Utility of plasma MicroRNA profiling as diagnostic biomarker in immune system activation and inflammation and early predictor of severity in patients with COVID-19 Authors: Martina Schiavello, Barbara Vizio, Tiziana Sanavia, Ornella Bosco, Chiara Dini, Paolo Cagna Vallino, Emanuele Pivetta, Fulvio Morello, Piero Fariselli, Giuseppe Montrucchio, Enrico Lupia Published: 2025 ABSTRACT AbstractThe clinical course of Coronavirus disease 2019 (COVID-19) ranges from mild symptoms to severe complications, including respiratory failure and thromboembolic events. MicroRNAs (miRNAs) are small non-coding RNAs involved in gene regulation and may serve as biomarkers. This study aimed to identify plasma miRNAs that could serve as biomarkers for diagnosing immune system disorders and inflammation, and for prognostic stratification of patients with COVID-19. We enrolled 40 patients with suspected COVID-19 at Emergency Room admission; infection was confirmed in 30 of them. Ten non-COVID-19 patients and 10 healthy subjects were included for comparison. Among the COVID-19 group, 26 hospitalized patients were followed and stratified by disease severity (good vs. poor prognosis). Plasma miRNA profiling and single-tube validation were performed using qRT-PCR, supported by in silico miRNA-mRNA prediction and pathway enrichment analysis. Expressions of miR-199a-5p, miR-142-3p, miR-133a-3p and miR-545-3p were higher in COVID-19 patients vs. healthy subjects. Moreover, miR-133a-3p and miR-545-3p were increased in COVID-19 vs. non-COVID-19 patients. miR-423-3p, miR-106b-5p, miR-142-3p and miR-369-3p were significantly elevated in patients who later developed respiratory failure. Bioinformatics analysis suggested that these miRNAs are involved in immune and inflammation pathways. Plasma miRNA profiling may be a promising non-invasive tool for COVID-19 diagnosis and prediction of severity.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-025-29736-2.Keywords: COVID-19, MiRNA profiling, Biomarkers, BioinformaticsSubject terms: Biomarkers, Computational biology and bioinformatics, Diseases, Genetics, Immunology, Medical research, Molecular biology Introduction Coronavirus disease 2019 (COVID-19) is still a threatening global disease1, with over 700 million confirmed cases and almost 7 million deaths worldwide2. The clinical course of COVID-19 may profoundly vary from mild to lethal forms according to the development of respiratory failure, target-organ damage, and thromboembolic events3–6. Predicting deterioration in vital status in COVID-19 patients, especially in the early phases of the disease, is a critical task in decision-making. The detection of early disease manifestations is currently challenging due to variable disease courses, leading to a long time to diagnose such patients, with different responses to the same treatment7,8. In recent years, the search for alternative clinical indicators, including those from emerging “omics” technologies, has provided new insights into the biological aspects of the disease, helping COVID-19 patient management9,10. MicroRNAs (miRNAs) are short non-coding RNAs involved in the post-transcriptional regulation of gene expression through mRNA silencing11. The modulation of gene expression by miRNAs plays a significant role in many human diseases, including cancer12, cardiovascular diseases13, respiratory infections14, and sepsis15. miRNAs are secreted into extracellular fluids and delivered to other target cells, where they can also act as chemical messengers to mediate cell-cell communication16. Cell- and tissue-specific miRNA expression and circulating miRNA signatures in various body fluids, including serum and plasma, are currently being evaluated as valuable diagnostic tools for specific diseases and for patient risk stratification16. Thus, the potential application of miRNAs as noninvasive biomarkers for the diagnosis and early prediction of poor prognosis in COVID-19 is a highly relevant research topic. Previous studies documented an altered plasma miRNA profile in COVID-19 patients17,18. For instance, Farr et al. profiled circulating miRNAs from 10 COVID-19 patients sampled longitudinally and 10 age and gender matched healthy donors, identifying a three-miRNA signature (miR-423-5p, miR-23a-3p and miR-195-5p) as diagnostic of COVID-19 with an accuracy of 99.9%18, which completely reverted at the levels found in healthy subjects after patient recovery17. Zeng et al. conducted a comprehensive analysis of miRNA expression in patients with different clinical presentations of COVID-19, showing that some miRNAs, such as miR-370, miR-1246 and miR-483, are associated with COVID-19 diagnosis, while miR-625 and miR-143 are associated with disease severity19. MiRNA sequencing from the plasma of 11 healthy subjects and from 18 non-severe and 18 severe COVID-19 patients revealed miR-150 and miR-21 associated to the diagnosis of COVID-19 disease, and miR-122 and miR-133 to COVID-19 severity20. In addition, Madè et al. identified miR-144-3p as one of the most significant miRNAs regarding COVID-19 disease severity21. In our study, we investigated the expression of circulating miRNAs in patients with COVID-19 at the time of diagnosis in the Emergency Room (ER) and their potential utility as biomarkers for the diagnosis of COVID-19. In addition, we aimed to identify miRNAs that were early markers of a higher risk of developing respiratory failure. Our results identified several miRNAs that may serve as promising diagnostic biomarkers and early predictors of severity in COVID-19 patients. Materials and methods Patient enrollment and study design A prospective observational cohort study was carried out on 40 patients admitted to the ER of the “Città della Salute e della Scienza di Torino – Molinette Site” University Hospital, Turin, Italy, from February to May 2021. We considered all adult patients (age > 18 years) who, at triage, were screened positive for acute symptoms commonly associated with severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) infection (i.e. fever, dyspnea, cough, sore throat, diarrhea, ageusia, anosmia). Exclusion criteria were: death or orotracheal intubation (OTI) in the first 24 h following ER admission, and malignancies in active treatment. For all patients, SARS-CoV-2 detection was performed through RT-PCR on nasopharyngeal swab samples collected at the time of ER admission. In addition, COVID-19 was always considered present in patients with a positive qRT-PCR test result obtained within 5 days of ER presentation. Final case adjudication of COVID-19 presence or absence (alternative diagnosis) was performed by two expert physicians who independently reviewed each patient’s medical records, including clinical presentation, laboratory findings, and imaging tests. Cases were classified as COVID-19 present or COVID-19 absent (non-COVID-19, alternative diagnosis) based on this assessment. In the other patients, the final diagnosis was established considering all follow-up data. In the event of discordant adjudication between the two experts, a third expert provided the final diagnosis. Patients were further classified based on COVID-19 severity observed during hospitalization as follows: (i) good prognosis group: patients who presented moderate symptoms at enrollment and did not require oxygen supplementation at any point during hospitalization; (ii) poor prognosis group: patients who presented with moderate symptoms at enrollment but subsequently developed respiratory failure, defined as the need for increasing oxygen supplementation, starting continuous positive airway pressure (CPAP) or non-invasive ventilation (NIV), or proceeding to IOT/admission to the intensive care unit (ICU). At the time of initial evaluation in the ER, patients in both groups had comparable clinical severity, including similar oxygen saturation (SpO2), respiratory rate and absence of organ dysfunction or respiratory failure. Therefore, the classification into prognostic groups was based on the subsequent development of respiratory failure, defined as the need for increased oxygen supplementation or respiratory support during hospitalization. Of note, none of these patients presented respiratory failure or organ damage at the time of the first medical evaluation in the ER, when patients in both groups had comparable symptoms and clinical severity. The full patient selection and classification process is illustrated in Fig. 1. Patient enrollment flowchart. The present study was conducted in compliance with the World Medical Association Declaration of Helsinki. All patients provided oral informed consent (as specifically authorized by the Ethical Committee during the COVID-19 pandemic due to infection risk management in ER) to participate in the study, and all data were immediately de-identified. The study was approved by the Institutional Ethical Committee (A.O.U. Città della Salute e della Scienza di Torino – A.O. Ordine Mauriziano – A.S.L. Città di Torino, number of approval CS2/139). Finally, 10 healthy subjects (HS) were recruited from the laboratory staff and used as an additional control group. The experimental process was separated into three phases: (i) screening; (ii) data analysis, and (iii) validation (Fig. 2). Study flowchart. Blood collection and plasma Preparation Peripheral blood samples were collected by clean venipuncture using a 21-gauge infusion set in EDTA-vacutainers and, after discharging the first mL, centrifuged for 10 min at 1,600 × g at 4 °C within 1 h of collection. The plasma obtained was further centrifuged at 13,300 × g for 10 min at 4 °C, and stored at -80 °C until use. Freeze-thaw cycles were avoided. MiRNA isolation MiRNAs were extracted from 200 µL of plasma by using the miRNeasy Serum/Plasma Advanced Kit (Qiagen, Hilden, Germany), as described by the manufacturer’s datasheet. A spike-in exogenous control, Cel-miR-39-3p (478293_mir TaqMan Advanced Assay, Thermo Fisher Scientific, MA, USA) from Caenorhabditis elegans, lacking sequence homology to human miRNAs, was added to each sample as an external reference miRNA (4.8 × 108 copies/sample). After silica-membrane-based purification on RNeasy UCP MinElute spin column, total RNAs, primarily miRNAs and other small non-coding RNAs, were eluted into 20 µL of RNase-free water and stored at -80 °C. RNA concentration and purity were assessed using a Nanodrop ND-1000 (Thermo Fisher Scientific). The 260/280 and 260/230 ratios indicated the absence of protein or organic contamination. Although RNA integrity was not assessed with a Bionalyzer, RNA quality was indirectly confirmed by the successful and consistent amplification of all tested miRNAs in downstream qRT-PCR reactions. Only samples yielding robust and reproducible amplification curves were included in the analysis. cDNA synthesis and pre-amplification An equal amount (2 µL) of each RNA extract was processed using the TaqMan Advanced miRNA cDNA synthesis kit (Applied Biosystems, Thermo Fisher Scientific), according to the manufacturer’s instructions. Briefly, the cDNA templates were obtained through the following steps: poly(A) tailing, adapter ligation, reverse transcription (RT), and pre-amplification. The converted cDNAs were then stored at -20 °C. MiRNA relative quantification by TaqMan MiRNA array For the diagnostic and prognostic screening phase, four pooled samples composed of one pool obtained from HS (n = 8) and three pools obtained from COVID-19 patients, each one from 8 patients, were processed by using TaqMan™ Advanced miRNA Human Cards A and B (Applied Biosystems, Thermo Fisher Scientific), which allowed the evaluation of the expression of 754 miRNAs (Supplementary Fig. 1). Specifically, pooled cDNA was dispensed into each of the 8 loading ports per card. Plates were then centrifuged twice at 330 × g (1 min at room temperature) and sealed to ensure a uniform distribution across the plate. qRT-PCR reaction was performed under the following cycling conditions: 92 °C (10 min), 40 cycles of 95 °C (1 s) and 60 °C (20 s) by using a QuantStudio 12 K Flex Real-Time PCR system (Applied Biosystems, Thermo Fisher Scientific). Endogenous and exogenous miRNA controls for data normalization were included in the array. Data were analyzed by QuantStudio Design and Analysis Software v1.4.1 (Applied Biosystems, Thermo Fisher Scientific). Data preprocessing To ensure optimal data quality, we first analyzed spike-in miRNA templates to assess the uniformity of the RNA extraction procedure and the efficiency of qRT-PCR. miRNAs with threshold cycles (Ct) > 33 were considered undetectable and excluded from the analysis. Relative quantification (Rq) was then performed using the 2–ΔΔCt method. Normalization was carried out using the exogenous spike-in control Cel-miR-39, added to all samples before RNA extraction. Endogenous reference miRNAs were not used, in line with previous studies on circulating miRNAs that highlight the lack of stable endogenous controls in plasma and serum, particularly in inflammatory or pathological conditions such as COVID-1922,23. Bioinformatics analysis A consensus analysis was applied to associate potential gene targets to each miRNA using R package multiMiR24. Specifically, the package provides both experimentally validated miRNA-mRNA targets from miRecords25, miRTarBase26 and TarBase27 and predicted targets from Diana28, ElMMo29, MicroCosm30, Miranda31, MirDB32, Pictar33, Pita34, and Targetscan35. We considered those target genes that were either annotated as experimentally validated in at least one database or predicted by at least three tools. We then functionally characterized the lists of selected miRNAs through a pathway enrichment analysis, collecting the annotations of the gene targets from 332 and 610 pathways available in Kyoto Encyclopedia of Genes and Genomes (KEGG)36 and WikiPathways37 databases, respectively. Based on this consensus analysis , each gene-to-pathway annotation was converted into the corresponding miRNA-to-pathway association. Fisher’s Exact test was then applied to each pathway, considering the number of annotated miRNAs and adjusting the p-values for multiple testing using the False Discovery Rate (FDR). Experimental validation of selected MiRNAs by single-tube MiRNA assay Validation of selected miRNAs was carried out by qRT-PCR on single-tube assays using the QuantStudio 1 Real-Time PCR system (Applied Biosystems, Thermo Fisher Scientific) for the following probes: hsa-miR-199a-5p (478231_mir), hsa-miR-200a-3p (478490_mir), hsa-miR-100-5p (478224_mir), hsa-miR-133a-3p (478511_mir), hsa-miR-28-5p (478000_mir), hsa-miR-142-3p (477910_mir), hsa-miR-545-3p (479002_mir); has-miR-495-3p (478136_mir); hsa-miR-221-5p (478778_mir); hsa-miR-376c-3p (478459_mir); hsa-miR-627-5p (478427_mir); hsa-miR-374b-5p (478389_mir); hsa-miR-154-3p (478725_mir); hsa-miR-423-3p (478327_mir); hsa-miR-106b-5p (478412_mir); hsa-miR-369-3p (478067_mir); hsa-miR-151b (477811_mir); miR-374a-5p (478238_mir); hsa-miR-103a-2-5p (477864_mir) and cel-miR-39-3p (478293_mir) (Applied Biosystems, Thermo Fisher Scientific). For all analyzed miRNAs, we performed qRT-PCR with three technical replicates through the following steps: 95 °C (10 min), followed by 40 cycles at 95 °C (15 s) and 60 °C (1 min). Sample-size Estimation The study was designed as a preliminary investigation. The required sample size was estimated using a one-sample test for proportions. Assuming a hypothesized population proportion of 0.50 (the null effect), a true proportion of 0.75, a one-sided α of 0.01, and a power of 0.80, a minimun of 38 subjects was required. The sample size calculation was performed using Stata 19 (StataCorp, College Station, TX, USA). Statistical analysis Descriptive statistics were used to summarize the characteristics of the study population. All data are reported as median (range) or mean ± standard error (SEM), according to data distribution, evaluated through the Shapiro-Wilk test. All other statistical analyses were performed by using Fisher’s Exact Test for categorical variables, One-Way ANOVA followed by Tukey multiple comparison test and Student’s t-test for normally distributed data, or by Mann-Whitney test for skewed data. Area under the receiver-operating-characteristic curve (AUC -ROC) and Likelihood ratio were applied to test miRNA’s accuracy as biomarkers. We used Youden’s index to estimate the optimal cut point38. The results were considered statistically significant for p-values < 0.05. All statistical analyses were performed using GraphPad Prism 8 software (GraphPad Software, CA, USA). Results Patient characteristics Forty patients with symptoms compatible with COVID-19 were included in the study. Of these, 30 patients were diagnosed with COVID-19 (COVID-19 patients), whereas the diagnosis of COVID-19 was excluded in the remaining 10 patients (classified as non-COVID-19 patients). Ten healthy subjects (HS) were also recruited and used as an additional control group. Demographic and clinical characteristics of the study subjects are reported in Tables 1 and 2. Briefly, the mean age was 70.4 (± 8.9) years for HS, 58.67 (± 11.2) years for non-COVID-19 patients and 69.1 (± 12.4) years for COVID-19 patients. The proportion of males/females was 6/4 for HS, 2/8 for non-COVID-19 patients and 17/13 for COVID-19 patients. Table 1Characteristics of patients at the moment of first medical evaluation in the ED.Healthy Subjects(n = 10)non-COVID-19(n = 10)COVID-19(n = 30)Gender (M/F), N (%)6/4 (60/40)2/8 (20.0/80.0)17/13 (56.7/43.3)Age, mean (± SD), yr70.4 (± 8.9)58.67 (± 11.2)69.1 (± 12.4)Symptoms, N (%)Fever-8 (80.0)27 (90.0)Cough-6 (60.0)19 (63.3)Shortness of breath-5 (50.0)22 (73.3)Sore throat-00Diarrhea-3 (30.0)11 (36.6)Ageusia-1 (10.0)7 (23.3)Anosmia-03 (10.0)Comorbidities, N (%)Hypertension-1 (10.0)13 (43.3)Diabetes Mellitus-05 (16.7)CAD-01 (3.0)Atrial fibrillation/flutter-1 (10.0)4 (13.3)Asthma/COPD-00Cancer-00Cerebrovascular disease-02 (6.7)CKD-01 (3.0)DVT/PE-00Home discharge, N (%)NA10 (100)0Ward admission, N (%)NA030 (100)ICU/HDU Admission, N (%)NA08 (26.7)NIV, N (%)NA016 (53.3)IMV, N (%)NA08 (26.7)In-hospital Mortality, N (%)NA04 (13.3)Day of hospitalization, mean (± SD)NA013.0 (± 9.4)ED, Emergency Department; CAD, Coronary Artery Disease; COPD, Chronic Obstructive Pulmonary Disease; CKD, Chronic Kidney Disease; DVT, Deep Vein Thrombosis; PE, Pulmonary Embolism; ICU, Intensive Care Unit; HDU, High Dependency Unit; NIV, Non-invasive Ventilation; IMV, Invasive Mechanical Ventilation; NA, not applicable. Characteristics of patients at the moment of first medical evaluation in the ED. ED, Emergency Department; CAD, Coronary Artery Disease; COPD, Chronic Obstructive Pulmonary Disease; CKD, Chronic Kidney Disease; DVT, Deep Vein Thrombosis; PE, Pulmonary Embolism; ICU, Intensive Care Unit; HDU, High Dependency Unit; NIV, Non-invasive Ventilation; IMV, Invasive Mechanical Ventilation; NA, not applicable. Table 2Laboratory findings.Healthy Subjects(n = 10)Non-COVID-19(n = 10)COVID-19(n = 30)White blood cell count (109/L)5.4 (± 0.9)7.2 (± 1.6)§7.6 (± 3.1)Neutrophil count (109/L)3.3 (± 0.8)4.6 (± 1.5)6.0 (± 3.1)§Lymphocyte count (109/L)1.6 (± 0.5)2.0 (1.5–2.4)0.9 (0.6–1.3)** §Neutrophil/Lymphocyte (ratio)2.3 (± 1.0)2.1 (1.9–3.3)5.2 (3.9–8.3)** §§§Leukocyte/Lymphocyte (ratio)3.7 (± 1.1)3.5 (3.3–4.8)6.7 (4.9–9.8)** §§§Platelet count (109/L)210.0 (± 75.3)263.6 (± 44.1)210.3 (± 72.8)*Mean platelet volume (fL)10.9 (± 0.74)10.4 (10.0-11.1)11.0 (10.1–11.6)Platelet distribution width (fL)13.4 (± 1.8)12.4 (± 1.5)12.7 (± 2.0)C-Reactive Protein (mg/L)1.9 (± 1.8)3.9 (0.4–15.0)76.9 (37.9-133.9)*** §§§P/F ratio--283.0 (± 62.3)Mean (± SD) or median (range) as appropriate. ***p < 0.0001; **p < 0.001; *p < 0.05 vs. Non-COVID-19. §§§ p < 0.0001; §§ p < 0.001; §p < 0.05 vs. HS. Laboratory findings. Mean (± SD) or median (range) as appropriate. ***p < 0.0001; **p < 0.001; *p < 0.05 vs. Non-COVID-19. §§§ p < 0.0001; §§ p < 0.001; §p < 0.05 vs. HS. Among non-COVID-19 patients, the final diagnosis was flu-like syndrome in 7 patients, bacterial pneumonia in 2, and acute gastroenteritis in 1 patient. COVID-19 patients had lower lymphocyte counts than non-COVID-19 patients and HS, whereas they showed increased neutrophil/lymphocyte ratio, leukocyte/lymphocyte ratio, and C-reactive protein values (Table 2). COVID-19 is associated with distinct plasma MiRNA signatures To identify specific miRNA signatures in plasma samples of COVID-19 patients, we performed a high-throughput screening of 754 miRNAs using TaqMan Human microRNA Arrays. Out of 754 miRNAs screened, 219 were detectable in the pooled plasma samples with Ct values < 33. Among these, we excluded: (i) 22 miRNAs since they were exclusively detected in COVID-19 patients, (ii) 17 miRNAs only present in plasma samples of HS, (iii) 39 miRNAs not expressed in any of the three COVID-19 plasma pools examined. Among the remaining 141 miRNAs, we finally considered only those with an absolute fold change (FC) of at least 33, resulting in 11 miRNAs differentially expressed in COVID-19 patients vs. HS, which were further analyzed through in silico analysis (Table 3). Table 3MiRNA fold change in plasma samples from COVID-19 patients vs. HS.miRNAFC hsa-miR-495-3p 33,3120 hsa-miR-221-5p 55,6321 hsa-miR-142-3p 65,3279 hsa-miR-545-3p 66,6032 hsa-miR-28-5p 72,7087 hsa-miR-133a-3p 129,1016 hsa-miR-100-5p 177,1509 hsa-miR-376c-3p 206,5762 hsa-miR-200a-3p 239,2815 hsa-miR-627-5p 312,8237 hsa-miR-199a-5p 1105,1478FC: fold change. MiRNA fold change in plasma samples from COVID-19 patients vs. HS. FC: fold change. To explore the pathways potentially associated to the selected miRNAs in COVID-19 patients, we performed an enrichment analysis using miRNA-pathway associations to the target genes, as described in the Methods section. The 11 miRNAs selected were found significantly enriched (FDR < 0.05) in 35 pathways from KEGG database (Supplemental Table S1) and in 5 pathways from WikiPathways database (Supplemental Table S2). Interestingly, from the list of the most significantly enriched pathways, we can infer five key biological mechanisms that might characterize the diagnosis of COVID-19 from the selected miRNAs: Immune System Inflammation: IL-17 signaling pathway and IL-6 signaling pathway highlight inflammation, cytokine storm, and excessive immune activation, which are hallmarks of severe COVID-1939, while chemokine signaling pathway is involved in immune cell recruitment and inflammation40; Immune System Inflammation: IL-17 signaling pathway and IL-6 signaling pathway highlight inflammation, cytokine storm, and excessive immune activation, which are hallmarks of severe COVID-1939, while chemokine signaling pathway is involved in immune cell recruitment and inflammation40; Immune System Activation: T-cell receptor (TCR) signaling pathway and B-cell receptor signaling pathway indicate an adaptive immune response, with T-cell activation playing a crucial role in viral defense41, while Natural killer (NK) cell-mediated cytotoxicity suggests activation of innate immunity, which is crucial in early viral response42; Immune System Activation: T-cell receptor (TCR) signaling pathway and B-cell receptor signaling pathway indicate an adaptive immune response, with T-cell activation playing a crucial role in viral defense41, while Natural killer (NK) cell-mediated cytotoxicity suggests activation of innate immunity, which is crucial in early viral response42; Cell Death & Tissue Damage: there is the activation of ferroptosis, a type of regulated cell death caused by oxidative stress, suggesting that COVID-19 might induce iron-dependent cell death, leading to lung and organ damage43; in addition, apoptosis-related network due to altered Notch3 suggests programmed cell death, which may contribute to tissue injury and immune dysregulation44. Cell Death & Tissue Damage: there is the activation of ferroptosis, a type of regulated cell death caused by oxidative stress, suggesting that COVID-19 might induce iron-dependent cell death, leading to lung and organ damage43; in addition, apoptosis-related network due to altered Notch3 suggests programmed cell death, which may contribute to tissue injury and immune dysregulation44. Metabolic & Stress-Related Pathways: mTOR signaling pathway and autophagy suggest disruptions in cellular metabolism and stress response, which are crucial for viral replication and immune regulation45, while HIF-1 signaling pathway is involved in oxygen regulation and could indicate hypoxia-related complications in COVID-19 patients46. Finally, the presence of the AGE-RAGE signaling pathway in diabetic complications suggests a link to metabolic disorders, which are known risk factors for severe COVID-1947; Metabolic & Stress-Related Pathways: mTOR signaling pathway and autophagy suggest disruptions in cellular metabolism and stress response, which are crucial for viral replication and immune regulation45, while HIF-1 signaling pathway is involved in oxygen regulation and could indicate hypoxia-related complications in COVID-19 patients46. Finally, the presence of the AGE-RAGE signaling pathway in diabetic complications suggests a link to metabolic disorders, which are known risk factors for severe COVID-1947; Cardiovascular & Systemic Effects: Cardiac Hypertrophic Response suggests potential cardiovascular complications linked to COVID-19, including heart inflammation and dysfunction48, while Gap junction and Adherens junction pathways indicate possible disruptions in cellular communication and tissue integrity49. Cardiovascular & Systemic Effects: Cardiac Hypertrophic Response suggests potential cardiovascular complications linked to COVID-19, including heart inflammation and dysfunction48, while Gap junction and Adherens junction pathways indicate possible disruptions in cellular communication and tissue integrity49. Beyond the specific KEGG pathway of Coronavirus disease selected by the enrichment analysis (FDR-adjusted p-value 0.016), we decided to focus our investigations on those significantly enriched pathways that are mostly involved with the immune system inflammation and the adaptive immune response: IL-6 signaling pathway from WikiPathways and Chemokine signaling pathway, IL-17 signaling pathway, B cell receptor signaling pathway and T cell receptor signaling pathway from KEGG database. Considering the potentially associated miRNAs, we focused on those that were annotated with at least five out of the six aforementioned pathways, restricting the list of candidate diagnostic miRNAs to the following 7 miRNAs: miR-100-5p, miR-133a-3p, miR-142-3p, miR-199a-5p, miR-200a-3p, miR-28-5p, and miR-545-3p. The bubble plot displayed in Fig. 3 shows, for each of the seven selected miRNAs, the annotated enriched pathways, the FDR-adjusted p-values resulting from the enrichment analysis (bubble color) and the number of genes that each miRNA targets in each pathway (bubble size). Bubble plot of the enriched pathways mainly associated with the 7 candidate diagnostic miRNAs. We next used qRT-PCR assay to confirm the differential expression of the seven candidate diagnostic miRNAs in plasma individual samples from 30 COVID-19 patients, 10 non-COVID-19 patients, and 10 HS. With this approach, four miRNAs, i.e. miR-199a-5p, miR-142-3p, miR-133a-3p and miR-545-3p, confirmed their differential expression in COVID-19 vs. HS (Fig. 4a-d), whereas other miRNAs did not show a significant difference between COVID-19 patients and HS (data not shown). In addition, miR-133a-3p and miR-545-3p were significantly higher in COVID-19 patients compared to non-COVID-19 patients. The expression of miR-199a-5p, miR-133a-3p and miR-545-3p did not differ significantly between non-COVID-19 patients and HS while miR-142-3p that was significantly higher in non-COVID-19 vs. HS (Fig. 4a-d). miRNA expression levels and ROC curves. Receiver operating characteristic (ROC) curves of miR-133a-3p [AUC = 0.857 (95% CI, 0.74–0.97), p = 0.0009] and miR-545-3p [AUC = 0.844 (95% CI, 0.69-1.00), p = 0.0030] showed a high diagnostic accuracy of these miRNAs in discriminating COVID-19 patients from non-COVID-19 patients, with a specificity of 100% and a sensitivity of 71.4% for miR-133a-3p, while a specificity of 100% and a sensitivity of 55.6% for miR-545-3p based on an optimal threshold value of 5.791 and 34.030, respectively (Fig. 4e). The combination of the two miRNAs achieved the highest diagnostic performance, AUC of 0.867 (95% CI, 0.724-1.000) (p = 0.0016), with a specificity of 70% and a sensitivity of 94.4% at the cut-off of 13.13 (Fig. 4f), outperforming either biomarker alone. This finding suggests that their combined use may improve diagnostic accuracy. miRNA profiling and integration of bioinformatics analysis reveal specific circulating miRNAs as potential early predictors of severity in COVID-19 patients. 26 of the 30 COVID-19 patients who did not present severe respiratory symptoms at the time of first clinical evaluation in the ER were subsequently separated into two groups based on the development of respiratory failure and the need for mechanical ventilation during the hospital stay. Based on this criterion, 13 COVID-19 patients had a poor prognosis and 13 a good prognosis (as detailed in the Methods section). Patients with a poor prognosis were significantly younger than patients with a good prognosis. No significant differences in other demographic and clinical characteristics were observed between COVID-19 patients with good and poor prognoses (Tables 4 and 5). Table 4Characteristics of patients at the moment of first medical evaluation in the ED.COVID-19Good Prognosis(n = 13)COVID-19Poor Prognosis(n = 13)Gender (M/F), N (%)6/7 (46.1/53.8)9/4 (69.2/38.8)Age, mean (± SD), yr75.2 (± 10.7)60.9 (± 9.6)°°Symptoms, N (%)Fever11 (84.6)13 (100)Cough7 (53.8)10 (76.9)Shortness of breath9 (69.2)11 (84.6)Sore throat00Diarrhea4 (30.8)6 (46.2)Ageusia2 (15.4)4 (30.8)Anosmia03 (23.1)Comorbidities, N (%)Hypertension6 (46.2)6 (46.2)Diabetes Mellitus2 (15.4)3 (23.8)CAD1 (7.7)0Atrial fibrillation/flutter3 (23.1)0Asthma/COPD00Cancer00Cerebrovascular disease1 (7.7)0CKD1 (7.7)0DVT/PE00 Home discharge (%) 00 Ward admission (%) 13 (100)13 (100) ICU/HDU Admission (%) 00 NIV (%) 5 (38.5)7 (53.8) MIV (%) 00 In-hospital Mortality (%) 02 (15.4) Day of hospitalization 14.0 (± 10.9)18.1 (± 11.0)CAD, Coronary Artery Disease; COPD, Chronic Obstructive Pulmonary Disease; CKD, Chronic Kidney Disease; DVT, Deep Vein Thrombosis; PE, Pulmonary Embolism; ED, Emergency Department; ICU, Intensive Care Unit; HDU, High Dependency Unit; MV, Mechanical Ventilation; NIV, Non-invasive Ventilation; IMV, Invasive Mechanical Ventilation; NA, not applicable. mean (± SD) or median (range) as appropriate. °°p < 0.001 vs. good prognosis. Characteristics of patients at the moment of first medical evaluation in the ED. CAD, Coronary Artery Disease; COPD, Chronic Obstructive Pulmonary Disease; CKD, Chronic Kidney Disease; DVT, Deep Vein Thrombosis; PE, Pulmonary Embolism; ED, Emergency Department; ICU, Intensive Care Unit; HDU, High Dependency Unit; MV, Mechanical Ventilation; NIV, Non-invasive Ventilation; IMV, Invasive Mechanical Ventilation; NA, not applicable. mean (± SD) or median (range) as appropriate. °°p < 0.001 vs. good prognosis. Table 5Laboratory findings.COVID-19Good prognosis(n = 13)COVID-19Poor prognosis(n = 13)White blood cell count (109/L)7.1 (4.6–9.5)6.8 (5.0-8.7)Neutrophil count (109/L)5.6 (± 2.3)5.6 (± 3.4)Lymphocyte count (109/L)1.0 (0.6–1.3)0.8 (0.6–1.4)Neutrophil/Lymphocyte (ratio)5.2 (4.2-9.0)5.1 (3.1–7.7)Leukocyte/Lymphocyte (ratio)6.5 (4.7–10.6)6.5 (4.6–9.3)Platelet count (109/L)191.0 (149.5–243.0)190.0 (159.0-249.5)Mean platelet volume (fL)10.7 (10.0-11.3)11 (9.9–11.9)Platelet distribution width (fL)12.4 (± 2.1)12.8 (± 2.1)C-Reactive Protein (mg/L)62.8 (± 37.1)87.2 (± 56.6)P/F ratio288.4 (± 45.7)294.7 (± 53.5)Mean (± SD) or median (range) as appropriate. Laboratory findings. Mean (± SD) or median (range) as appropriate. Out of 754 miRNAs screened, 221 were detectable in the pooled plasma samples using the same criteria detailed in the previous section. Of these, we excluded: (i) 13 miRNAs exclusively expressed in COVID-19 patients with poor prognosis, and (ii) 41 miRNAs exclusively expressed in those with good prognosis. After selecting only those miRNAs with an absolute fold change of at least 33, 12 miRNAs were considered for further analysis (Table 6), which were found in 26 and 15 enriched pathways from KEGG (Supplemental Table S3) and WikiPathways (Supplemental Table S4), respectively. The obtained enriched pathways primarily indicate dysregulation in immune response, metabolism, oxidative stress, and cardiovascular function, which are critical determinants of COVID-19 severity. Figure 5 displays a bubble plot of the enriched pathways, mostly involved in immune system inflammation and the adaptive immune response. In particular, for the adaptive immune response, also here we observed the enrichment of B cell receptor signaling pathway, suggesting that this process might be compromised in patients with a poor prognosis50. In terms of immune system inflammation, Toll-like receptor (TLR) signaling pathway with its regulation and Fc epsilon RI signaling pathway indicate hyperactivation of innate immunity, contributing to excessive inflammation in severe cases51,52; in addition, it is interesting to observe the pathway Neutrophil extracellular (NET) trap formation, which is associated with severe COVID-19 as excessive NETs contribute to inflammation, thrombosis, and lung damage53. Another intriguing pathway involved in the immune regulation is RANKL-RANK signaling pathway, which may play a role in inflammatory tissue damage54. For the evaluation of the impact of potential miRNAs in the severity of COVID-19 patients, we decided to focus on these pathways related to immune system response and inflammation. Table 6MiRNAs fold change in plasma samples from COVID-19 patients with poor vs. those with good prognosis.miRNAFC hsa-miR-374b-5p 0,0009 hsa-miR-100-5p 0,0042 hsa-miR-154-3p 0,0053 hsa-miR-423-3p 0,0055 hsa-miR-106b-5p 0,0086 hsa-miR-369-3p 0,0199 hsa-miR-151b 0,0232 hsa-miR-142-3p 0,0234 hsa-miR-103a-2-5p 0,0245 hsa-miR-374a-5p 50,5626 hsa-miR-627-5p 57,8803 hsa-miR-199a-5p 109,0616FC, fold change. MiRNAs fold change in plasma samples from COVID-19 patients with poor vs. those with good prognosis. FC, fold change. Bubble plot of the enriched pathways mainly associated with the 7 candidate prognostic miRNAs. Considering the potentially associated miRNAs, we focused on those that miRNAs were annotated with all these aforementioned pathways, identifying seven miRNAs as potential predictors of clinical worsening: miR-106b-5p, miR-142-3p, miR-151b, miR-199a-5p, miR-369-3p, miR-374a-5p, and miR-423-3p. Experimental validation through qRT-PCR assay showed that miR-423-3p (p = 0.047), miR-106-5p (p = 0.007), miR-142-3p (p = 0.017), and miR-369-3p (p = 0.038) resulted in lower expression in COVID-19 patients with poor prognoses than in those with good prognoses (Fig. 6a-d). miRNA expression levels and ROC curves. Using ROC analysis, these same miRNAs miR-423-3p [AUC = 0.752 (95% CI 0.536–0.968) p = 0.0452], miR-106-5p [AUC = 0.805 (95% CI, 0.629–0.981) p = 0.0083], miR-142-3p [AUC = 0.822 (95% CI 0.628-1.000) p = 0.0179], and miR-369-3p [AUC = 0.778 (95% CI 0.566–0.990) p = 0.0367] showed a good accuracy in discriminating COVID-19 patients based on their clinical prognosis (Fig. 6e). Setting the cut-off values at 7.911 for miR-142-3p and at 0.630 for miR-369-3p, these two miRNAs had a specificity of 100% and a sensitivity of 87.5%, whereas a cut-off of 9.823 for miR-423-3p resulted in a specificity of 81.82% and a sensitivity of 63.64. Finally, miR-106-5p achieved a specificity of 92.31% and a sensitivity of 61.54% with a cut-off value set at 3.583. The combination of miR-142-3p and miR-423-3p showed the highest AUC = 0.931, (95% CI 0.792-1.000) (p = 0.003), compared to all the other combinations (Fig. 6f; Table 7). In particular, a cut-off value of 20.22 allows the discrimination of COVID-19 patients with poor prognosis, achieving a specificity of 100% and a sensitivity of 87.50%. Table 7Area under the curve (AUC), 95% confidence interval and p-value calculated from combined MiRNA expression level.miRNAsAUC95% CIp-valuemiR142-3p + miR-369-3p + miR-423-3p + miR-106-5p0,9170,772-1,0000,004miR142-3p + miR-369-3p + miR-423-3p0,9030,738-1,0000,005miR142-3p + miR-423-3p + miR-106-5p0,9310,792-1,0000,003miR142-3p + miR-369-3p + miR-106-5p0,8670,695-1,0000,007miR-369-3p + miR-423-3p + miR-106-5p0,8880,723-1,0000,006miR142-3p + miR-369-3p0,8630,664-1,0000,010 miR142-3p + miR-423-3p 0,9310,792-1,0000,003miR142-3p + miR-106-5p0,8670,685-1,0000,007miR-369-3p + miR-423-3p0,8880,723-1,0000,006miR-369-3p + miR-106-5p0,8690,704-1,0000,006miR-423-3p + miR-106-5p0,7770,572-1,0000,028AUC: Area under the curve; CI: confidence interval. Area under the curve (AUC), 95% confidence interval and p-value calculated from combined MiRNA expression level. AUC: Area under the curve; CI: confidence interval. Discussion COVID-19 is an acute infectious respiratory disease caused by SARS-CoV-21. Patients with COVID-19 often worse rapidly, developing acute respiratory distress syndrome (ARDS) and septic shock, which are difficult to treat and heavily affect patient’s survival1. While most patients with severe respiratory disorders successfully recover, a substantial number of them die of respiratory failure and/or systemic complications55,56. Since the start of the pandemic, significant progress has been made in the treatment and outcome of patients with COVID-19. However, there is still room for improvement in early diagnosis and identifying patients at the highest risk of adverse outcomes. However, despite the decreased incidence of the disease, the identification of novel biomarkers predictive of severity and a deeper understanding of COVID-19 pathogenic mechanisms remain necessary. In particular, this may also be relevant to understanding the onset of the syndrome referred to as “long COVID-19”56. Several studies have focused on the possible use of miRNAs as biomarkers in diagnosing COVID-19, which have demonstrated good accuracy in distinguishing between COVID-19 patients and healthy controls57. In our study, based on bioinformatics analysis and limited to specific pathways identified through literature review, we aimed to identify a circulating miRNA signature associated with COVID-19 diagnosis. For this purpose, we evaluated differentially expressed miRNAs in COVID-19 patients compared to healthy subjects; in the validation phase, we compared them in COVID-19 patients vs. non-COVID-19 patients, thereby avoiding overestimating the performance of the potential novel biomarkers identified. We studied a cohort of COVID-19 patients enrolled at hospital admission, and, after a comprehensive bioinformatics analysis of miRNA arrays, selected seven candidate diagnostic miRNAs based on their high expression and association with pathways involved in the adaptive immune response and inflammation, processes that are key to COVID-19 pathophysiology. We then selected those miRNAs that mainly occur in these pathways and assessed them through qRT-PCR assay. The validation phase confirmed miR-199a-5p, miR-142-3p, miR-133a-3p and miR-545-3p as diagnostic biomarkers of COVID-19 patients vs. healthy subjects, but more interesting is the finding of the diagnostic power of miR-133a-3p and miR-545-3p in discriminating COVID-19 patients from non-COVID-19 patients; in particular, miR-133a-3p shows the best diagnostic accuracy, also if compared to the analysis of co-expression with miR-545-3p. Notably, the combined evaluation of miR-133a-3p and miR-545-3p yielded the highest diagnostic power, underscoring the complementary nature of these biomarkers. This represents the main strength and practical implication of our study, as it highlights the potential advantage of a combined miRNA signature over single-marker approaches. Our results are in line with and expand those of previous studies18,19,58,59 that investigated differentially expressed miRNAs and evaluated their clinical significance in COVID-19 patients. For instance, miR-199a-5p is known to be a key effector of TGF-β signaling in lung fibroblasts and it has been implicated in modulating inflammatory responses and fibrosis60. The up-regulation of miR-199a-5p has been found in COVID-19 patients admitted to ICU61 and has been suggested to predict and to participate in the development of fibrotic lesions in the lung of post-COVID-19 patients62. This supports a mechanistic link between miR-199a-5p and fibroproliferative remodeling, a critical pathological feature of severe and long-term COVID-19 sequelae. MiR-142-3p plays a key role in regulating immune responses, including T-cell differentiation and cytokine production. It has been demonstrated that this miRNA can influence the production of pro-inflammatory cytokines such as IL-6, which is elevated in severe COVID-19 cases, thereby potentially contributing to the cytokine storm and immune dysregulation observed in critical illness. Supporting our findings, Farr et colleague’s analysis of nasal swabs and plasma from COVID-19 patients revealed a panel of differentially expressed miRNAs, among which miR-142-3p was the most strongly upregulated and statistically different from control subjects63. In contrast, miR-545 had not previously been directly associated with SARS-CoV-2 infection. However, its involvement in regulating key biological processes such as cancer development64 and in epithelial-mesenchymal transition under hypoxic conditions65 suggests a broader role in cellular stress responses and tissue remodeling. These processes are also highly relevant in the context of lung injury and repair, which are critical aspects of COVID-19 pathogenesis. Moreover, Bertrams and collaborators66identified several miRNAs, including miR-545-3p able to distinguish between Acute Exacerbation of COPD and CAP, highlighting miR-545-3p potential role as a regulator of immune responses and inflammation in respiratory diseases. Although direct evidence linking miR-543-3p to COVID-19 remains lacking, its established role in related respiratory and inflammatory conditions support its candidacy as a biomarker reflecting immune dysregulation in COVID-19. In addition, this aligns with our finding that miR-545-3p, especially in combination with miR-133a-3p, enhances the diagnostic accuracy for COVID-19, possibly by capturing distinct but complementary pathological pathways. Interestingly, Molinero et al. analyzed the miRNA profile of bronchial aspirate samples from COVID-19 and non-COVID-19 patients and identified multiple miRNAs, including miR-133a-3p and miR-199a-5p, that can be proposed as markers to distinguish between the two groups67. Moreover, baseline levels of myocyte-derived miR-133a were associated with COVID-19 severity and 28-day ICU mortality19. Previous studies also identified increased levels of miR-133a related to inflammation and organ dysfunction in sepsis68, in atherosclerotic thrombotic cerebral infarction and in cardio-embolic stroke69. For instance, miR-133a-3p has been shown to modulate apoptotic pathways and NF-kB-mediated inflammatory signaling, which are central to tissue injury and systemic inflammation in COVID-19. Taken together, these evidences support our results and confirm that the increased expression of specific miRNAs reflects activation of inflammatory, apoptotic, and fibrotic pathways relevant to COVID-19 pathogenesis. The diagnostic power of these miRNAs likely stems from their activity as molecular sensors of host immune response and tissue damage, making them promising biomarkers for disease severity and progression. Identifying which patients are at the highest risk of experiencing adverse outcomes, whether through clinically recognized risk factors and/or biomarkers for severe illness, is undoubtedly crucial in clinical practice, as well as to optimize the intensive medical management of patients with COVID-19 or other critical conditions. Therefore, an additional and more ambitious goal of our study was to identify plasma miRNA profiles linked to the risk of developing respiratory failure and to assess their usefulness as clinical predictors of poor prognosis in COVID-19 patients. In SARS-CoV-2 infection, host protein translation is almost completely suppressed by the coronavirus in favor of viral proteins70 and, since miRNAs are the main regulators of gene expression, their expression alteration could characterize the patient’s clinical state and the severity of the disease71. Therefore, using the same approach developed for the diagnostic miRNA value, we selected seven miRNAs as potential biomarkers for clinical deterioration in COVID-19 patients. Among these, four miRNAs (miR-423-3p, miR-106b-5p, miR-142-3p, and miR-369-3p) passed the validation phase through qRT-PCR assay. The ROC curve analysis of individual and combined miRNAs showed that pairing miR-142-3p with miR-423-3p achieved the highest prognostic accuracy. Some evidence supporting our results originated from a previous study by Tang et al., which identified a panel of miRNAs, including miR-142‐3p, as potential biomarkers and contributors to the pathogenesis and severity of COVID‐19, as well as candidate therapeutic targets72. Notably, miR-142 is involved in the post-transcriptional negative regulation of IL-6, a cytokine centrally involved in the COVID-19 cytokine storm and hyperinflammatory state73. This regulatory role links miR-142-3p directly to the modulation of immune response intensity and inflammation, which are critical determinants of disease severity. Furthermore, altered expression of miR-142-3p has been observed in patients exhibiting rheumatological symptoms associated with long COVID-1974, suggesting its involvement in chronic inflammatory sequelae. Our data also agree with prior reports describing a downregulation of miR-106a expression in COVID-19 patients compared to controls. This decrease correlates with the increased production of pro-inflammatory cytokines, such as TNF-alpha, IL-1beta and IL-675, reinforcing the role of miR-106a in negative regulation of inflammatory pathways and highlighting its potential as a marker of dysregulated immune activation in COVID-19. Similarly, miR-423-3p has been observed to be upregulated in early COVID-19 patients18, as well as in conditions like heart failure76 and pulmonary tuberculosis77. Its recognized role as a master regulator of inflammatory processes75–79 suggests that miR-423-3p helps coordinate systemic inflammation and tissue responses during COVID-19. Analogously, miR-369-3p expression decreases in serum samples of COVID-19 patients requiring mechanical ventilation compared to those with milder disease80. Since miR-369-3p has been predicted to target the SARS-CoV-2 genome directly81, it may exert antiviral effects, and its reduction during disease worsening could reflect a loss of this protective mechanism, thus contributing to viral persistence and disease progression. Finally, although there are no direct studies linking miR-106b-5p to COVID-19 severity, Martínez-Fleta et al. identified a combination of four miRNAs, including miR-106-5p, that distinguished COVID-19 and community-acquired pneumonia through multivariate analysis82. This suggests that miR-106-5p could be involved in pathways that differentiate COVID-19 from other respiratory infections, potentially linked to different immune or inflammatory responses. Our study has certain limitations. First, although the study population may be considered representative of COVID-19 patients and prognosis, the number of patients enrolled is limited. However, it should be emphasized that patient selection was rigorous, as we excluded several pre-existing pathological conditions, such as immunodeficiency, active cancers, and cardiovascular diseases, to minimize confounding factors. Therefore, further validation of our results in larger and international cohorts is encouraged and necessary. Second, we evaluated the biological and molecular functions of the identified miRNAs solely through in silico bioinformatics analysis on publicly available databases. Although we were able to suggest potential target genes and biological processes involved, it remains unclear whether the observed changes in circulating miRNA levels are mediators or consequences of the pathophysiological mechanisms involved in COVID-19. Consequently, further experimental studies are required to determine if these miRNAs can mechanistically participate in the pathophysiology of COVID-19 complications. Conclusions This study allowed us to better characterize the miRNA profile associated with COVID-19, particularly its role in immune system activation and inflammation, while also considering the disease’s heterogeneous clinical course, which ranges from mild symptoms to severe complications such as septic shock and ARDS. Our findings may support the use of miRNA signatures for early diagnosis, patient stratification, and prediction of COVID-19 severity. Notably, the combined analysis of multiple miRNAs achieved the highest diagnostic and prognostic accuracy, representing the most significant finding and the most meaningful clinical implication of this study. Furthermore, these results may contribute to clarifying the molecular mechanisms underlying miRNA involvement in the development of COVID-19 complications. Further studies in larger cohorts with broader disease severity, age, and ethnic distributions, along with functional experiments, will be necessary to validate our findings and provide insights into the utility of circulating miRNAs as biomarkers and functional mediators of COVID-19. Future therapeutic strategies should consider the effect of pharmacological therapies on the circulating miRNA pattern in COVID-19 as well as in other critical illnesses. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Acknowledgements The authors would like to thank all the clinicians and staff members in the Emergency Department of the A.O.U. “Città della Salute e della Scienza di Torino” – Molinette Hospital, Turin, Italy, for their work and collaboration. We would also like to thank the Department of Medical Science – University of Turin – for the BioRender License provided to E.L. T.S. thanks PRIN project “Investigating the role of NF-YA isoform/lncRNA axis in mesoderm specification” of the Italian Ministry for Education, University and Research (Grant ID: 20224TWKNJ). Author contributions Conceptualization, M.S., B.V., T.S., E.L., G.M.; methodology, M.S., B.V., T.S., O.B., C.D.; formal analysis, M.S., B.V., T.S., O.B.; writing-original, M.S., B.V., T.S., E.L. drafted manuscript. M.S., B.V. T.S., O.B., P.C.V, E.P., F.M., P.F., E.L., G.M. revised manuscript. B.V., P.F. were occupied in founding acquisition. All authors read and approved the final manuscript. Funding This work was supported by Ministero dell’Università e della Ricerca Scientifica e Tecnologica (MURST) ex 60% to B.V. and P.F and Research Projects of National Relevance (PRIN) prot. 2022BPNY3E to EL. Data availability Datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. The datasets generated and/or analysed during the current study are available in the Gene Expression Omnibus (GEO) repository (GSE302626). Declarations Ethics approval and consent to participate The present study was conducted in compliance with the World Medical Association Declaration of Helsinki. All patients provided oral informed consent (as specifically authorized by the Ethical Committee during the COVID-19 pandemic due to infection risk management in ER) for the participation to the study, and all data were immediately de-identified. The study was approved by the Institutional Ethical Committee (A.O.U. Città della Salute e della Scienza di Torino – A.O. Ordine Mauriziano – A.S.L. Città di Torino) (number of approval CS2/139). Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Martina Schiavello and Barbara Vizio contributed equally to this work. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Data Availability Statement Datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. The datasets generated and/or analysed during the current study are available in the Gene Expression Omnibus (GEO) repository (GSE302626).