003eSample\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAll miRs\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 246px;\"\u003e\n \u003cp\u003e\u003cstrong\u003emiRs related to MoA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd style=\"width: 0px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBaseline/ Plasma\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003eCombination 1\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e ENTROPY + CORRELATION \u0026nbsp; \u0026nbsp; \u0026nbsp;\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 78.95%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-218-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-6754-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-4767\u003c/li\u003e\n \u003cli\u003ehsa-miR-939-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-548s\u003c/li\u003e\n \u003cli\u003ehsa-miR-34c-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 246px;\"\u003e\n \u003cp\u003eCombination 2\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e SIMPLE REGRESSION\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 73.68%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-224-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-324-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-10a-5p\u003c/li\u003e\n \u003cli\u003ehsa-let-7d-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDay 3/ Plasma\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003eCombination 1\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e SIMPLE REGRESSION\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 70.00%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-181A-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-548at-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 246px;\"\u003e\n \u003cp\u003eCombination 2\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e SIMPLE REGRESSION\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 70.00%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-181a-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-433-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-20a-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eBaseline/ Biopsy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003eCombination 1\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e ROC\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 93.33%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-769-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-6516-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-625-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 246px;\"\u003e\n \u003cp\u003eCombination 2\u0026nbsp;\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e ROC\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 93.33%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-10B-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-449a\u003c/li\u003e\n \u003cli\u003ehsa-miR-494-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-224-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-339-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDay 3/ Biopsy\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 255px;\"\u003e\n \u003cp\u003eCombination 2\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e RELIEF\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 100%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-140-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-431-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-25-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-151a-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-7706\u003c/li\u003e\n \u003cli\u003ehsa-miR-299-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-362-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd rowspan=\"3\" valign=\"top\" style=\"width: 246px;\"\u003e\n \u003cp\u003eCombination 2\u003c/p\u003e\n \u003cp\u003e\u003cstrong\u003e\u003cu\u003eFeature Selection Method:\u003c/u\u003e\u003c/strong\u003e RELIEF\u003c/p\u003e\n \u003cp\u003eGeneralization capability: 100%\u003c/p\u003e\n \u003cp\u003eAccuracy: 100%\u003c/p\u003e\n \u003cp\u003eSelected features:\u003c/p\u003e\n \u003cul\u003e\n \u003cli\u003ehsa-miR-140-5p\u003c/li\u003e\n \u003cli\u003ehsa-miR-25-3p\u003c/li\u003e\n \u003cli\u003ehsa-miR-130b-5p\u003c/li\u003e\n \u003c/ul\u003e\n \u003c/td\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd height=\"45\" style=\"width: 0px;\"\u003e\u003cbr\u003e\u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003emiR, microRNA; MoA, mechanism of action; ROC, receiver operating characteristic.\u003c/p\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":"Ulcerative colitis, TNBS-colitis, Biomarkers, Steroid-refractoriness, miR, Machine Learning, Systems Biology","lastPublishedDoi":"10.21203/rs.3.rs-7401947/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-7401947/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground and Aims:\u003c/strong\u003e Approximately 60% of patients with ulcerative colitis (UC) exhibit steroid resistance or dependence, highlighting the need for reliable biomarkers to predict therapeutic response. This study employed a systems biology and machine learning approach to integrate microRNA (miR) and mRNA expression data from both rectal biopsies and plasma samples from UC patients undergoing corticosteroid therapy, aiming to uncover the molecular mechanisms underlying steroid refractoriness.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods:\u003c/strong\u003e Whole-transcriptome and miR profiling were performed at baseline and after three days of corticosteroid therapy. Corticosteroid-treated UC patients were classified as responders (R) or non-responders (NR) after seven days of treatment. Mathematical modelling and protein-miR interaction mapping were used to identify mechanistically relevant biomarker candidates. Selected findings were validated in a TNBS-induced colitis mouse model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults:\u003c/strong\u003e Key transcriptional co-regulators such as NCOA3, CBP, NCOR1, and NRIP1 were differentially expressed between R and NR, influencing glucocorticoid receptor (GCR) signalling. Multiple miRNAs, including miR-145-5p, miR-10b-5p, and miR-16-5p, were identified as potential biomarkers and regulators of inflammatory and GCR-related pathways. The cross-correlation between plasma and tissue miRs revealed consistent molecular patterns, some of which were also conserved in the murine model, supporting the existence of cross-species steroid response mechanisms.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e This integrative multi-omic approach provides new insights into molecular steroid-refractoriness in UC and offers a promising framework for developing predictive tools and advancing personalised therapeutic strategies in inflammatory bowel disease.\u003c/p\u003e","manuscriptTitle":"Systems biology-driven miR-mRNA integration identifies potential steroid- refractoriness biomarkers in ulcerative colitis and reveals a conserved mechanism across species","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-09-17 05:11:35","doi":"10.21203/rs.3.rs-7401947/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":"3babc300-7563-488a-9545-a08f7fe75cf6","owner":[],"postedDate":"September 17th, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":54674871,"name":"Health sciences/Biomarkers"},{"id":54674872,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":54674873,"name":"Health sciences/Diseases"},{"id":54674874,"name":"Health sciences/Gastroenterology"},{"id":54674875,"name":"Biological sciences/Immunology"}],"tags":[],"updatedAt":"2025-10-22T13:38:21+00:00","versionOfRecord":[],"versionCreatedAt":"2025-09-17 05:11:35","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-7401947","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-7401947","identity":"rs-7401947","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}