Integrative blood transcriptomics identifies U1-snRNP gene repression and miRNA–mRNA regulatory hubs in pre-clinical Parkinson’s Disease

Integrative blood transcriptomics identifies U1-snRNP gene repression and miRNA–mRNA regulatory hubs in pre-clinical Parkinson’s Disease | 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 Integrative blood transcriptomics identifies U1-snRNP gene repression and miRNA–mRNA regulatory hubs in pre-clinical Parkinson’s Disease Davide D'Angelo, Alfonso De Simone, Nunzio D'Agostino This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9074426/v1 This work is licensed under a CC BY 4.0 License Status: Under Revision Version 1 posted 9 You are reading this latest preprint version Abstract Parkinson’s disease (PD) is a progressive neurodegenerative disorder characterised by the loss of dopaminergic neurons and the accumulation of intraneuronal misfolded α-synuclein aggregates. Despite advances in understanding the underlying molecular mechanisms of PD, diagnosis still relies largely on clinical symptoms that appear relatively late in disease progression. To advance both the mechanistic understanding of PD and the development of new diagnostic tools, we here performed an integrative analysis of matched whole-blood mRNA-seq and miRNA-seq data from 2,604 PPMI participants (Healthy n = 658; PD n = 1,946), stratified into pre-clinical (n = 561), mild, moderate, and severe stages. Differential expression (adjusted p 1.5) revealed a U-shaped trajectory, with maximal alterations in pre-clinical (7,427 DEGs; 157 DE-miRNAs) and severe PD. Pre-clinical signatures were dominated by widespread downregulation, notably involving multiple U1 snRNA genes, indicating early repression of U1-snRNP spliceosomal components. Integration of predicted targets with Spearman anti-correlation identified 252 miRNA–mRNA pairs, driven by hub miRNAs including miR-34a-5p, miR-1299, miR-3120-5p, and miR-412-3p. Using pre-clinical DE-miRNAs for classification (n = 1,219), Naive Bayes and GLMNet achieved AUCs of 0.886 and 0.883 with minimal overfitting. Taken together, these findings highlight the possible employment of early spliceosomal dysfunction and miRNA-mediated regulation as novel blood-based biomarkers of pre-clinical PD. Health sciences/Biomarkers Biological sciences/Computational biology and bioinformatics Health sciences/Diseases Biological sciences/Genetics Biological sciences/Molecular biology Health sciences/Neurology Biological sciences/Neuroscience transcriptomics microRNA parkinson's disease biomarkers multi-omics blood-based diagnostics Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Introduction Parkinson’s disease (PD) is the second most prevalent age-related neurodegenerative disorder and represents a growing global health burden driven by population aging (Dorsey et al., 2018). Clinically, PD affects more than 1% of individuals over the age of 60 and more than 10 million people worldwide, a figure expected to increase substantially in the coming decades (Su et al., 2025). The disease is characterised by a core triad of motor symptoms – bradykinesia, resting tremor, and muscular rigidity - often accompanied by postural instability. These manifestations arise primarily from the progressive degeneration of dopaminergic neurons within the substantia nigra pars compacta and the resulting depletion of dopamine within basal ganglia circuits (Jankovic, 2008). Beyond its cardinal motor features, PD is increasingly recognized as a multisystem disorder encompassing a wide range of non-motor symptoms, including olfactory dysfunction, sleep disturbances (e.g. rapid eye movement sleep behavior disorder, RBD), gastrointestinal dysfunction, mood disorders, and cognitive impairment (Chaudhuri and Schapira, 2009). Notably, many of these non-motor symptoms can emerge years or even decades before the onset of overt motor deficits, defining a prodromal phase of PD (Berg et al., 2015). By the time motor symptoms lead to clinical diagnosis, substantial neurodegeneration has already occurred, with approximately 50–60% of dopaminergic neurons in the substantia nigra loss and striatal dopamine levels reduced of approximately 80% (Bernheimer et al., 1973; Fearnley and Lees, 1991). This delayed clinical recognition highlights a critical need for strategies of early detection. Identifying PD during its pre-clinical or prodromal stages could allow the implementation of neuroprotective or disease-modifying interventions prior to extensive neuronal damage occurring (Berg et al., 2015). However, early diagnosis remains a major challenge, as no definitive biomarkers or diagnostic tests are currently available, and diagnosis continues to rely primarily on clinical features that manifest relatively late in the disease course. The pathophysiology of PD is driven by the progressive degeneration of nigrostriatal dopaminergic neurons and the accumulation of misfolded alpha-synuclein (α-syn) protein aggregates in vulnerable brain regions. A pathological hallmark of PD is the presence of intraneuronal inclusions known as Lewy bodies, which are composed predominantly of fibrillar α-syn and are closely associated with neuronal dysfunction and cell death (Spillantini et al., 1997; Surmeier et al., 2017). Misfolded α-syn proteins overwhelm neuronal proteostasis mechanisms, leading to impaired protein degradation and the accumulation of toxic oligomeric species. These aggregates disrupt synaptic transmission, compromise mitochondrial function, and impair lysosomal-autophagic pathways, thereby promoting cellular stress and progressive neurodegeneration (Breydo et al., 2012; Lashuel et al., 2013). Mitochondrial dysfunction and oxidative stress are widely regarded as central and converging drivers to PD pathology. This is underscored by familial PD mutations (e.g. in SNCA, PINK1, GBA, DJ-1, LRRK2 genes), which are involved in mitochondrial quality control, lysosomal function, and reactive oxygen species homeostasis (Pickrell and Youle, 2015; Sobhifar and Brown, 2025). In parallel, chronic neuroinflammation has emerged as a significant pathogenic component of PD. Activated microglia and infiltrating peripheral immune cells release pro-inflammatory cytokines and neurotoxic mediators that further exacerbate dopaminergic neuron vulnerability and loss (Hirsch et al., 2012). Together, α-syn aggregation, mitochondrial failure, oxidative stress, and sustained inflammation form a network of mutually reinforcing processes that ultimately drive progressive neuronal loss in the substantia nigra and widespread dysfunction of neural networks underlying the clinical manifestations of PD (Hirsch et al., 2012). Importantly, PD is increasingly recognized as a multisystem disorder, with pathological changes extending beyond the central nervous system. According to Braak’s staging hypothesis, α-syn pathology may originate in peripheral sites such as the olfactory epithelium or enteric nervous system years before the onset of motor symptoms, subsequently propagating to the brain via neural or humoral pathways (Braak et al., 2003a; Braak et al., 2003b). This systemic view of PD pathogenesis supports the concept that peripheral biological alterations accompany central neurodegeneration. In particular, accumulating evidence of immune dysregulation in PD suggests that blood-based molecular changes may reflect ongoing disease processes (Tönges et al., 2022). Among emerging molecular regulators, microRNAs (miRNAs) have garnered significant attention in PD research. miRNAs are small non-coding RNAs that fine-tune gene expression at the post-transcriptional level, and multiple miRNAs have been found dysregulated in PD patient brains as well as in peripheral biofluids (Shaheen et al., 2024). Collectively, these findings provide a strong rationale for investigating blood-based transcriptomic alterations, including both mRNA and miRNA expression profiles, as accessible biomarkers capable of capturing key molecular processes associated with the early, pre-clinical stages of PD. Despite substantial advances in our understanding of PD, major challenges remain in translating this knowledge into reliable tools for early diagnosis and prognosis. A critical unmet need is the identification of validated biomarkers capable of detecting PD during its preclinical stages, when therapeutic interventions may be most effective and before extensive, irreversible neurodegeneration has occurred. Although neuroimaging modalities and cerebrospinal fluid assays (e.g., measurements of α-synuclein species or neurotransmitter metabolites) have shown promise, their application in early PD remains limited by invasiveness, high cost, limited scalability, and insufficient disease specificity (Cheslow et al., 2021; Posavi et al., 2019). In this context, blood-based biomarkers represent an especially attractive alternative due to their minimal invasiveness, suitability for longitudinal monitoring, and broad clinical applicability. However, to date no blood-based diagnostic test for PD has received regulatory approval, and reported candidate biomarkers, including gene expression signature, protein markers, and miRNAs, have not yet been consistently validated in large, well-characterized prospective cohorts. Moreover, transcriptomic and proteomic alterations in peripheral blood are often subtle and susceptible to confounding by demographic, clinical and environmental factors, highlighting the difficulty of distinguishing disease-related signals from background biological and technical variability. A further gap in the field concerns the limited understanding of how peripheral molecular changes reflect central nervous system pathology, particularly during the earliest stages of PD. The biological mechanisms linking blood-based biomarkers to neurodegenerative processes remain incompletely defined. For example, alterations in leukocyte gene expression may mirror neuroinflammatory activity within the brain, while circulating miRNAs could indicate neuronal injury or α-syn–related pathology; however, these associations remain poorly characterised. Addressing this knowledge gap will require integrative, systems-level approaches that combine multi-omics data with clinical phenotypes and neurobiological endpoints. The present study was designed to address these challenges by systematically investigating blood-based transcriptomic and miRNA alterations across stages of PD, with a particular emphasis on the pre-clinical phase. Specifically, our objectives were to: ( 1 ) integrate multiple independent blood-based transcriptomic datasets, including both mRNA and miRNA expression profiles, from large and heterogeneous PD cohorts to enhance statistical power and reproducibility; ( 2 ) identify differentially expressed genes and miRNAs associated with pre-clinical PD and characterize their regulatory interactions; ( 3 ) perform pathway enrichment analyses to elucidate biological processes perturbed during the early stage of the disease; and ( 4 ) propose and validate a panel of candidate blood-based biomarkers (comprising both mRNA and miRNA markers) — using machine learning classification models to assess their potential for distinguishing individuals with preclinical PD from healthy controls. Materials and methods Data collection and pre-processing Data were obtained from the Parkinson's Progression Markers Initiative (PPMI) database (Marek et al., 2011) and comprised demographic, clinical, hematological, bulk RNA-sequencing, and small non-coding RNA (sncRNA)-sequencing data derived from blood samples. RNA-sequencing data were quality-filtered by PPMI prior to release, and raw count matrices were downloaded for downstream analyses. To construct a unified cohort, samples with missing values (NA) in any dataset were excluded, resulting in a final dataset of 2,604 samples with complete data coverage. Individuals were originally classified according to the Neuronal Synuclein Disease (NSD) stage and were subsequently regrouped into five categories: Healthy, Pre-clinical, Mild, Moderate, and Severe. The Pre-clinical stage was defined by the presence of neuronal α-sync detected in cerebrospinal fluid (CSF) using the seed amplification assay (SAA), together with evidence of dopaminergic dysfunction (e.g., imaging abnormalities), in the absence of overt functional impairment. Hematological variables with isolated missing values were imputed using category-specific means. Patient identifiers and clinical events (longitudinal samplings) were consolidated into unique sample identifiers, formatted as patient ID combined with event code. Ensembl gene identifiers in the raw count matrices were trimmed to remove version numbers. Gene annotations, including gene symbols and biotypes, were retrieved from the latest GRCh38 Ensembl-based annotation database using the AnnotationHub package v3.16.1 (Martin Morgan, 2017) and the ensembldb package v2.32.0 (Johannes Rainer, 2017). Gene symbols were validated and corrected for outdated aliases using the checkGeneSymbols() function from the HGNChelper package v0.8.15 (Oh et al., 2012). Approved HGNC symbols were assigned where available; unresolved entries were set to NA, and correction frequencies were recorded. A summary annotation table was generated (Supplementary table 1 ), filtering out unmapped genes. For genes with duplicate HGNC symbols, entries were ordered by mean expression and the highest-expressed transcript was retained, resulting in a final expression matrix with unique HGNC gene symbols as row identifiers. RNA differential expression analysis Differential expression analyses of both mRNAs and miRNAs were performed using the MIRit package v1.4.2 (Ronchi and Foti, 2023), which implements the statistical framework of edgeR. Prior to differential testing, Multidimensional Scaling (MDS) analysis was conducted to assess sample variability and identify potential outliers. Low-abundance features were subsequently filtered using stringent criteria: min.count = 5, min.total.count = 10, large.n = 10, min.prop = 0.3. Differential expression was assessed using a generalized linear model (GLM) quasi-likelihood F-test. A robust design matrix was constructed with disease stage as the primary variable of interest, while adjusting for potential confounding covariates including age, sex, body mass index (BMI), white blood cell (WBC) count, platelet count, and neutrophil-to-lymphocyte ratio (NLR). Pairwise contrasts were performed by comparing each pathological stage against the Healthy control group. Features were considered significantly differentially expressed based on a nominal p-value < 0.05 and an absolute fold-change threshold of 1.5. Identification of miRNA–mRNA regulatory interactions Potential regulatory targets of the differentially expressed miRNAs (DEMs) were identified through an integrative approach combining target prediction and expression correlation analysis. Putative targets for each DEM were retrieved using the getTargets() function from the MIRit package. To assess the functional relevance of miRNA-mRNA interactions, Spearman correlation analyses were performed between miRNA and mRNA expression profiles of the identified DEGs and DEMs. Prior to correlation analysis, batch correction was applied to mitigate the influence of confounding technical and biological factors using a Weighted Surrogate Variable Analysis. Specifically, patient-specific effects and sex were adjusted for, while preserving variability associated with relevant biological covariates, including age, BMI, WBC, platelet count, and NLR. Additionally, WSVA was used to estimate and remove two latent surrogate variables, with features weighted by their standard deviation to more effectively capture systematic sources of unwanted variation. Following batch correction, miRNA and gene expression datasets were integrated using the mirnaIntegration() function from MIRit . This integration was performed separately for each disease stage, linking batch-corrected DEM and DEG profiles to identify statistically significant negative correlations consistent with miRNA-mediated post-transcriptional repression. Finally, relevant correlation networks were visualized using Fluorish ( https://flourish.studio/ ). Functional enrichment analysis To elucidate the biological relevance of the identified molecular signatures, functional enrichment analyses were performed with shinyGO v0.85 (Ge et al., 2020) on (i) the DEGs identified in the Pre-clinical stage and (ii) the subset of target DEGs showing strong negative correlations with upregulated miRNAs. An Over-Representation Analysis (ORA) was performed to identify significantly enriched biological processes (BP) by querying the Gene Ontology (GO) database for Homo sapiens . Only BPs supported by at least three associated genes were retained for downstream interpretation. Machine learning framework for pre-clinical stage classification To assess the predictive potential of the identified miRNA biomarkers, a machine learning-based classification framework was implemented to distinguish subjects in the Pre-clinical and Healthy stages. All analyses were performed in R using the caret package v7.0.1 (Kuhn, 2008). A dedicated modeling dataset was generated by filtering the original miRNA count matrix to retain only significant differentially expressed miRNAs (DEMs) identified in the "Pre-clinical vs. Healthy" contrast (adjusted p-value 0.58). This resulting expression matrix was then merged with clinical metadata to incorporate disease stage labels. The dataset comprised 1,219 samples (Healthy, n = 658; Pre-clinical, n = 561). Using a stratified sampling strategy to preserve disease-stage class proportions, the data was divided into a training set (70%, n = 854; Healthy, n = 461; Pre-clinical, n = 393) and an independent test set (30%, n = 365; Healthy, n = 197; Pre-clinical, n = 168). Eight supervised learning algorithms were trained and evaluated: Random Forest (RF), Support Vector Machine with Radial Basis Function kernel (SVM-Radial), Generalized Linear Model with Elastic Net regularized logistic regression (GLMNet), Gradient Boosting machine (GBM), k-Nearest Neighbors (KNN), Naive Bayes (NB), Linear Discriminant Analysis (LDA) and Neural Network (NN). Model training employed 10-fold cross-validation repeated five times to ensure robust performance and to minimize overfitting. Model performance was evaluated on the independent test set using confusion matrices and multiple evaluation metrics, including Accuracy, Kappa, Sensitivity, Specificity, Precision, F1 Score, and Balanced Accuracy. The Receiver Operating Characteristic (ROC) curves were estimated, and Area Under the Curve (AUC) values were calculated and visualized with pROC v1.19 (Robin et al., 2010). In addition, feature importance measures were extracted from caret’s available algorithm to identify miRNAs contributing most strongly to the classification task. Finally, to assess potential overfitting, the models were also evaluated on the training data, and the AUC was computed. Overfitting was quantified as the difference between the training and test AUCs (the AUC gap). Results Cohort characteristics and data exploration Following quality control, filtering, and cross-platform harmonization of transcriptomic, miRNA, clinical, and hematological datasets, we retained only samples with complete multi-omics coverage for downstream analyses (Fig. 1 ). The final cohort comprised 2,604 individuals, including 658 healthy controls and 1,946 PD-related samples. PD cases were stratified into pre-clinical (n = 561), mild (n = 197), moderate (n = 910), and severe (n = 278) stages after regrouping (Fig. 1 A). Across all disease stages, the cohort exhibited a consistent male predominance (~ 60%), consistent with the known PD epidemiology (Fig. 2 B) along with a predominance of participants aged 60–70 years (Fig. 2 C). The mean age increased from healthy controls to severe PD stages. Hematological parameters, including white blood cell count, platelet count, neutrophil count, lymphocyte count, and the neutrophil-to-lymphocyte ratio, demonstrated stage-dependent variability, suggesting systematic alterations associated with disease progression. Importantly, the same set of samples were used for both RNA-seq and miRNA-seq analyses, resulting in a fully integrated cohort and eliminating potential biases arising from partial data overlap. Dynamic patterns of mRNA and miRNA expression across Parkinson’s disease progression Differential expression analysis revealed extensive transcriptional dysregulation across PD stages relative to healthy controls (Supplementary table 2 ). MDS analysis on mRNA expression profiles identified ‘sex’ as the primary source of variance in the dataset (Supplementary Fig. 1). Among disease stages,the pre-clinical group exhibited the most pronounced transcriptional alterations, with 7,427 DEGs. Substantial transcriptome reprogramming was also observed in the severe stage, where 3,393 DEGs were identified. In contrast, the mild and moderate stages showed markedly fewer changes, with only 29 and 33 DEGs detected, respectively (Fig. 2 ). In the pre-clinical stage, more than two-thirds of DEGs were downregulated, indicating widespread repression of transcriptional programs prior to the onset of overt clinical symptoms. A similar bias toward downregulation was observed in the severe stage. Notably, only 1,531 DEGs were common between the pre-clinical and severe stages, while other stage comparisons shared a maximum of just 8 DEGs, suggesting that transcriptional alterations are largely stage-specific (Fig. 3 ). Collectively, these findings highlight a U-shaped pattern of transcriptional disruption, with maximal perturbation occurring at the pre-clinical stage and re-emerging at advanced disease stages. MiRNA expression analysis revealed a stage-dependent pattern that paralleled the mRNA dysregulation. Consistent with the mRNA results, the pre-clinical stage showed the largest number of differentially expressed miRNAs (DEmiRNAs), with 157 identified relative to healthy controls (Supplementary Table 3). The severe stage again showed notable dysregulation with 35 DEmiRNAs detected, whereas the mild and moderate stages exhibited only 13 and 7 DEmiRNAs, respectively (Fig. 2 ). In the pre-clinical stage, the majority of DEmiRNAs were upregulated, consistent with a potential repressive effect on target mRNA expression. This pattern aligns with the predominance of downregulated mRNAs observed at the same stage and is consistent with a regulatory relationship between miRNA overexpression and mRNA repression. Early and late disease miRNA–mRNA interaction patterns Integration of DEGs and DEMs revealed a clear stage-dependent pattern of miRNA–mRNA associations. No significant correlations were detected in the Mild and Moderate stages. In contrast, multiple miRNA–mRNA interactions emerged in the preclinical and severe stages, with the strongest signal observed in the preclinical stage. To focus on the most biologically relevant interactions, analyses were restricted to strong negative correlations (Spearman’s ρ < −0.8). Under this stringent threshold, 22 miRNA–mRNA correlations were identified in the Severe stage, all driven by only two miRNAs, hsa-miR-412-3p and hsa-miR-1262, suggesting a highly centralized regulatory architecture in advanced disease. In the Pre-clinical stage, 252 strong negative correlations were detected, only six miRNAs (Fig. 4 ), consistent with an early regulatory network dominated by a limited set of miRNA regulators. Among these, four miRNAs, namely hsa-miR-1299, hsa-miR-34a-5p, hsa-miR-3120-5p, and hsa-miR-412-3p, emerged as major hubs, accounting for the largest number of anti-correlated target transcripts. Notably, hsa-miR-34a-5p also ranked among the most upregulated miRNAs, supporting concordance between miRNA overexpression and repression of target genes. In addition, several target genes exhibited negative correlations with two or three distinct miRNAs, suggesting coordinated post-transcriptional regulation during the pre-clinical phase. Functional characterization Functional enrichment analysis of pre-clinical DEGs revealed a heterogeneous biological landscape, encompassing epithelial differentiation and keratin-related processes, chemosensory processes, and broader nervous-system-associated functions (Fig. 5 A). This diversity suggests that early molecular alterations affect multiple biological systems rather than being confined solely to neuronal pathways. Consistent with this observation, examination of the top 20 pre-clinical DEGs (Table 1 ) showed that the strongest expression changes were dominated by U1 small nuclear RNA (snRNA) genes. These transcripts are not captured by standard enrichment analyses due to the absence of curated functional annotations U1 snRNA genes encode components of the U1 small nuclear ribonucleoprotein (snRNP) complex, which plays a central role in pre-mRNA splicing and transcriptional regulation. Notably, the long non-coding RNA LINC03126, also highly ranked among preclinical DEGs, is predicted to be associated with the U1 snRNP complex, further supporting a potential role for RNA-processing mechanisms in early disease stages. In contrast, functional enrichment restricted to genes involved in the miRNA–mRNA interactions with strong negative correlations, yielded a markedly more neurocentric signature. Enriched GO terms converged predominantly on processes related neurodevelopment and neuronal communication, including nervous system development, neurogenesis, neuronal differentiation, cell morphogenesis associated with neuronal fate, and synapse-related signaling pathways (Fig. 5 B). This shift toward neuronal specificity underscores the additional biological resolution provided by integrating miRNA regulation, uncovering molecular programs more directly linked to early Parkinson’s disease pathophysiology. Table 1 Top 20 differentially expressed genes in the pre-clinical stage of Parkinson's disease ranked by magnitude of expression change. The list includes U1 small nuclear RNAs (snRNAs) and long non-coding RNAs (lncRNAs) among the top candidates. Gene ID Gene type Log 2 (fold change) P-value Adjusted p-value LINC03126 lncRNA -5.75 9.80E-73 1.83E-68 RNU1-83P snRNA -3.07 2.20E-70 2.73E-66 RNU1-28P snRNA -2.74 4.94E-51 1.02E-47 RNU1-88P snRNA -2.67 2.09E-52 4.58E-49 RNU1-67P snRNA -2.61 5.61E-45 6.53E-42 RNU1-18P snRNA -2.54 1.05E-33 3.45E-31 RNU1-27P snRNA -2.43 1.39E-34 5.33E-32 RNU1-11P snRNA -2.41 4.54E-36 2.14E-33 RNVU1-7 snRNA -2.27 4.62E-30 9.79E-28 RNU1-3 snRNA -2.26 1.36E-35 5.95E-33 RNU2-36P snRNA -2.25 6.22E-47 1.01E-43 RNU1-2 snRNA -2.09 1.60E-39 1.08E-36 RNVU1-29 snRNA -2.06 1.47E-57 5.49E-54 RNU6ATAC snRNA -1.83 9.91E-54 2.46E-50 RNU1-89P snRNA -1.80 6.62E-21 5.44E-19 RNU4-2 snRNA -1.76 9.62E-46 1.24E-42 RNU4-1 snRNA -1.74 9.72E-33 2.81E-30 RNVU1-31 snRNA -1.72 3.80E-33 1.11E-30 RNU1-148P snRNA -1.72 1.59E-25 2.13E-23 Machine learning classification of the pre-clinical stage All eight binary classification models achieved predicted performance above chance on the test set, with area under the ROC curve (AUC) values ranging from 0.791 to 0.886. The highest overall performance was observed for Naive Bayes (Accuracy = 0.814, F1 = 0.800, Balanced Accuracy = 0.813, AUC = 0.886), closely followed by GLMNet (Accuracy = 0.800, F1 = 0.786, Balanced Accuracy = 0.800, AUC = 0.883) and SVM-Radial (Accuracy = 0.789, F1 = 0.763, Balanced Accuracy = 0.785, AUC = 0.860). Gradient Boosting and the Neural Network models showed intermediate performance, while k-NN and LDA displayed lower AUC values. Finally, Naive Bayes and, particularly, GLMnet demonstrated the lowest degree of overfitting among the evaluated models (Table 2 ). To explore mode interpretability, feature-importance analyses were performed for models that support such estimates (Random Forest, Gradient Boosting, and GLMNet). These analyses consistently highlighted a small subset of DEMs as key contributors to classification. Notably, hsa-miR-4433a-5p and hsa-miR-34a-5p ranked among the top three features across all three models. Importantly, these miRNAs were also identified as major regulatory hubs in the miRNA-mRNA interaction analysis (Supplementary Table 4), underscoring their potential biological relevance in the pre-clinical stage of Parkinson’s disease. Table 2 Performance of eight binary classification models on the test set in distinguishing pre-clinical from healthy subjects using differentially expressed miRNAs as features. Metrics include accuracy, sensitivity, specificity, and area under the ROC curve (AUC). Model Accuracy Kappa Sensitivity Specificity Positive Predictive Value Negative Predictive Value Precision Recall F1 Balanced Accuracy AUC AUC Gap Naive Bayes 0.81 0.63 0.81 0.82 0.79 0.83 0.79 0.81 0.80 0.81 0.89 0.02 GLMNet 0.80 0.60 0.80 0.80 0.78 0.82 0.78 0.80 0.79 0.80 0.88 0.01 SVM 0.79 0.57 0.74 0.83 0.79 0.79 0.79 0.74 0.76 0.79 0.86 0.11 Gradient Boosting 0.78 0.56 0.76 0.80 0.77 0.80 0.77 0.76 0.76 0.78 0.85 0.1 Neural Network 0.78 0.56 0.78 0.78 0.75 0.81 0.75 0.78 0.76 0.78 0.85 0.15 Random Forest 0.75 0.49 0.77 0.73 0.71 0.79 0.71 0.77 0.74 0.75 0.83 0.16 k-NN 0.73 0.46 0.78 0.68 0.68 0.78 0.68 0.78 0.72 0.73 0.80 0.02 Linear Discriminant Analysis 0.69 0.40 0.84 0.57 0.62 0.81 0.62 0.84 0.72 0.70 0.79 -0.01 Discussion In this study, we performed an integrative analysis of whole-blood mRNA and miRNA expression profiles across multiple stages of PD to identify early molecular signatures and regulatory networks associated with the pre-clinical phase of the disorder. This large-scale in silico analysis of publicly available multi-omics datasets underscores the power of integrative computational approaches to elucidate early molecular mechanisms in neurodegenerative diseases. By systematically mining whole-blood mRNA and miRNA expression profiles from ~ 2,600 individuals spanning healthy controls, pre-clinical, and clinically manifest PD stages, we identified stage-specific molecular signatures that precede overt motor symptoms. Our findings build upon earlier blood-based transcriptomic studies demonstrating the possibility of identifying molecular markers of early PD in peripheral blood (Scherzer et al., 2007) as well as studies characterizing transcriptional alterations in PD blood mononuclear cells, including in genetic forms such as LRRK2 mutation carriers (Mutez et al., 2011). The scale and heterogeneity of the dataset of our study enabled robust detection of subtle yet biologically coherent regulatory patterns that may not be apparent in smaller cohorts, consistent with previous large-scale blood-based transcriptomic and multi-omics integration studies demonstrating that harmonized, high-dimensional datasets increase sensitivity for detecting disease-associated RNA signatures and improve predictive modeling performance in PD (Dong et al., 2025; Irmady et al., 2023; Tng et al., 2024) A key aspect of our study is that it focuses on the pre-clinical phase of PD, a critical but poorly characterized window during which neuropathological processes are likely already underway despite the absence of classical motor manifestations. Molecular alterations detected at this stage may reflect early systemic responses to incipient neurodegeneration, compensatory regulatory mechanisms, or primary pathogenic drivers. By exploiting integrated mRNA–miRNA network analyses, not only we were able to catalogue differentially expressed transcripts but we could also infer regulatory relationships that may actively shape early disease biology. Together, these results demonstrate how computational multi-omics can move beyond descriptive transcriptomics toward mechanistic insight, offering a scalable strategy to identify candidate biomarkers and molecular pathways that may enable earlier diagnosis, improved assessment of risk stratification, and ultimately preventive intervention in PD. One of the most striking findings of the present study was to observe a robust downregulation of multiple U1 small nuclear ribonucleoprotein (U1-snRNP) complex genes in the pre-clinical stage of PD. U1-snRNP is a core component of the spliceosome and plays a key role in the regulation of gene expression (Campagne, 2024). It initiates spliceosome assembly by recognizing the 5′ splice site through base pairing between U1 snRNA and pre-mRNA (Mount et al., 1983), a step that represents a key rate-limiting stage in regulated splicing. Beyond splice-site recognition, U1 also contributes to proper 3′-end mRNA formation by suppressing premature cleavage and polyadenylation and coordinating productive transcription with RNA polymerase II (Almada et al., 2013; Berg et al., 2012). Through these dual functions, U1 maintains transcript integrity and helps preserve global transcriptional stability. Within this framework, the early downregulation of U1-related genes observed in our dataset may reflect impaired RNA-processing capacity preceding overt clinical symptoms. Even modest reductions in U1 availability could indeed destabilize splicing fidelity and transcript length control, leading to subtle but widespread transcriptomic remodeling. Spliceosomal dysfunction is increasingly recognized as a driver of neurodegeneration (Nikom and Zheng, 2023), whereas the disruption of U1-snRNP biogenesis has been linked to RNA metabolism defects in disorders such as spinal muscular atrophy, ontocerebellar hypoplasia type 7, and FUS-associated amyotrophic lateral sclerosis (Campagne, 2024). In Alzheimer’s disease (AD), U1 snRNP components redistribute to cytoplasmic aggregates and are associated with aberrant RNA processing and increased premature cleavage/polyadenylation (Zhu et al., 2020). Furthermore, U1 snRNP aggregation has been directly demonstrated in AD cases with autosomal dominant mutations and trisomy 21, reinforcing the concept that spliceosomal instability may represent a shared vulnerability across neurodegenerative conditions (Hales et al., 2014) Although similar U1 aggregates have not been detected in PD brains (Bai et al., 2013), our findings suggest that PD may involve a more subtle, functional reduction in spliceosomal capacity rather than overt protein mislocalization. Importantly, mis-splicing of PD-relevant genes has been documented. In cellular models of PARK7 (DJ-1) mutation, U1-dependent splicing defects lead to exon 3 exclusion and loss of the DJ-1 protein, resulting in mitochondrial dysfunction, whereas restoration of U1-snRNP levels rescues normal splicing and protein expression (Boussaad et al., 2020). These observations provide mechanistic plausibility that early U1-snRNP downregulation could reduce expression of neuroprotective factors such as DJ-1 or broadly perturb mitochondrial and stress-response pathways. Additional evidence supporting the link between splicing alterations and PD comes from leukocyte small RNA sequencing studies showing that deep brain stimulation modulates splicing-related signatures and that peripheral RNA profiles can classify brain-region transcriptomes (Soreq et al., 2013), further indicating systemic RNA-processing involvement in PD. Whether the observed blood U1-snRNP suppression reflects a systemic response to early neurodegeneration or a primary regulatory disturbance remains unclear. It may arise from disease-related epigenetic remodeling of housekeeping gene networks or from immune-cell adaptations to incipient pathology. Regardless of origin, our data position spliceosomal regulation as a potential early molecular vulnerability in PD. Future studies should validate these findings in independent cohorts and determine whether therapeutic modulation of RNA splicing can mitigate disease progression. Another key finding of our study is the identification of a U-shaped trajectory in blood transcriptomic alterations across PD progression. Specifically, we observed a widespread downregulation of mRNAs in the pre-clinical stage, followed by a marked reduction in the number of DEGs during the mild and moderate clinical stages, and finally a renewed and pronounced transcriptional downregulation in severe disease. This non-linear dynamic suggests that peripheral transcriptomic remodelling in PD does not progress in a simple monotonic fashion, but instead follows a stage-dependent pattern of molecular engagement and disengagement. Although explicit descriptions of U-shaped dynamics in blood transcriptomics remain limited, multiple studies support the concept of dynamic gene expression changes across PD progression. Whole-blood and PBMC transcriptomic analyses have consistently identified alterations in immune and stress-response pathways, with reproducible DEG signatures across cohorts, supporting the presence of stage-specific peripheral molecular signals (Navarrete et al., 2025). The widespread downregulation observed in the pre-clinical phase may reflect an early systemic response to emerging neuropathology, consistent with accumulating evidence that peripheral immune dysregulation and chronic inflammatory signaling are integral components of PD pathophysiology, even at early disease stages (Tansey et al., 2022). Such suppression could be driven by oxidative stress, inflammation, or shifts in immune cell composition and activation states, in line with single-cell studies demonstrating peripheral immune reprogramming in PD (Moquin-Beaudry et al., 2025). The relative reduction in transcriptomic perturbation during mild and moderate stages may indicate compensatory homeostatic mechanisms, immunological adaptation, or treatment-related effects that transiently reduce detectable differential expression. Longitudinal studies further support the notion that blood gene expression profiles shift dynamically over time rather than progressing linearly (Kõks, 2025). In severe PD, the observed re-emergence of broad downregulation may reflect failure of compensatory processes and the onset of systemic dysregulation, potentially driven by chronic inflammation, sustained metabolic stress, and advanced neurodegeneration. Similar stage-dependent transcriptional shifts have also been observed in brain tissue, reinforcing the concept that PD molecular pathology evolves dynamically across disease stages (Cappelletti et al., 2023). The integrative miRNA–mRNA network analysis provided important mechanistic insight into pre-clinical PD by identifying regulatory hubs rather than isolated differentially expressed transcripts. We detected several highly connected miRNAs, most prominentlymiR-34a-5p, miR-1299, miR-3120-5p, miR-412-3p, and miR-4433a-5p, that were markedly upregulated in the pre-clinical stage and inversely correlated with numerous downregulated mRNA targets. These anti-correlated patterns suggest that a small number of miRNAs may act as upstream regulators driving the widespread gene repression observed early in disease. This regulatory architecture aligns with growing evidence that miRNA-mRNA networks contribute to PD pathogenesis, influencing mitochondrial quality control, inflammatory signaling, and neuronal survival. Multiple independent studies have identified dysregulated circulating miRNAs in PD blood and plasma (Khoo et al., 2012; Vallelunga et al., 2014; Shaheen et. al, 2024). For example, miR-34a-5p has been shown to suppress PINK1 expression and inhibit PINK1-mediated mitophagy, a pathway critical for mitochondrial function and neuronal health (Tai et al., 2021). Additionally, circulating miRNAs are increasingly recognized as candidate blood-based biomarkers for PD, with reproducible expression changes reported across multiple cohorts and sample types (Zhang et al., 2024). The convergence between network centrality and prior functional evidence strengthens the biological plausibility of these miRNAs may act as early molecular drivers rather than passive biomarkers. Collectively, these findings support a model in which coordinated upregulation of a limited set of miRNAs contributes to early transcriptomic remodelling in PD, potentially amplifying subtle regulatory changes into broader pathway-level dysregulation. MiR-34a-5p represents a particularly compelling example among the identified hub miRNAs.. In our pre-clinical PD dataset, it was among the most strongly overexpressed miRNAs and exhibited extensive connectivity within the regulatory network, targeting a broad array of downregulated mRNAs. This high degree of centrality suggests that miR-34a-5p may play a significant role in early transcriptomic remodeling. Our findings are consistent with independent studies reporting elevated miR-34a levels in PD.. Notably, miR-34a has been shown to be significantly increased in plasma-derived extracellular vesicles (EVs) from PD patients, where it demonstrated moderate discriminative performance (AUC ≈ 0.74 in purified EVs) (Grossi et al., 2020). Mechanistically, miR-34a-5p has been directly linked to mitochondrial dysfunction through the negative regulation of PINK1, a key mediator of mitophagy (Tai et al., 2021). Suppression of PINK1-dependent mitochondrial quality control represents a central pathogenic axis in PD, thereby linking miR-34a-5p activity to one of the most well-established molecular pathways in the disease. The convergence between our network-based identification of miR-34a-5p as a highly connected regulatory hub and its experimentally supported role in mitochondrial homeostasis further strengthens the biological plausibility of its involvement in the prodromal phase of PD. As for miR-1299 and miR-3120-5p, direct evidence in PD patient biofluids remains limited. However, data from biologically relevant central nervous system models support their dysregulation. In midbrain organoids derived from induced pluripotent stem cells (iPSCs) of sporadic PD patients, both miR-1299 and miR-3120-5p exhibited robust and reproducible upregulation during neuronal differentiation and maturation (Valente et al., 2025). This convergence between peripheral blood and patient-derived midbrain models strengthens the plausibility that these miRNAs participate in disease-relevant regulatory programs rather than representing incidental findings. Notably, miR-3120-5p has clear mechanistic relevance to neural function, as it can modulate Hsc70/auxilin-mediated uncoating of clathrin-coated vesicles, a process essential for synaptic vesicle recycling and efficient neurotransmission (Scott et al., 2012). Such functions link this miRNA to pathways directly implicated in synaptic dysfunction, a hallmark of early PD pathophysiology. As for miR-4433a-5p, emerging clinical evidence intersects more directly with the prodromal continuum of synucleinopathies. In plasma EVs, miR-4433a-5p levels increase progressively along the trajectory from healthy individuals to idiopathic REM sleep behavior disorder (iRBD) to PD, with iRBD representing a recognized prodromal stage (Li et al., 2024). This stage-dependent elevation closely mirrors our observation of upregulation in pre-clinical PD. Additionally, in silico analyses of human cortical datasets have identified a closely related family member (hsa-miR-4443, also referred to as “microRNA-4433”) as co-upregulated in PD and enriched within immune and synaptic regulatory networks (Ge et al., 2022). Together, these findings suggest that the miR-4433 family may contribute to early neuroinflammatory processes and synaptic regulatory processes relevant to disease progression. In contrast, miR-412-3p currently lacks direct evidence linking it to PD. Existing literature, however, associates this miRNA with other neurodegenerative conditions. In particular, increased expression has been reported in cerebellar tissue from murine models of spinocerebellar ataxia type 2 (SCA2), and EV-related signals have been described in Huntington’s disease contexts, although these observations require cautious interpretation (Beatriz et al., 2023; Paul et al., 2024). In view of these literature reports, miR-412-3p should therefore be considered a relatively novel and exploratory candidate in PD research. Its identification as a network hub in the present analysis suggests the need of specific functional investigations to clarify whether it plays a previously unrecognized role in peripheral or central PD-related pathways. From a translational perspective, our findings provide concrete leads for the development of blood-based biomarkers capable of detecting PD at a pre-clinical stage. Using differentially expressed miRNAs as features, we demonstrated that supervised machine learning models can reliably distinguish pre-clinical PD individuals from healthy controls with strong predictive performance. Across multiple algorithms, including regularized regression and Naive Bayes classifiers, area-under-ROC-curve values consistently ranged between ~ 0.85 and 0.90, underscoring the substantial diagnostic potential of circulating miRNA profiles. Importantly, the most informative features driving classifier performance, such as miR-34a-5p and miR-4433a-5p, were also identified as highly connected hubs in our regulatory network analysis. This convergence between predictive modeling and network centrality strengthens confidence that these miRNAs represent biologically meaningful regulators rather than purely statistical markers within early PD molecular circuitry. Notably, miR-34a-5p alone showed strong classification capacity, consistent with prior reports linking it to PD pathophysiology (Grossi et al., 2020). When combined with additional hub miRNAs, it contributed to a multivariate signature capable of robust discrimination of the pre-clinical PD state. Together, these results highlight the promise of integrating systems biology with machine learning approaches to advance toward clinically actionable early-detection strategies. Despite the strength of the multi-omics integration and the size of the analyzed cohorts, several limitations should be acknowledged. First, the regulatory relationships inferred in our miRNA–mRNA networks are based primarily on correlation and computational target prediction. Although we prioritized anti-correlated pairs and incorporated established miRNA targeting rules to increase biological plausibility, some predicted interactions may represent false positives. Functional validation is therefore essential. Experimental approaches such as luciferase reporter assays to confirm 3′UTR binding, miRNA overexpression or inhibition in cellular systems, and in vivo perturbation in PD-relevant animal models will be necessary to determine whether altering hub miRNAs produces measurable effects on mitochondrial function, splicing regulation, or neurodegenerative phenotypes. Second, although our blood-based findings hold promise for diagnostic applications, they also raise important mechanistic questions—namely, whether these peripheral molecular alterations reflect pathological processes occurring in the central nervous system or instead represent parallel systemic responses independent of primary brain pathology. Recent integrative studies suggest that circulating non-coding RNAs may mirror brain regulatory networks through competing endogenous RNA interactions, supporting the possibility of cross-tissue molecular coupling (Chun and Kim, 2024). To address this question, future studies should examine whether key markers identified here, such as U1-snRNP genes or other strongly downregulated transcripts,are similarly altered in prodromal or early-stage PD brain tissue. Cross-tissue analyses integrating blood, post-mortem brain samples, and induced pluripotent stem cells–derived neuronal models could help determine whether these signatures represent surrogate biomarkers or active participants in PD pathogenesis. Longitudinal studies will also be critical to establish temporal dynamics and assess whether these molecular alterations predict disease conversion or progression. In conclusion this study demonstrates how integrative analysis of blood-based miRNA–mRNA regulatory networks can uncover candidate biomarkers and pathways associated with the pre-clinical phase of PD. By exploiting large-scale multi-omics data, we identified a coordinated pattern of transcriptomic remodeling characterized by widespread downregulation, most notably involving U1-snRNP spliceosomal components, together with the upregulation of specific hub miRNAs that may act as upstream regulatory drivers. These findings suggest that systemic RNA-processing alterations and miRNA-mediated repression may emerge early in the disease course, potentially before the onset of overt motor symptoms. While these results provide biologically plausible insights into prodromal PD, further work is required to strengthen and extend these observations. Integration of additional omics layers, such as proteomics, epigenomics, or single-cell transcriptomics, could further refine the regulatory landscape and clarify causal relationships. Independent cohort validation and prospective longitudinal studies will be essential to determine the robustness, temporal stability, and predictive value of the identified blood signatures. In parallel, mechanistic investigations are needed to establish whether modulation of key miRNAs or spliceosomal components can influence PD-relevant cellular phenotypes or modify neurodegenerative trajectories. Together, these advances may enable the translation of these molecular signatures into improved strategies for early diagnosis while also opening new directions for investigating the underlying molecular mechanisms in PD pathology. Declarations Code availability The code for this study is available in GitHub and can be accessed via https://github.com/dangelodavid/BLOMOPARKED Competing Interests The authors declare that they have no financial or non-financial competing interests Fundings This research was conducted without specific funding from any public, commercial, or not-for-profit funding agencies. Author Contribution D.D.A. conceptualized the study, performed the formal analysis, and wrote the original draft of the manuscript. N.D. and A.D.S. reviewed and edited the original draft. All authors read and approved the final manuscript. Acknowledgement PPMI – a public-private partnership – is funded by the Michael J. Fox Foundation for Parkinson’s Research, and funding partners; including 4D Pharma, Abbvie, AcureX, Allergan, Amathus Therapeutics, Aligning Science Across Parkinson’s, AskBio, Avid Radiopharmaceuticals, BIAL, Biogen, Biohaven, BioLegend, BlueRock Therapeutics, Bristol-Myers Squibb, Calico Labs, Celgene, Cerevel Therapeutics, Coave Therapeutics, DaCapo Brainscience, Denali, Edmond J. Safra Foundation, Eli Lilly, Gain Therapeutics, GE HealthCare, Genentech, GSK, Golub Capital, Handl Therapeutics, Insitro, Janssen Neuroscience, Lundbeck, Merck, Meso Scale Discovery, Mission Therapeutics, Neurocrine Biosciences, Pfizer, Piramal, Prevail Therapeutics, Roche, Sanofi, Servier, Sun Pharma Advanced Research Company, Takeda, Teva, UCB, Vanqua Bio, Verily, Voyager Therapeutics, the Weston Family Foundation and Yumanity Therapeutics.We acknowledge Dr. Carmen Biancaniello for discussions. Data Availability Data used in the preparation of this article was obtained on 2025-09-05 from the Parkinson’s Progression Markers Initiative (PPMI) database (www.ppmi-info.org/access-data-specimens/download-data), RRID:SCR\_006431. For up-to-date information on the study, visit [www.ppmi-info.org](http:/www.ppmi-info.org) . References Almada, A. E., Wu, X., Kriz, A. J., Burge, C. B. & Sharp, P. A. Promoter directionality is controlled by U1 snRNP and polyadenylation signals. Nature 499, 360–363 (2013). Bai, B. et al. U1 small nuclear ribonucleoprotein complex and RNA splicing alterations in Alzheimer's disease. Proc. Natl. Acad. Sci. U.S.A. 110, 16562–16567 (2013). Beatriz, M. et al. Extracellular vesicles improve GABAergic transmission in Huntington's disease iPSC-derived neurons. Theranostics 13, 3707–3724 (2023). Berg, D. et al. MDS research criteria for prodromal Parkinson's disease. Mov Disord 30, 1600–1611 (2015). Berg, M. G. et al. U1 snRNP Determines mRNA Length and Regulates Isoform Expression. Cell 150, 53–64 (2012). Bernheimer, H., Birkmayer, W., Hornykiewicz, O., Jellinger, K. & Seitelberger, F. Brain dopamine and the syndromes of Parkinson and Huntington. Clinical, morphological and neurochemical correlations. J Neurol Sci 20, 415–455 (1973). Boussaad, I. et al. A patient-based model of RNA mis-splicing uncovers treatment targets in Parkinson's disease. Sci. Transl. Med. 12, eaau3960 (2020). Braak, H., Rüb, U., Gai, W. P. & Del Tredici, K. Idiopathic Parkinson's disease: possible routes by which vulnerable neuronal types may be subject to neuroinvasion by an unknown pathogen. J Neural Transm (Vienna) 110, 517–536 (2003). Braak, H. et al. Staging of brain pathology related to sporadic Parkinson's disease. Neurobiol Aging 24, 197–211 (2003). Breydo, L., Wu, J. W. & Uversky, V. N. Α-synuclein misfolding and Parkinson's disease. Biochim Biophys Acta 1822, 261–285 (2012). Campagne, S. U1 snRNP Biogenesis Defects in Neurodegenerative Diseases. ChemBioChem 25, e202300864 (2024). Cappelletti, C. et al. Transcriptomic profiling of Parkinson's disease brains reveals disease stage specific gene expression changes. Acta Neuropathol 146, 227–244 (2023). Chaudhuri, K. R. & Schapira, A. H. V. Non-motor symptoms of Parkinson's disease: dopaminergic pathophysiology and treatment. Lancet Neurol 8, 464–474 (2009). Cheslow, L., Snook, A. E. & Waldman, S. A. Emerging targets for the diagnosis of Parkinson's disease: examination of systemic biomarkers. Biomark Med 15, 597–608 (2021). Chun, K.-Y. & Kim, S.-N. Integrative analysis of plasma and substantia nigra in Parkinson's disease: unraveling biomarkers and insights from the lncRNA–miRNA–mRNA ceRNA network. Front. Aging Neurosci. 16, 1388655 (2024). Dong, X. et al. Multi-omics machine learning classifier and blood transcriptomic signature of Parkinson's disease. Preprint at https://doi.org/ 10.21203/rs.3.rs-6837659/v1 (2025). Dorsey, E. R. et al. Global, regional, and national burden of Parkinson's disease, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016. The Lancet Neurology 17, 939–953 (2018). Fearnley, J. M. & Lees, A. J. Ageing and Parkinson's disease: substantia nigra regional selectivity. Brain 114 (Pt 5), 2283–2301 (1991). Ge, S. X., Jung, D. & Yao, R. ShinyGO: a graphical gene-set enrichment tool for animals and plants. Bioinformatics 36, 2628–2629 (2020). Ge, X. et al. Human microRNA-4433 (hsa-miR-4443) Targets 18 Genes to be a RiskFactor of Neurodegenerative Diseases. CAR 19, 511–522 (2022). Grossi, I. et al. MicroRNA–34a–5p expression in the plasma and in its extracellular vesicle fractions in subjects with Parkinson's disease: An exploratory study. Int J Mol Med 47, 533–546 (2020). Hales, C. M. et al. U1 small nuclear ribonucleoproteins (snRNPs) aggregate in Alzheimer's disease due to autosomal dominant genetic mutations and trisomy 21. Mol Neurodegeneration 9, 15 (2014). Hirsch, E. C., Vyas, S. & Hunot, S. Neuroinflammation in Parkinson's disease. Parkinsonism & Related Disorders 18, S210–S212 (2012). Irmady, K. et al. Blood transcriptomic signatures associated with molecular changes in the brain and clinical outcomes in Parkinson's disease. Nat Commun 14, 3956 (2023). Jankovic, J. Parkinson's disease: clinical features and diagnosis. J Neurol Neurosurg Psychiatry 79, 368–376 (2008). Rainer, J., Gatto, L., Weichenberger, C.X.. Bioconductor https://doi.org/10.18129/B9.BIOC.ENSEMBLDB (2017). Khoo, S. K. et al. Plasma-Based Circulating MicroRNA Biomarkers for Parkinson's Disease. Journal of Parkinson's Disease 2, 321–331 (2012). Kõks, S. The Exon-Based Transcriptomic Analysis of Parkinson's Disease. Biomolecules 15, 440 (2025). Kuhn, M. Building Predictive Models in R Using the caret Package. J. Stat. Soft. 28, (2008). Lashuel, H. A., Overk, C. R., Oueslati, A. & Masliah, E. The many faces of α-synuclein: from structure and toxicity to therapeutic target. Nat Rev Neurosci 14, 38–48 (2013). Li, Y. et al. Candidate biomarkers of EV-microRNA in detecting REM sleep behavior disorder and Parkinson's disease. npj Parkinsons Dis. 10, 18 (2024). Marek, K. et al. The Parkinson Progression Marker Initiative (PPMI). Progress in Neurobiology 95, 629–635 (2011). Morgan, M., Shepherd, L. AnnotationHub. Bioconductor https://doi.org/10.18129/B9.BIOC.ANNOTATIONHUB (2017). Moquin-Beaudry, G. et al. Mapping the peripheral immune landscape of Parkinson's disease patients with single-cell sequencing. Brain 148, 2847–2860 (2025). Mount, S. M., Pettersson, I., Hinterberger, M., Karmas, A. & Steitz, J. A. The U1 small nuclear RNA-protein complex selectively binds a 5′ splice site in vitro. Cell 33, 509–518 (1983). Mutez, E. et al. Transcriptional profile of Parkinson blood mononuclear cells with LRRK2 mutation. Neurobiology of Aging 32, 1839–1848 (2011). Navarrete, F. et al. Whole transcriptome analysis of peripheral blood mononuclear cells from de novo and drug-naive Parkinson's disease patients. Neurotherapeutics 22, e00762 (2025). Nikom, D. & Zheng, S. Alternative splicing in neurodegenerative disease and the promise of RNA therapies. Nat. Rev. Neurosci. 24, 457–473 (2023). Oh, S., Aggarwal, A., Riester, M. & Waldron, L. HGNChelper: Identify and Correct Invalid HGNC Human Gene Symbols and MGI Mouse Gene Symbols. 0.8.15 https://doi.org/10.32614/CRAN.package.HGNChelper (2012). Paul, S. et al. Cerebellar Micro-RNA Profile in a Mouse Model of Spinocerebellar Ataxia Type 2. Neurol Genet 10, e200144 (2024). Pickrell, A. M. & Youle, R. J. The roles of PINK1, parkin, and mitochondrial fidelity in Parkinson's disease. Neuron 85, 257–273 (2015). Posavi, M. et al. Characterization of Parkinson's disease using blood-based biomarkers: A multicohort proteomic analysis. PLoS Med 16, e1002931 (2019). Robin, X. et al. pROC: Display and Analyze ROC Curves. 1.19.0.1 https://doi.org/10.32614/CRAN.package.pROC (2010). Ronchi, J. & Foti, M. MIRit: an integrative R framework for the identification of impaired miRNA-mRNA regulatory networks in complex diseases. Preprint at https://doi.org/10.1101/2023.11.24.568528 (2023). Scherzer, C. R. et al. Molecular markers of early Parkinson's disease based on gene expression in blood. Proc. Natl. Acad. Sci. U.S.A. 104, 955–960 (2007). Scott, H. et al. MiR-3120 Is a Mirror MicroRNA That Targets Heat Shock Cognate Protein 70 and Auxilin Messenger RNAs and Regulates Clathrin Vesicle Uncoating. Journal of Biological Chemistry 287, 14726–14733 (2012). Shaheen, N. et al. MicroRNAs regulation in Parkinson's disease, and their potential role as diagnostic and therapeutic targets. NPJ Parkinsons Dis 10, 186 (2024). Sobhifar, P. & Brown, D. R. Interaction Between α-Synuclein and DJ-1 in Parkinson's Disease. Brain Sci 15, 899 (2025). Soreq, L. et al. Small RNA sequencing-microarray analyses in Parkinson leukocytes reveal deep brain stimulation-induced splicing changes that classify brain region transcriptomes. Front. Mol. Neurosci. 6, (2013). Spillantini, M. G. et al. Alpha-synuclein in Lewy bodies. Nature 388, 839–840 (1997). Su, D. et al. Projections for prevalence of Parkinson's disease and its driving factors in 195 countries and territories to 2050: modelling study of Global Burden of Disease Study 2021. BMJ 388, e080952 (2025). Surmeier, D. J., Obeso, J. A. & Halliday, G. M. Selective neuronal vulnerability in Parkinson disease. Nat Rev Neurosci 18, 101–113 (2017). Tai, Y. et al. miR-34a-5p regulates PINK1-mediated mitophagy via multiple modes. Life Sciences 276, 119415 (2021). Tansey, M. G. et al. Inflammation and immune dysfunction in Parkinson disease. Nat Rev Immunol 22, 657–673 (2022). Tng, T. J. W. et al. Systematic analysis of multi-omics data reveals component-specific blood-based biomarkers for Parkinson's disease. transl med commun 9, 12 (2024). Tönges, L. et al. Blood-based biomarker in Parkinson's disease: potential for future applications in clinical research and practice. J Neural Transm (Vienna) 129, 1201–1217 (2022). Valente, D. et al. Distinct microRNA signatures define sporadic PSP-RS and PD in patient-derived midbrain organoids. iScience 28, 113162 (2025). Vallelunga, A. et al. Identification of circulating microRNAs for the differential diagnosis of Parkinson's disease and Multiple System Atrophy. Front. Cell. Neurosci. 8, (2014). Zhang, W. T. et al. Circulating microRNAs as potential biomarkers for the diagnosis of Parkinson's disease: A meta-analysis. Neurología (English Edition) 39, 573–583 (2024). Zhu, W. et al. Effects of U1 Small Nuclear Ribonucleoprotein Inhibition on the Expression of Genes Involved in Alzheimer's Disease. ACS Omega 5, 25306–25311 (2020). Additional Declarations No competing interests reported. Supplementary Files SupplementaryFigure1.png SupplementaryTable1.csv SupplementaryTable2.xlsx SupplementaryTable3.xlsx SupplementaryTable4.xlsx SupplementaryFilesCaptions.txt Cite Share Download PDF Status: Under Revision Version 1 posted Editorial decision: Revision requested 27 Apr, 2026 Reviews received at journal 08 Apr, 2026 Reviews received at journal 04 Apr, 2026 Reviewers agreed at journal 30 Mar, 2026 Reviewers agreed at journal 30 Mar, 2026 Reviewers invited by journal 29 Mar, 2026 Editor assigned by journal 11 Mar, 2026 Submission checks completed at journal 11 Mar, 2026 First submitted to journal 09 Mar, 2026 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-9074426","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":614926327,"identity":"be685393-7fd9-495a-8150-a0e9df295033","order_by":0,"name":"Davide D'Angelo","email":"","orcid":"","institution":"University of Naples Federico II","correspondingAuthor":false,"prefix":"","firstName":"Davide","middleName":"","lastName":"D'Angelo","suffix":""},{"id":614926328,"identity":"e949f968-58f6-47c2-bd0d-ee07cdb32786","order_by":1,"name":"Alfonso De Simone","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAABAUlEQVRIie3PMWvCQBTA8RcO4nIxa4IlfoWTQKeAX8UuTmdxdBAMCHFzTj9HobPtA7PcB7ihoBJoFzfBKZTeJQot9TJ3uD83vOF+3D0Am+0/1nH2zbCpD4DfzECNhBD2m4TphRjNT1LHroOJ+EvilNMqgW6xOrzJ2XsUy8krTudwNzSQAAmJn7IxhEIw5OIjvpePI8y3LR9Df9vzUgQmOeAkw4cXyRlS10z6SNwerRTZfe41WTznmnyZCauJq19RsyIjFijiZWYy0Lt42ZiGgutdcJCLoyLrgNLNbRIVS6ekVRJ1i6I88Rn2/RWPT/ScDDupaf+mP78I2u/bbDabrbVvurdYdd9jQEEAAAAASUVORK5CYII=","orcid":"","institution":"University of Naples Federico II","correspondingAuthor":true,"prefix":"","firstName":"Alfonso","middleName":"","lastName":"De Simone","suffix":""},{"id":614926329,"identity":"056542b0-7d0e-454d-87fb-d997b7232c70","order_by":2,"name":"Nunzio D'Agostino","email":"","orcid":"","institution":"University of Naples Federico II","correspondingAuthor":false,"prefix":"","firstName":"Nunzio","middleName":"","lastName":"D'Agostino","suffix":""}],"badges":[],"createdAt":"2026-03-09 14:53:30","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9074426/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9074426/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":105907632,"identity":"917f5497-a302-4049-b1e5-85bbbcf3eae7","added_by":"auto","created_at":"2026-04-01 10:33:10","extension":"jpg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":109452,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOverview of the PPMI cohort.\u003c/strong\u003e (A) Flow diagram of the PPMI cohort (n = 2,604) illustarting the reclassification of eight Neuronal Synuclein Disease stages into five clinical groups. (B) Sex distribution of the population under investigation. (C) Age distribution of the cohort. (D) Body Mass Index (BMI) distribution of the population under study.\u003c/p\u003e","description":"","filename":"1.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/91e39476118d89138067e905.jpg"},{"id":105908712,"identity":"435062e5-f9f7-4f78-8892-030585fd3bd4","added_by":"auto","created_at":"2026-04-01 10:39:19","extension":"jpg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":155181,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eDifferential gene expression results for mRNA (top row) and miRNA (bottom row) across the four stages of Parkinson’s disease.\u003c/strong\u003e Genes meeting the significance thresholds of |fold change| \u0026gt; 1.5 and adjusted \u003cem\u003eP\u003c/em\u003e \u0026lt; 0.05 are shown, with down-regulated genes indicated in blue and up-regulated genes in red.\u003c/p\u003e","description":"","filename":"2.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/044578424be8f69619a480f3.jpg"},{"id":105907575,"identity":"dec93e31-d717-4749-8b50-c84c38f01054","added_by":"auto","created_at":"2026-04-01 10:32:47","extension":"jpg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":63433,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eOverlap of differentially expressed genes (DEGs) across Parkinson’s disease stages\u003c/strong\u003e. Vertical bars indicate the number of shared and stage-specific DEGs, while horizontal bars represent the total number of DEGs identified at each disease stage.\u003c/p\u003e","description":"","filename":"3.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/c2d5c30bee71fad4af74af06.jpg"},{"id":105911811,"identity":"8bda69c0-6afe-4f35-a270-9defbc9a74dd","added_by":"auto","created_at":"2026-04-01 10:55:13","extension":"jpg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":195054,"visible":true,"origin":"","legend":"\u003cp\u003eNetwork visualization of the differential expressed miRNAs (purple nodes) and their negatively correlated to differentially expressed target genes (green nodes). Edges represent strong inverse associations (Spearman’s ρ \u0026lt; −0.8), consistent with putative miRNA-mediated post-transcriptional regulation.\u003c/p\u003e","description":"","filename":"4.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/612f9764366902390017e111.jpg"},{"id":105907640,"identity":"b7c2b68b-03f4-4b6c-b5a3-4b3de11dded8","added_by":"auto","created_at":"2026-04-01 10:33:11","extension":"jpg","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":135352,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eFunctional enrichment results of pre-clinical Parkinson’s disease gene signatures. \u003c/strong\u003eGene Ontology (GO) Biological Process terms enriched among: (A) downregulated genes in the pre-clinical stage and (B) downregulated genes in the preclinical stage exhibiting strong miRNA–mRNA negative correlations (correlation coefficient \u0026lt; -0.8)\u003c/p\u003e","description":"","filename":"5.jpg","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/e1a71163593014664cb8a3fa.jpg"},{"id":105912846,"identity":"27b83607-af96-4722-83d0-77d71e0abc63","added_by":"auto","created_at":"2026-04-01 11:03:14","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1693785,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/2b9633b6-ea11-4cdf-b280-7edc613b9d7c.pdf"},{"id":105907720,"identity":"a3482d34-8d60-4f19-a2bd-8685b1a2c86a","added_by":"auto","created_at":"2026-04-01 10:33:23","extension":"png","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":775131,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFigure1.png","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/918ec16e6f7ffd8e760d8b6c.png"},{"id":105907726,"identity":"5a1063cd-4524-42c2-b568-c3c5eeda27d3","added_by":"auto","created_at":"2026-04-01 10:33:23","extension":"csv","order_by":2,"title":"","display":"","copyAsset":false,"role":"supplement","size":2518719,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable1.csv","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/3fcecbe036727d9a9f3456dc.csv"},{"id":105907519,"identity":"9528c480-2afa-4b48-95d8-ac71c723ba71","added_by":"auto","created_at":"2026-04-01 10:32:26","extension":"xlsx","order_by":3,"title":"","display":"","copyAsset":false,"role":"supplement","size":8849853,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable2.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/7ce12465e7cfc9bbf018cbfb.xlsx"},{"id":105907576,"identity":"417c2318-bf00-4dcb-b0a3-5484bc568f72","added_by":"auto","created_at":"2026-04-01 10:32:47","extension":"xlsx","order_by":4,"title":"","display":"","copyAsset":false,"role":"supplement","size":352392,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable3.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/1fb9f256244ebd1d6ac3e657.xlsx"},{"id":105907577,"identity":"6e2ab3a6-c9ff-4bad-8541-4c4e64682481","added_by":"auto","created_at":"2026-04-01 10:32:47","extension":"xlsx","order_by":5,"title":"","display":"","copyAsset":false,"role":"supplement","size":266075,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryTable4.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/a2380f6a8ab9b19ed6577c07.xlsx"},{"id":105907639,"identity":"2a6fc61c-fdb3-4513-b1ed-4110bca7e526","added_by":"auto","created_at":"2026-04-01 10:33:11","extension":"txt","order_by":6,"title":"","display":"","copyAsset":false,"role":"supplement","size":1486,"visible":true,"origin":"","legend":"","description":"","filename":"SupplementaryFilesCaptions.txt","url":"https://assets-eu.researchsquare.com/files/rs-9074426/v1/144be8a0c5b7f64662c7f90a.txt"}],"financialInterests":"No competing interests reported.","formattedTitle":"Integrative blood transcriptomics identifies U1-snRNP gene repression and miRNA–mRNA regulatory hubs in pre-clinical Parkinson’s Disease","fulltext":[{"header":"Introduction","content":"\u003cp\u003eParkinson\u0026rsquo;s disease (PD) is the second most prevalent age-related neurodegenerative disorder and represents a growing global health burden driven by population aging (Dorsey et al., 2018). Clinically, PD affects more than 1% of individuals over the age of 60 and more than 10\u0026nbsp;million people worldwide, a figure expected to increase substantially in the coming decades (Su et al., 2025). The disease is characterised by a core triad of motor symptoms \u0026ndash; bradykinesia, resting tremor, and muscular rigidity - often accompanied by postural instability. These manifestations arise primarily from the progressive degeneration of dopaminergic neurons within the substantia nigra pars compacta and the resulting depletion of dopamine within basal ganglia circuits (Jankovic, 2008). Beyond its cardinal motor features, PD is increasingly recognized as a multisystem disorder encompassing a wide range of non-motor symptoms, including olfactory dysfunction, sleep disturbances (e.g. rapid eye movement sleep behavior disorder, RBD), gastrointestinal dysfunction, mood disorders, and cognitive impairment (Chaudhuri and Schapira, 2009). Notably, many of these non-motor symptoms can emerge years or even decades before the onset of overt motor deficits, defining a prodromal phase of PD (Berg et al., 2015). By the time motor symptoms lead to clinical diagnosis, substantial neurodegeneration has already occurred, with approximately 50\u0026ndash;60% of dopaminergic neurons in the substantia nigra loss and striatal dopamine levels reduced of approximately 80% (Bernheimer et al., 1973; Fearnley and Lees, 1991). This delayed clinical recognition highlights a critical need for strategies of early detection. Identifying PD during its pre-clinical or prodromal stages could allow the implementation of neuroprotective or disease-modifying interventions prior to extensive neuronal damage occurring (Berg et al., 2015). However, early diagnosis remains a major challenge, as no definitive biomarkers or diagnostic tests are currently available, and diagnosis continues to rely primarily on clinical features that manifest relatively late in the disease course.\u003c/p\u003e \u003cp\u003eThe pathophysiology of PD is driven by the progressive degeneration of nigrostriatal dopaminergic neurons and the accumulation of misfolded alpha-synuclein (α-syn) protein aggregates in vulnerable brain regions. A pathological hallmark of PD is the presence of intraneuronal inclusions known as Lewy bodies, which are composed predominantly of fibrillar α-syn and are closely associated with neuronal dysfunction and cell death (Spillantini et al., 1997; Surmeier et al., 2017). Misfolded α-syn proteins overwhelm neuronal proteostasis mechanisms, leading to impaired protein degradation and the accumulation of toxic oligomeric species. These aggregates disrupt synaptic transmission, compromise mitochondrial function, and impair lysosomal-autophagic pathways, thereby promoting cellular stress and progressive neurodegeneration (Breydo et al., 2012; Lashuel et al., 2013). Mitochondrial dysfunction and oxidative stress are widely regarded as central and converging drivers to PD pathology. This is underscored by familial PD mutations (e.g. in SNCA, PINK1, GBA, DJ-1, LRRK2 genes), which are involved in mitochondrial quality control, lysosomal function, and reactive oxygen species homeostasis (Pickrell and Youle, 2015; Sobhifar and Brown, 2025). In parallel, chronic neuroinflammation has emerged as a significant pathogenic component of PD. Activated microglia and infiltrating peripheral immune cells release pro-inflammatory cytokines and neurotoxic mediators that further exacerbate dopaminergic neuron vulnerability and loss (Hirsch et al., 2012). Together, α-syn aggregation, mitochondrial failure, oxidative stress, and sustained inflammation form a network of mutually reinforcing processes that ultimately drive progressive neuronal loss in the substantia nigra and widespread dysfunction of neural networks underlying the clinical manifestations of PD (Hirsch et al., 2012). Importantly, PD is increasingly recognized as a multisystem disorder, with pathological changes extending beyond the central nervous system. According to Braak\u0026rsquo;s staging hypothesis, α-syn pathology may originate in peripheral sites such as the olfactory epithelium or enteric nervous system years before the onset of motor symptoms, subsequently propagating to the brain via neural or humoral pathways (Braak et al., 2003a; Braak et al., 2003b). This systemic view of PD pathogenesis supports the concept that peripheral biological alterations accompany central neurodegeneration. In particular, accumulating evidence of immune dysregulation in PD suggests that blood-based molecular changes may reflect ongoing disease processes (T\u0026ouml;nges et al., 2022).\u003c/p\u003e \u003cp\u003eAmong emerging molecular regulators, microRNAs (miRNAs) have garnered significant attention in PD research. miRNAs are small non-coding RNAs that fine-tune gene expression at the post-transcriptional level, and multiple miRNAs have been found dysregulated in PD patient brains as well as in peripheral biofluids (Shaheen et al., 2024). Collectively, these findings provide a strong rationale for investigating blood-based transcriptomic alterations, including both mRNA and miRNA expression profiles, as accessible biomarkers capable of capturing key molecular processes associated with the early, pre-clinical stages of PD.\u003c/p\u003e \u003cp\u003eDespite substantial advances in our understanding of PD, major challenges remain in translating this knowledge into reliable tools for early diagnosis and prognosis. A critical unmet need is the identification of validated biomarkers capable of detecting PD during its preclinical stages, when therapeutic interventions may be most effective and before extensive, irreversible neurodegeneration has occurred. Although neuroimaging modalities and cerebrospinal fluid assays (e.g., measurements of α-synuclein species or neurotransmitter metabolites) have shown promise, their application in early PD remains limited by invasiveness, high cost, limited scalability, and insufficient disease specificity (Cheslow et al., 2021; Posavi et al., 2019).\u003c/p\u003e \u003cp\u003eIn this context, blood-based biomarkers represent an especially attractive alternative due to their minimal invasiveness, suitability for longitudinal monitoring, and broad clinical applicability. However, to date no blood-based diagnostic test for PD has received regulatory approval, and reported candidate biomarkers, including gene expression signature, protein markers, and miRNAs, have not yet been consistently validated in large, well-characterized prospective cohorts. Moreover, transcriptomic and proteomic alterations in peripheral blood are often subtle and susceptible to confounding by demographic, clinical and environmental factors, highlighting the difficulty of distinguishing disease-related signals from background biological and technical variability.\u003c/p\u003e \u003cp\u003eA further gap in the field concerns the limited understanding of how peripheral molecular changes reflect central nervous system pathology, particularly during the earliest stages of PD. The biological mechanisms linking blood-based biomarkers to neurodegenerative processes remain incompletely defined. For example, alterations in leukocyte gene expression may mirror neuroinflammatory activity within the brain, while circulating miRNAs could indicate neuronal injury or α-syn\u0026ndash;related pathology; however, these associations remain poorly characterised. Addressing this knowledge gap will require integrative, systems-level approaches that combine multi-omics data with clinical phenotypes and neurobiological endpoints.\u003c/p\u003e \u003cp\u003eThe present study was designed to address these challenges by systematically investigating blood-based transcriptomic and miRNA alterations across stages of PD, with a particular emphasis on the pre-clinical phase. Specifically, our objectives were to: (\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e) integrate multiple independent blood-based transcriptomic datasets, including both mRNA and miRNA expression profiles, from large and heterogeneous PD cohorts to enhance statistical power and reproducibility; (\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e) identify differentially expressed genes and miRNAs associated with pre-clinical PD and characterize their regulatory interactions; (\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e) perform pathway enrichment analyses to elucidate biological processes perturbed during the early stage of the disease; and (\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e) propose and validate a panel of candidate blood-based biomarkers (comprising both mRNA and miRNA markers) \u0026mdash; using machine learning classification models to assess their potential for distinguishing individuals with preclinical PD from healthy controls.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eData collection and pre-processing\u003c/h2\u003e \u003cp\u003eData were obtained from the Parkinson's Progression Markers Initiative (PPMI) database (Marek et al., 2011) and comprised demographic, clinical, hematological, bulk RNA-sequencing, and small non-coding RNA (sncRNA)-sequencing data derived from blood samples. RNA-sequencing data were quality-filtered by PPMI prior to release, and raw count matrices were downloaded for downstream analyses. To construct a unified cohort, samples with missing values (NA) in any dataset were excluded, resulting in a final dataset of 2,604 samples with complete data coverage.\u003c/p\u003e \u003cp\u003eIndividuals were originally classified according to the Neuronal Synuclein Disease (NSD) stage and were subsequently regrouped into five categories: Healthy, Pre-clinical, Mild, Moderate, and Severe. The Pre-clinical stage was defined by the presence of neuronal α-sync detected in cerebrospinal fluid (CSF) using the seed amplification assay (SAA), together with evidence of dopaminergic dysfunction (e.g., imaging abnormalities), in the absence of overt functional impairment. Hematological variables with isolated missing values were imputed using category-specific means. Patient identifiers and clinical events (longitudinal samplings) were consolidated into unique sample identifiers, formatted as patient ID combined with event code.\u003c/p\u003e \u003cp\u003eEnsembl gene identifiers in the raw count matrices were trimmed to remove version numbers. Gene annotations, including gene symbols and biotypes, were retrieved from the latest GRCh38 Ensembl-based annotation database using the \u003cem\u003eAnnotationHub\u003c/em\u003e package v3.16.1 (Martin Morgan, 2017) and the \u003cem\u003eensembldb\u003c/em\u003e package v2.32.0 (Johannes Rainer, 2017). Gene symbols were validated and corrected for outdated aliases using the \u003cem\u003echeckGeneSymbols()\u003c/em\u003e function from the \u003cem\u003eHGNChelper\u003c/em\u003e package v0.8.15 (Oh et al., 2012). Approved HGNC symbols were assigned where available; unresolved entries were set to NA, and correction frequencies were recorded. A summary annotation table was generated (Supplementary table \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), filtering out unmapped genes. For genes with duplicate HGNC symbols, entries were ordered by mean expression and the highest-expressed transcript was retained, resulting in a final expression matrix with unique HGNC gene symbols as row identifiers.\u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eRNA differential expression analysis\u003c/h3\u003e\n\u003cp\u003eDifferential expression analyses of both mRNAs and miRNAs were performed using the \u003cem\u003eMIRit\u003c/em\u003e package v1.4.2 (Ronchi and Foti, 2023), which implements the statistical framework of edgeR. Prior to differential testing, Multidimensional Scaling (MDS) analysis was conducted to assess sample variability and identify potential outliers. Low-abundance features were subsequently filtered using stringent criteria: min.count\u0026thinsp;=\u0026thinsp;5, min.total.count\u0026thinsp;=\u0026thinsp;10, large.n\u0026thinsp;=\u0026thinsp;10, min.prop\u0026thinsp;=\u0026thinsp;0.3. Differential expression was assessed using a generalized linear model (GLM) quasi-likelihood F-test. A robust design matrix was constructed with disease stage as the primary variable of interest, while adjusting for potential confounding covariates including age, sex, body mass index (BMI), white blood cell (WBC) count, platelet count, and neutrophil-to-lymphocyte ratio (NLR).\u003c/p\u003e \u003cp\u003ePairwise contrasts were performed by comparing each pathological stage against the Healthy control group. Features were considered significantly differentially expressed based on a nominal p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and an absolute fold-change threshold of 1.5.\u003c/p\u003e\n\u003ch3\u003eIdentification of miRNA–mRNA regulatory interactions\u003c/h3\u003e\n\u003cp\u003ePotential regulatory targets of the differentially expressed miRNAs (DEMs) were identified through an integrative approach combining target prediction and expression correlation analysis. Putative targets for each DEM were retrieved using the \u003cem\u003egetTargets()\u003c/em\u003e function from the \u003cem\u003eMIRit\u003c/em\u003e package.\u003c/p\u003e \u003cp\u003eTo assess the functional relevance of miRNA-mRNA interactions, Spearman correlation analyses were performed between miRNA and mRNA expression profiles of the identified DEGs and DEMs. Prior to correlation analysis, batch correction was applied to mitigate the influence of confounding technical and biological factors using a Weighted Surrogate Variable Analysis. Specifically, patient-specific effects and sex were adjusted for, while preserving variability associated with relevant biological covariates, including age, BMI, WBC, platelet count, and NLR. Additionally, WSVA was used to estimate and remove two latent surrogate variables, with features weighted by their standard deviation to more effectively capture systematic sources of unwanted variation.\u003c/p\u003e \u003cp\u003eFollowing batch correction, miRNA and gene expression datasets were integrated using the \u003cem\u003emirnaIntegration()\u003c/em\u003e function from \u003cem\u003eMIRit\u003c/em\u003e. This integration was performed separately for each disease stage, linking batch-corrected DEM and DEG profiles to identify statistically significant negative correlations consistent with miRNA-mediated post-transcriptional repression. Finally, relevant correlation networks were visualized using Fluorish (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://flourish.studio/\u003c/span\u003e\u003cspan address=\"https://flourish.studio/\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\n\u003ch3\u003eFunctional enrichment analysis\u003c/h3\u003e\n\u003cp\u003eTo elucidate the biological relevance of the identified molecular signatures, functional enrichment analyses were performed with \u003cem\u003eshinyGO\u003c/em\u003e v0.85 (Ge et al., 2020) on (i) the DEGs identified in the Pre-clinical stage and (ii) the subset of target DEGs showing strong negative correlations with upregulated miRNAs. An Over-Representation Analysis (ORA) was performed to identify significantly enriched biological processes (BP) by querying the Gene Ontology (GO) database for \u003cem\u003eHomo sapiens\u003c/em\u003e. Only BPs supported by at least three associated genes were retained for downstream interpretation.\u003c/p\u003e\n\u003ch3\u003eMachine learning framework for pre-clinical stage classification\u003c/h3\u003e\n\u003cp\u003eTo assess the predictive potential of the identified miRNA biomarkers, a machine learning-based classification framework was implemented to distinguish subjects in the Pre-clinical and Healthy stages. All analyses were performed in R using the \u003cem\u003ecaret\u003c/em\u003e package v7.0.1 (Kuhn, 2008). A dedicated modeling dataset was generated by filtering the original miRNA count matrix to retain only significant differentially expressed miRNAs (DEMs) identified in the \"Pre-clinical vs. Healthy\" contrast (adjusted p-value\u0026thinsp;\u0026lt;\u0026thinsp;0.05 and |logFC| \u0026gt; 0.58). This resulting expression matrix was then merged with clinical metadata to incorporate disease stage labels.\u003c/p\u003e \u003cp\u003eThe dataset comprised 1,219 samples (Healthy, n\u0026thinsp;=\u0026thinsp;658; Pre-clinical, n\u0026thinsp;=\u0026thinsp;561). Using a stratified sampling strategy to preserve disease-stage class proportions, the data was divided into a training set (70%, n\u0026thinsp;=\u0026thinsp;854; Healthy, n\u0026thinsp;=\u0026thinsp;461; Pre-clinical, n\u0026thinsp;=\u0026thinsp;393) and an independent test set (30%, n\u0026thinsp;=\u0026thinsp;365; Healthy, n\u0026thinsp;=\u0026thinsp;197; Pre-clinical, n\u0026thinsp;=\u0026thinsp;168).\u003c/p\u003e \u003cp\u003eEight supervised learning algorithms were trained and evaluated: Random Forest (RF), Support Vector Machine with Radial Basis Function kernel (SVM-Radial), Generalized Linear Model with Elastic Net regularized logistic regression (GLMNet), Gradient Boosting machine (GBM), k-Nearest Neighbors (KNN), Naive Bayes (NB), Linear Discriminant Analysis (LDA) and Neural Network (NN). Model training employed 10-fold cross-validation repeated five times to ensure robust performance and to minimize overfitting.\u003c/p\u003e \u003cp\u003eModel performance was evaluated on the independent test set using confusion matrices and multiple evaluation metrics, including Accuracy, Kappa, Sensitivity, Specificity, Precision, F1 Score, and Balanced Accuracy. The Receiver Operating Characteristic (ROC) curves were estimated, and Area Under the Curve (AUC) values were calculated and visualized with \u003cem\u003epROC\u003c/em\u003e v1.19 (Robin et al., 2010). In addition, feature importance measures were extracted from caret\u0026rsquo;s available algorithm to identify miRNAs contributing most strongly to the classification task. Finally, to assess potential overfitting, the models were also evaluated on the training data, and the AUC was computed. Overfitting was quantified as the difference between the training and test AUCs (the AUC gap).\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003eCohort characteristics and data exploration\u003c/h2\u003e \u003cp\u003eFollowing quality control, filtering, and cross-platform harmonization of transcriptomic, miRNA, clinical, and hematological datasets, we retained only samples with complete multi-omics coverage for downstream analyses (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e). The final cohort comprised 2,604 individuals, including 658 healthy controls and 1,946 PD-related samples. PD cases were stratified into pre-clinical (n\u0026thinsp;=\u0026thinsp;561), mild (n\u0026thinsp;=\u0026thinsp;197), moderate (n\u0026thinsp;=\u0026thinsp;910), and severe (n\u0026thinsp;=\u0026thinsp;278) stages after regrouping (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003eA). Across all disease stages, the cohort exhibited a consistent male predominance (~\u0026thinsp;60%), consistent with the known PD epidemiology (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eB) along with a predominance of participants aged 60\u0026ndash;70 years (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eC). The mean age increased from healthy controls to severe PD stages. Hematological parameters, including white blood cell count, platelet count, neutrophil count, lymphocyte count, and the neutrophil-to-lymphocyte ratio, demonstrated stage-dependent variability, suggesting systematic alterations associated with disease progression.\u003c/p\u003e \u003cp\u003eImportantly, the same set of samples were used for both RNA-seq and miRNA-seq analyses, resulting in a fully integrated cohort and eliminating potential biases arising from partial data overlap.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eDynamic patterns of mRNA and miRNA expression across Parkinson’s disease progression\u003c/h3\u003e\n\u003cp\u003eDifferential expression analysis revealed extensive transcriptional dysregulation across PD stages relative to healthy controls (Supplementary table \u003cspan refid=\"MOESM2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). MDS analysis on mRNA expression profiles identified \u0026lsquo;sex\u0026rsquo; as the primary source of variance in the dataset (Supplementary Fig.\u0026nbsp;1). Among disease stages,the pre-clinical group exhibited the most pronounced transcriptional alterations, with 7,427 DEGs. Substantial transcriptome reprogramming was also observed in the severe stage, where 3,393 DEGs were identified. In contrast, the mild and moderate stages showed markedly fewer changes, with only 29 and 33 DEGs detected, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn the pre-clinical stage, more than two-thirds of DEGs were downregulated, indicating widespread repression of transcriptional programs prior to the onset of overt clinical symptoms. A similar bias toward downregulation was observed in the severe stage. Notably, only 1,531 DEGs were common between the pre-clinical and severe stages, while other stage comparisons shared a maximum of just 8 DEGs, suggesting that transcriptional alterations are largely stage-specific (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). Collectively, these findings highlight a U-shaped pattern of transcriptional disruption, with maximal perturbation occurring at the pre-clinical stage and re-emerging at advanced disease stages.\u003c/p\u003e \u003cp\u003eMiRNA expression analysis revealed a stage-dependent pattern that paralleled the mRNA dysregulation. Consistent with the mRNA results, the pre-clinical stage showed the largest number of differentially expressed miRNAs (DEmiRNAs), with 157 identified relative to healthy controls (Supplementary Table\u0026nbsp;3). The severe stage again showed notable dysregulation with 35 DEmiRNAs detected, whereas the mild and moderate stages exhibited only 13 and 7 DEmiRNAs, respectively (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eIn the pre-clinical stage, the majority of DEmiRNAs were upregulated, consistent with a potential repressive effect on target mRNA expression. This pattern aligns with the predominance of downregulated mRNAs observed at the same stage and is consistent with a regulatory relationship between miRNA overexpression and mRNA repression.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eEarly and late disease miRNA\u0026ndash;mRNA interaction patterns\u003c/h2\u003e \u003cp\u003eIntegration of DEGs and DEMs revealed a clear stage-dependent pattern of miRNA\u0026ndash;mRNA associations. No significant correlations were detected in the Mild and Moderate stages. In contrast, multiple miRNA\u0026ndash;mRNA interactions emerged in the \u003cem\u003epreclinical\u003c/em\u003e and \u003cem\u003esevere\u003c/em\u003e stages, with the strongest signal observed in the \u003cem\u003epreclinical\u003c/em\u003e stage.\u003c/p\u003e \u003cp\u003eTo focus on the most biologically relevant interactions, analyses were restricted to strong negative correlations (Spearman\u0026rsquo;s ρ \u0026lt; \u0026minus;0.8). Under this stringent threshold, 22 miRNA\u0026ndash;mRNA correlations were identified in the Severe stage, all driven by only two miRNAs, hsa-miR-412-3p and hsa-miR-1262, suggesting a highly centralized regulatory architecture in advanced disease. In the Pre-clinical stage, 252 strong negative correlations were detected, only six miRNAs (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e), consistent with an early regulatory network dominated by a limited set of miRNA regulators. Among these, four miRNAs, namely hsa-miR-1299, hsa-miR-34a-5p, hsa-miR-3120-5p, and hsa-miR-412-3p, emerged as major hubs, accounting for the largest number of anti-correlated target transcripts. Notably, hsa-miR-34a-5p also ranked among the most upregulated miRNAs, supporting concordance between miRNA overexpression and repression of target genes. In addition, several target genes exhibited negative correlations with two or three distinct miRNAs, suggesting coordinated post-transcriptional regulation during the pre-clinical phase.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eFunctional characterization\u003c/h2\u003e \u003cp\u003eFunctional enrichment analysis of pre-clinical DEGs revealed a heterogeneous biological landscape, encompassing epithelial differentiation and keratin-related processes, chemosensory processes, and broader nervous-system-associated functions (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eA). This diversity suggests that early molecular alterations affect multiple biological systems rather than being confined solely to neuronal pathways. Consistent with this observation, examination of the top 20 pre-clinical DEGs (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e) showed that the strongest expression changes were dominated by U1 small nuclear RNA (snRNA) genes. These transcripts are not captured by standard enrichment analyses due to the absence of curated functional annotations U1 snRNA genes encode components of the U1 small nuclear ribonucleoprotein (snRNP) complex, which plays a central role in pre-mRNA splicing and transcriptional regulation. Notably, the long non-coding RNA LINC03126, also highly ranked among preclinical DEGs, is predicted to be associated with the U1 snRNP complex, further supporting a potential role for RNA-processing mechanisms in early disease stages. In contrast, functional enrichment restricted to genes involved in the miRNA\u0026ndash;mRNA interactions with strong negative correlations, yielded a markedly more neurocentric signature. Enriched GO terms converged predominantly on processes related neurodevelopment and neuronal communication, including nervous system development, neurogenesis, neuronal differentiation, cell morphogenesis associated with neuronal fate, and synapse-related signaling pathways (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003eB). This shift toward neuronal specificity underscores the additional biological resolution provided by integrating miRNA regulation, uncovering molecular programs more directly linked to early Parkinson\u0026rsquo;s disease pathophysiology.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eTop 20 differentially expressed genes in the pre-clinical stage of Parkinson's disease ranked by magnitude of expression change. The list includes U1 small nuclear RNAs (snRNAs) and long non-coding RNAs (lncRNAs) among the top candidates.\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGene ID\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGene type\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eLog\u003csub\u003e2\u003c/sub\u003e(fold change)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eP-value\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAdjusted p-value\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLINC03126\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003elncRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-5.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9.80E-73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.83E-68\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-83P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-3.07\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.20E-70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.73E-66\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-28P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.94E-51\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.02E-47\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-88P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.67\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e2.09E-52\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e4.58E-49\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-67P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.61\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5.61E-45\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e6.53E-42\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-18P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.05E-33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e3.45E-31\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-27P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.43\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.39E-34\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.33E-32\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-11P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.41\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.54E-36\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.14E-33\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNVU1-7\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.27\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.62E-30\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e9.79E-28\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-3\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.26\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.36E-35\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.95E-33\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU2-36P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.22E-47\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.01E-43\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.09\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.60E-39\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.08E-36\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNVU1-29\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-2.06\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.47E-57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.49E-54\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU6ATAC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-1.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9.91E-54\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.46E-50\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-89P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-1.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e6.62E-21\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e5.44E-19\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU4-2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-1.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9.62E-46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.24E-42\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU4-1\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-1.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e9.72E-33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.81E-30\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNVU1-31\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-1.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3.80E-33\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e1.11E-30\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRNU1-148P\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003esnRNA\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e-1.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.59E-25\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e2.13E-23\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eMachine learning classification of the pre-clinical stage\u003c/h2\u003e \u003cp\u003eAll eight binary classification models achieved predicted performance above chance on the test set, with area under the ROC curve (AUC) values ranging from 0.791 to 0.886. The highest overall performance was observed for Naive Bayes (Accuracy\u0026thinsp;=\u0026thinsp;0.814, F1\u0026thinsp;=\u0026thinsp;0.800, Balanced Accuracy\u0026thinsp;=\u0026thinsp;0.813, AUC\u0026thinsp;=\u0026thinsp;0.886), closely followed by GLMNet (Accuracy\u0026thinsp;=\u0026thinsp;0.800, F1\u0026thinsp;=\u0026thinsp;0.786, Balanced Accuracy\u0026thinsp;=\u0026thinsp;0.800, AUC\u0026thinsp;=\u0026thinsp;0.883) and SVM-Radial (Accuracy\u0026thinsp;=\u0026thinsp;0.789, F1\u0026thinsp;=\u0026thinsp;0.763, Balanced Accuracy\u0026thinsp;=\u0026thinsp;0.785, AUC\u0026thinsp;=\u0026thinsp;0.860). Gradient Boosting and the Neural Network models showed intermediate performance, while k-NN and LDA displayed lower AUC values. Finally, Naive Bayes and, particularly, GLMnet demonstrated the lowest degree of overfitting among the evaluated models (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e \u003cp\u003eTo explore mode interpretability, feature-importance analyses were performed for models that support such estimates (Random Forest, Gradient Boosting, and GLMNet). These analyses consistently highlighted a small subset of DEMs as key contributors to classification. Notably, hsa-miR-4433a-5p and hsa-miR-34a-5p ranked among the top three features across all three models. Importantly, these miRNAs were also identified as major regulatory hubs in the miRNA-mRNA interaction analysis (Supplementary Table\u0026nbsp;4), underscoring their potential biological relevance in the pre-clinical stage of Parkinson\u0026rsquo;s disease.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003ePerformance of eight binary classification models on the test set in distinguishing pre-clinical from healthy subjects using differentially expressed miRNAs as features. Metrics include accuracy, sensitivity, specificity, and area under the ROC curve (AUC).\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"13\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c7\" colnum=\"7\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c8\" colnum=\"8\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c9\" colnum=\"9\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c10\" colnum=\"10\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c11\" colnum=\"11\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c12\" colnum=\"12\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c13\" colnum=\"13\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eModel\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eKappa\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eSensitivity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eSpecificity\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c6\"\u003e \u003cp\u003ePositive\u003c/p\u003e \u003cp\u003ePredictive\u003c/p\u003e \u003cp\u003eValue\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c7\"\u003e \u003cp\u003eNegative\u003c/p\u003e \u003cp\u003ePredictive\u003c/p\u003e \u003cp\u003eValue\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c8\"\u003e \u003cp\u003ePrecision\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c9\"\u003e \u003cp\u003eRecall\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c10\"\u003e \u003cp\u003eF1\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c11\"\u003e \u003cp\u003eBalanced\u003c/p\u003e \u003cp\u003eAccuracy\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c12\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c13\"\u003e \u003cp\u003eAUC\u003c/p\u003e \u003cp\u003eGap\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNaive Bayes\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.63\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.89\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e0.02\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGLMNet\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.60\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.82\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.88\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSVM\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.86\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e0.11\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGradient Boosting\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e0.1\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eNeural Network\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.56\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.76\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.85\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e0.15\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eRandom Forest\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.49\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.71\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.77\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.74\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.75\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.83\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e0.16\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003ek-NN\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.46\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.68\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.78\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.73\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.80\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e0.02\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eLinear Discriminant Analysis\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e0.69\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e \u003cp\u003e0.40\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e \u003cp\u003e0.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e \u003cp\u003e0.57\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e \u003cp\u003e0.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c7\"\u003e \u003cp\u003e0.81\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c8\"\u003e \u003cp\u003e0.62\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c9\"\u003e \u003cp\u003e0.84\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c10\"\u003e \u003cp\u003e0.72\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c11\"\u003e \u003cp\u003e0.70\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c12\"\u003e \u003cp\u003e0.79\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c13\"\u003e \u003cp\u003e-0.01\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we performed an integrative analysis of whole-blood mRNA and miRNA expression profiles across multiple stages of PD to identify early molecular signatures and regulatory networks associated with the pre-clinical phase of the disorder. This large-scale in silico analysis of publicly available multi-omics datasets underscores the power of integrative computational approaches to elucidate early molecular mechanisms in neurodegenerative diseases. By systematically mining whole-blood mRNA and miRNA expression profiles from ~\u0026thinsp;2,600 individuals spanning healthy controls, pre-clinical, and clinically manifest PD stages, we identified stage-specific molecular signatures that precede overt motor symptoms. Our findings build upon earlier blood-based transcriptomic studies demonstrating the possibility of identifying molecular markers of early PD in peripheral blood (Scherzer et al., 2007) as well as studies characterizing transcriptional alterations in PD blood mononuclear cells, including in genetic forms such as LRRK2 mutation carriers (Mutez et al., 2011). The scale and heterogeneity of the dataset of our study enabled robust detection of subtle yet biologically coherent regulatory patterns that may not be apparent in smaller cohorts, consistent with previous large-scale blood-based transcriptomic and multi-omics integration studies demonstrating that harmonized, high-dimensional datasets increase sensitivity for detecting disease-associated RNA signatures and improve predictive modeling performance in PD (Dong et al., 2025; Irmady et al., 2023; Tng et al., 2024) A key aspect of our study is that it focuses on the pre-clinical phase of PD, a critical but poorly characterized window during which neuropathological processes are likely already underway despite the absence of classical motor manifestations. Molecular alterations detected at this stage may reflect early systemic responses to incipient neurodegeneration, compensatory regulatory mechanisms, or primary pathogenic drivers. By exploiting integrated mRNA\u0026ndash;miRNA network analyses, not only we were able to catalogue differentially expressed transcripts but we could also infer regulatory relationships that may actively shape early disease biology. Together, these results demonstrate how computational multi-omics can move beyond descriptive transcriptomics toward mechanistic insight, offering a scalable strategy to identify candidate biomarkers and molecular pathways that may enable earlier diagnosis, improved assessment of risk stratification, and ultimately preventive intervention in PD.\u003c/p\u003e \u003cp\u003eOne of the most striking findings of the present study was to observe a robust downregulation of multiple U1 small nuclear ribonucleoprotein (U1-snRNP) complex genes in the pre-clinical stage of PD. U1-snRNP is a core component of the spliceosome and plays a key role in the regulation of gene expression (Campagne, 2024). It initiates spliceosome assembly by recognizing the 5\u0026prime; splice site through base pairing between U1 snRNA and pre-mRNA (Mount et al., 1983), a step that represents a key rate-limiting stage in regulated splicing.\u003c/p\u003e \u003cp\u003eBeyond splice-site recognition, U1 also contributes to proper 3\u0026prime;-end mRNA formation by suppressing premature cleavage and polyadenylation and coordinating productive transcription with RNA polymerase II (Almada et al., 2013; Berg et al., 2012). Through these dual functions, U1 maintains transcript integrity and helps preserve global transcriptional stability.\u003c/p\u003e \u003cp\u003eWithin this framework, the early downregulation of U1-related genes observed in our dataset may reflect impaired RNA-processing capacity preceding overt clinical symptoms. Even modest reductions in U1 availability could indeed destabilize splicing fidelity and transcript length control, leading to subtle but widespread transcriptomic remodeling. Spliceosomal dysfunction is increasingly recognized as a driver of neurodegeneration (Nikom and Zheng, 2023), whereas the disruption of U1-snRNP biogenesis has been linked to RNA metabolism defects in disorders such as spinal muscular atrophy, ontocerebellar hypoplasia type 7, and FUS-associated amyotrophic lateral sclerosis (Campagne, 2024). In Alzheimer\u0026rsquo;s disease (AD), U1 snRNP components redistribute to cytoplasmic aggregates and are associated with aberrant RNA processing and increased premature cleavage/polyadenylation (Zhu et al., 2020). Furthermore, U1 snRNP aggregation has been directly demonstrated in AD cases with autosomal dominant mutations and trisomy 21, reinforcing the concept that spliceosomal instability may represent a shared vulnerability across neurodegenerative conditions (Hales et al., 2014)\u003c/p\u003e \u003cp\u003eAlthough similar U1 aggregates have not been detected in PD brains (Bai et al., 2013), our findings suggest that PD may involve a more subtle, functional reduction in spliceosomal capacity rather than overt protein mislocalization. Importantly, mis-splicing of PD-relevant genes has been documented. In cellular models of PARK7 (DJ-1) mutation, U1-dependent splicing defects lead to exon 3 exclusion and loss of the DJ-1 protein, resulting in mitochondrial dysfunction, whereas restoration of U1-snRNP levels rescues normal splicing and protein expression (Boussaad et al., 2020). These observations provide mechanistic plausibility that early U1-snRNP downregulation could reduce expression of neuroprotective factors such as DJ-1 or broadly perturb mitochondrial and stress-response pathways. Additional evidence supporting the link between splicing alterations and PD comes from leukocyte small RNA sequencing studies showing that deep brain stimulation modulates splicing-related signatures and that peripheral RNA profiles can classify brain-region transcriptomes (Soreq et al., 2013), further indicating systemic RNA-processing involvement in PD.\u003c/p\u003e \u003cp\u003eWhether the observed blood U1-snRNP suppression reflects a systemic response to early neurodegeneration or a primary regulatory disturbance remains unclear. It may arise from disease-related epigenetic remodeling of housekeeping gene networks or from immune-cell adaptations to incipient pathology. Regardless of origin, our data position spliceosomal regulation as a potential early molecular vulnerability in PD. Future studies should validate these findings in independent cohorts and determine whether therapeutic modulation of RNA splicing can mitigate disease progression.\u003c/p\u003e \u003cp\u003eAnother key finding of our study is the identification of a U-shaped trajectory in blood transcriptomic alterations across PD progression. Specifically, we observed a widespread downregulation of mRNAs in the pre-clinical stage, followed by a marked reduction in the number of DEGs during the mild and moderate clinical stages, and finally a renewed and pronounced transcriptional downregulation in severe disease.\u003c/p\u003e \u003cp\u003eThis non-linear dynamic suggests that peripheral transcriptomic remodelling in PD does not progress in a simple monotonic fashion, but instead follows a stage-dependent pattern of molecular engagement and disengagement. Although explicit descriptions of U-shaped dynamics in blood transcriptomics remain limited, multiple studies support the concept of dynamic gene expression changes across PD progression. Whole-blood and PBMC transcriptomic analyses have consistently identified alterations in immune and stress-response pathways, with reproducible DEG signatures across cohorts, supporting the presence of stage-specific peripheral molecular signals (Navarrete et al., 2025).\u003c/p\u003e \u003cp\u003eThe widespread downregulation observed in the pre-clinical phase may reflect an early systemic response to emerging neuropathology, consistent with accumulating evidence that peripheral immune dysregulation and chronic inflammatory signaling are integral components of PD pathophysiology, even at early disease stages (Tansey et al., 2022). Such suppression could be driven by oxidative stress, inflammation, or shifts in immune cell composition and activation states, in line with single-cell studies demonstrating peripheral immune reprogramming in PD (Moquin-Beaudry et al., 2025).\u003c/p\u003e \u003cp\u003eThe relative reduction in transcriptomic perturbation during mild and moderate stages may indicate compensatory homeostatic mechanisms, immunological adaptation, or treatment-related effects that transiently reduce detectable differential expression. Longitudinal studies further support the notion that blood gene expression profiles shift dynamically over time rather than progressing linearly (K\u0026otilde;ks, 2025).\u003c/p\u003e \u003cp\u003eIn severe PD, the observed re-emergence of broad downregulation may reflect failure of compensatory processes and the onset of systemic dysregulation, potentially driven by chronic inflammation, sustained metabolic stress, and advanced neurodegeneration. Similar stage-dependent transcriptional shifts have also been observed in brain tissue, reinforcing the concept that PD molecular pathology evolves dynamically across disease stages (Cappelletti et al., 2023).\u003c/p\u003e \u003cp\u003eThe integrative miRNA\u0026ndash;mRNA network analysis provided important mechanistic insight into pre-clinical PD by identifying regulatory hubs rather than isolated differentially expressed transcripts. We detected several highly connected miRNAs, most prominentlymiR-34a-5p, miR-1299, miR-3120-5p, miR-412-3p, and miR-4433a-5p, that were markedly upregulated in the pre-clinical stage and inversely correlated with numerous downregulated mRNA targets. These anti-correlated patterns suggest that a small number of miRNAs may act as upstream regulators driving the widespread gene repression observed early in disease.\u003c/p\u003e \u003cp\u003eThis regulatory architecture aligns with growing evidence that miRNA-mRNA networks contribute to PD pathogenesis, influencing mitochondrial quality control, inflammatory signaling, and neuronal survival. Multiple independent studies have identified dysregulated circulating miRNAs in PD blood and plasma (Khoo et al., 2012; Vallelunga et al., 2014; Shaheen et. al, 2024). For example, miR-34a-5p has been shown to suppress PINK1 expression and inhibit PINK1-mediated mitophagy, a pathway critical for mitochondrial function and neuronal health (Tai et al., 2021). Additionally, circulating miRNAs are increasingly recognized as candidate blood-based biomarkers for PD, with reproducible expression changes reported across multiple cohorts and sample types (Zhang et al., 2024).\u003c/p\u003e \u003cp\u003eThe convergence between network centrality and prior functional evidence strengthens the biological plausibility of these miRNAs may act as early molecular drivers rather than passive biomarkers. Collectively, these findings support a model in which coordinated upregulation of a limited set of miRNAs contributes to early transcriptomic remodelling in PD, potentially amplifying subtle regulatory changes into broader pathway-level dysregulation.\u003c/p\u003e \u003cp\u003eMiR-34a-5p represents a particularly compelling example among the identified hub miRNAs.. In our pre-clinical PD dataset, it was among the most strongly overexpressed miRNAs and exhibited extensive connectivity within the regulatory network, targeting a broad array of downregulated mRNAs. This high degree of centrality suggests that miR-34a-5p may play a significant role in early transcriptomic remodeling. Our findings are consistent with independent studies reporting elevated miR-34a levels in PD.. Notably, miR-34a has been shown to be significantly increased in plasma-derived extracellular vesicles (EVs) from PD patients, where it demonstrated moderate discriminative performance (AUC\u0026thinsp;\u0026asymp;\u0026thinsp;0.74 in purified EVs) (Grossi et al., 2020).\u003c/p\u003e \u003cp\u003eMechanistically, miR-34a-5p has been directly linked to mitochondrial dysfunction through the negative regulation of PINK1, a key mediator of mitophagy (Tai et al., 2021). Suppression of PINK1-dependent mitochondrial quality control represents a central pathogenic axis in PD, thereby linking miR-34a-5p activity to one of the most well-established molecular pathways in the disease. The convergence between our network-based identification of miR-34a-5p as a highly connected regulatory hub and its experimentally supported role in mitochondrial homeostasis further strengthens the biological plausibility of its involvement in the prodromal phase of PD.\u003c/p\u003e \u003cp\u003eAs for miR-1299 and miR-3120-5p, direct evidence in PD patient biofluids remains limited. However, data from biologically relevant central nervous system models support their dysregulation. In midbrain organoids derived from induced pluripotent stem cells (iPSCs) of sporadic PD patients, both miR-1299 and miR-3120-5p exhibited robust and reproducible upregulation during neuronal differentiation and maturation (Valente et al., 2025).\u003c/p\u003e \u003cp\u003eThis convergence between peripheral blood and patient-derived midbrain models strengthens the plausibility that these miRNAs participate in disease-relevant regulatory programs rather than representing incidental findings.\u003c/p\u003e \u003cp\u003eNotably, miR-3120-5p has clear mechanistic relevance to neural function, as it can modulate Hsc70/auxilin-mediated uncoating of clathrin-coated vesicles, a process essential for synaptic vesicle recycling and efficient neurotransmission (Scott et al., 2012). Such functions link this miRNA to pathways directly implicated in synaptic dysfunction, a hallmark of early PD pathophysiology.\u003c/p\u003e \u003cp\u003eAs for miR-4433a-5p, emerging clinical evidence intersects more directly with the prodromal continuum of synucleinopathies. In plasma EVs, miR-4433a-5p levels increase progressively along the trajectory from healthy individuals to idiopathic REM sleep behavior disorder (iRBD) to PD, with iRBD representing a recognized prodromal stage (Li et al., 2024). This stage-dependent elevation closely mirrors our observation of upregulation in pre-clinical PD. Additionally, \u003cem\u003ein silico\u003c/em\u003e analyses of human cortical datasets have identified a closely related family member (hsa-miR-4443, also referred to as \u0026ldquo;microRNA-4433\u0026rdquo;) as co-upregulated in PD and enriched within immune and synaptic regulatory networks (Ge et al., 2022). Together, these findings suggest that the miR-4433 family may contribute to early neuroinflammatory processes and synaptic regulatory processes relevant to disease progression. In contrast, miR-412-3p currently lacks direct evidence linking it to PD. Existing literature, however, associates this miRNA with other neurodegenerative conditions. In particular, increased expression has been reported in cerebellar tissue from murine models of spinocerebellar ataxia type 2 (SCA2), and EV-related signals have been described in Huntington\u0026rsquo;s disease contexts, although these observations require cautious interpretation (Beatriz et al., 2023; Paul et al., 2024). In view of these literature reports, miR-412-3p should therefore be considered a relatively novel and exploratory candidate in PD research. Its identification as a network hub in the present analysis suggests the need of specific functional investigations to clarify whether it plays a previously unrecognized role in peripheral or central PD-related pathways.\u003c/p\u003e \u003cp\u003eFrom a translational perspective, our findings provide concrete leads for the development of blood-based biomarkers capable of detecting PD at a pre-clinical stage. Using differentially expressed miRNAs as features, we demonstrated that supervised machine learning models can reliably distinguish pre-clinical PD individuals from healthy controls with strong predictive performance. Across multiple algorithms, including regularized regression and Naive Bayes classifiers, area-under-ROC-curve values consistently ranged between ~\u0026thinsp;0.85 and 0.90, underscoring the substantial diagnostic potential of circulating miRNA profiles.\u003c/p\u003e \u003cp\u003eImportantly, the most informative features driving classifier performance, such as miR-34a-5p and miR-4433a-5p, were also identified as highly connected hubs in our regulatory network analysis. This convergence between predictive modeling and network centrality strengthens confidence that these miRNAs represent biologically meaningful regulators rather than purely statistical markers within early PD molecular circuitry. Notably, miR-34a-5p alone showed strong classification capacity, consistent with prior reports linking it to PD pathophysiology (Grossi et al., 2020). When combined with additional hub miRNAs, it contributed to a multivariate signature capable of robust discrimination of the pre-clinical PD state. Together, these results highlight the promise of integrating systems biology with machine learning approaches to advance toward clinically actionable early-detection strategies.\u003c/p\u003e \u003cp\u003eDespite the strength of the multi-omics integration and the size of the analyzed cohorts, several limitations should be acknowledged. First, the regulatory relationships inferred in our miRNA\u0026ndash;mRNA networks are based primarily on correlation and computational target prediction. Although we prioritized anti-correlated pairs and incorporated established miRNA targeting rules to increase biological plausibility, some predicted interactions may represent false positives. Functional validation is therefore essential. Experimental approaches such as luciferase reporter assays to confirm 3\u0026prime;UTR binding, miRNA overexpression or inhibition in cellular systems, and in vivo perturbation in PD-relevant animal models will be necessary to determine whether altering hub miRNAs produces measurable effects on mitochondrial function, splicing regulation, or neurodegenerative phenotypes.\u003c/p\u003e \u003cp\u003eSecond, although our blood-based findings hold promise for diagnostic applications, they also raise important mechanistic questions\u0026mdash;namely, whether these peripheral molecular alterations reflect pathological processes occurring in the central nervous system or instead represent parallel systemic responses independent of primary brain pathology.\u003c/p\u003e \u003cp\u003eRecent integrative studies suggest that circulating non-coding RNAs may mirror brain regulatory networks through competing endogenous RNA interactions, supporting the possibility of cross-tissue molecular coupling (Chun and Kim, 2024).\u003c/p\u003e \u003cp\u003eTo address this question, future studies should examine whether key markers identified here, such as U1-snRNP genes or other strongly downregulated transcripts,are similarly altered in prodromal or early-stage PD brain tissue. Cross-tissue analyses integrating blood, post-mortem brain samples, and induced pluripotent stem cells\u0026ndash;derived neuronal models could help determine whether these signatures represent surrogate biomarkers or active participants in PD pathogenesis. Longitudinal studies will also be critical to establish temporal dynamics and assess whether these molecular alterations predict disease conversion or progression.\u003c/p\u003e \u003cp\u003eIn conclusion this study demonstrates how integrative analysis of blood-based miRNA\u0026ndash;mRNA regulatory networks can uncover candidate biomarkers and pathways associated with the pre-clinical phase of PD. By exploiting large-scale multi-omics data, we identified a coordinated pattern of transcriptomic remodeling characterized by widespread downregulation, most notably involving U1-snRNP spliceosomal components, together with the upregulation of specific hub miRNAs that may act as upstream regulatory drivers.\u003c/p\u003e \u003cp\u003eThese findings suggest that systemic RNA-processing alterations and miRNA-mediated repression may emerge early in the disease course, potentially before the onset of overt motor symptoms.\u003c/p\u003e \u003cp\u003eWhile these results provide biologically plausible insights into prodromal PD, further work is required to strengthen and extend these observations. Integration of additional omics layers, such as proteomics, epigenomics, or single-cell transcriptomics, could further refine the regulatory landscape and clarify causal relationships. Independent cohort validation and prospective longitudinal studies will be essential to determine the robustness, temporal stability, and predictive value of the identified blood signatures.\u003c/p\u003e \u003cp\u003e \u003cspan type=\"SmallCaps\" class=\"SmallCaps\" name=\"Emphasis\"\u003eIn parallel, mechanistic investigations are needed to establish whether modulation of key miRNAs or spliceosomal components can influence PD-relevant cellular phenotypes or modify neurodegenerative trajectories. Together, these advances may enable the translation of these molecular signatures into improved strategies for early diagnosis while also opening new directions for investigating the underlying molecular mechanisms in PD pathology.\u003c/span\u003e \u003c/p\u003e "},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eCode availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe code for this study is available in GitHub and can be accessed via https://github.com/dangelodavid/BLOMOPARKED\u003c/p\u003e\u003cp\u003e \u003ch2\u003eCompeting Interests\u003c/h2\u003e \u003cp\u003eThe authors declare that they have no financial or non-financial competing interests\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eFundings\u003c/h2\u003e \u003cp\u003eThis research was conducted without specific funding from any public, commercial, or not-for-profit funding agencies.\u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eD.D.A. conceptualized the study, performed the formal analysis, and wrote the original draft of the manuscript. N.D. and A.D.S. reviewed and edited the original draft. All authors read and approved the final manuscript.\u003c/p\u003e\u003ch2\u003eAcknowledgement\u003c/h2\u003e\u003cp\u003ePPMI \u0026ndash; a public-private partnership \u0026ndash; is funded by the Michael J. Fox Foundation for Parkinson\u0026rsquo;s Research, and funding partners; including 4D Pharma, Abbvie, AcureX, Allergan, Amathus Therapeutics, Aligning Science Across Parkinson\u0026rsquo;s, AskBio, Avid Radiopharmaceuticals, BIAL, Biogen, Biohaven, BioLegend, BlueRock Therapeutics, Bristol-Myers Squibb, Calico Labs, Celgene, Cerevel Therapeutics, Coave Therapeutics, DaCapo Brainscience, Denali, Edmond J. Safra Foundation, Eli Lilly, Gain Therapeutics, GE HealthCare, Genentech, GSK, Golub Capital, Handl Therapeutics, Insitro, Janssen Neuroscience, Lundbeck, Merck, Meso Scale Discovery, Mission Therapeutics, Neurocrine Biosciences, Pfizer, Piramal, Prevail Therapeutics, Roche, Sanofi, Servier, Sun Pharma Advanced Research Company, Takeda, Teva, UCB, Vanqua Bio, Verily, Voyager Therapeutics, the Weston Family Foundation and Yumanity Therapeutics.We acknowledge Dr. Carmen Biancaniello for discussions.\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eData used in the preparation of this article was obtained on 2025-09-05 from the Parkinson\u0026rsquo;s Progression Markers Initiative (PPMI) database (www.ppmi-info.org/access-data-specimens/download-data), RRID:SCR\\_006431. For up-to-date information on the study, visit [www.ppmi-info.org](http:/www.ppmi-info.org) .\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eAlmada, A. E., Wu, X., Kriz, A. J., Burge, C. B. \u0026amp; Sharp, P. A. Promoter directionality is controlled by U1 snRNP and polyadenylation signals. Nature 499, 360\u0026ndash;363 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBai, B. et al. U1 small nuclear ribonucleoprotein complex and RNA splicing alterations in Alzheimer's disease. Proc. Natl. Acad. Sci. U.S.A. 110, 16562\u0026ndash;16567 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBeatriz, M. et al. Extracellular vesicles improve GABAergic transmission in Huntington's disease iPSC-derived neurons. Theranostics 13, 3707\u0026ndash;3724 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBerg, D. et al. MDS research criteria for prodromal Parkinson's disease. Mov Disord 30, 1600\u0026ndash;1611 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBerg, M. G. et al. U1 snRNP Determines mRNA Length and Regulates Isoform Expression. Cell 150, 53\u0026ndash;64 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBernheimer, H., Birkmayer, W., Hornykiewicz, O., Jellinger, K. \u0026amp; Seitelberger, F. Brain dopamine and the syndromes of Parkinson and Huntington. Clinical, morphological and neurochemical correlations. J Neurol Sci 20, 415\u0026ndash;455 (1973).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBoussaad, I. et al. A patient-based model of RNA mis-splicing uncovers treatment targets in Parkinson's disease. Sci. Transl. Med. 12, eaau3960 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraak, H., R\u0026uuml;b, U., Gai, W. P. \u0026amp; Del Tredici, K. Idiopathic Parkinson's disease: possible routes by which vulnerable neuronal types may be subject to neuroinvasion by an unknown pathogen. J Neural Transm (Vienna) 110, 517\u0026ndash;536 (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraak, H. et al. Staging of brain pathology related to sporadic Parkinson's disease. Neurobiol Aging 24, 197\u0026ndash;211 (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBreydo, L., Wu, J. W. \u0026amp; Uversky, V. N. Α-synuclein misfolding and Parkinson's disease. Biochim Biophys Acta 1822, 261\u0026ndash;285 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCampagne, S. U1 snRNP Biogenesis Defects in Neurodegenerative Diseases. ChemBioChem 25, e202300864 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCappelletti, C. et al. Transcriptomic profiling of Parkinson's disease brains reveals disease stage specific gene expression changes. Acta Neuropathol 146, 227\u0026ndash;244 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChaudhuri, K. R. \u0026amp; Schapira, A. H. V. Non-motor symptoms of Parkinson's disease: dopaminergic pathophysiology and treatment. Lancet Neurol 8, 464\u0026ndash;474 (2009).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCheslow, L., Snook, A. E. \u0026amp; Waldman, S. A. Emerging targets for the diagnosis of Parkinson's disease: examination of systemic biomarkers. Biomark Med 15, 597\u0026ndash;608 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChun, K.-Y. \u0026amp; Kim, S.-N. Integrative analysis of plasma and substantia nigra in Parkinson's disease: unraveling biomarkers and insights from the lncRNA\u0026ndash;miRNA\u0026ndash;mRNA ceRNA network. Front. Aging Neurosci. 16, 1388655 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDong, X. et al. Multi-omics machine learning classifier and blood transcriptomic signature of Parkinson's disease. Preprint at https://doi.org/\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003e10.21203/rs.3.rs-6837659/v1\u003c/span\u003e\u003cspan address=\"10.21203/rs.3.rs-6837659/v1\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eDorsey, E. R. et al. Global, regional, and national burden of Parkinson's disease, 1990\u0026ndash;2016: a systematic analysis for the Global Burden of Disease Study 2016. The Lancet Neurology 17, 939\u0026ndash;953 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFearnley, J. M. \u0026amp; Lees, A. J. Ageing and Parkinson's disease: substantia nigra regional selectivity. Brain 114 (Pt 5), 2283\u0026ndash;2301 (1991).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGe, S. X., Jung, D. \u0026amp; Yao, R. ShinyGO: a graphical gene-set enrichment tool for animals and plants. Bioinformatics 36, 2628\u0026ndash;2629 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGe, X. et al. Human microRNA-4433 (hsa-miR-4443) Targets 18 Genes to be a RiskFactor of Neurodegenerative Diseases. CAR 19, 511\u0026ndash;522 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGrossi, I. et al. MicroRNA\u0026ndash;34a\u0026ndash;5p expression in the plasma and in its extracellular vesicle fractions in subjects with Parkinson's disease: An exploratory study. Int J Mol Med 47, 533\u0026ndash;546 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHales, C. M. et al. U1 small nuclear ribonucleoproteins (snRNPs) aggregate in Alzheimer's disease due to autosomal dominant genetic mutations and trisomy 21. Mol Neurodegeneration 9, 15 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHirsch, E. C., Vyas, S. \u0026amp; Hunot, S. Neuroinflammation in Parkinson's disease. Parkinsonism \u0026amp; Related Disorders 18, S210\u0026ndash;S212 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eIrmady, K. et al. Blood transcriptomic signatures associated with molecular changes in the brain and clinical outcomes in Parkinson's disease. Nat Commun 14, 3956 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eJankovic, J. Parkinson's disease: clinical features and diagnosis. J Neurol Neurosurg Psychiatry 79, 368\u0026ndash;376 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRainer, J., Gatto, L., Weichenberger, C.X.. Bioconductor \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18129/B9.BIOC.ENSEMBLDB\u003c/span\u003e\u003cspan address=\"10.18129/B9.BIOC.ENSEMBLDB\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKhoo, S. K. et al. Plasma-Based Circulating MicroRNA Biomarkers for Parkinson's Disease. Journal of Parkinson's Disease 2, 321\u0026ndash;331 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eK\u0026otilde;ks, S. The Exon-Based Transcriptomic Analysis of Parkinson's Disease. Biomolecules 15, 440 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKuhn, M. Building Predictive Models in R Using the caret Package. J. Stat. Soft. 28, (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLashuel, H. A., Overk, C. R., Oueslati, A. \u0026amp; Masliah, E. The many faces of α-synuclein: from structure and toxicity to therapeutic target. Nat Rev Neurosci 14, 38\u0026ndash;48 (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, Y. et al. Candidate biomarkers of EV-microRNA in detecting REM sleep behavior disorder and Parkinson's disease. npj Parkinsons Dis. 10, 18 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMarek, K. et al. The Parkinson Progression Marker Initiative (PPMI). Progress in Neurobiology 95, 629\u0026ndash;635 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMorgan, M., Shepherd, L. AnnotationHub. Bioconductor \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.18129/B9.BIOC.ANNOTATIONHUB\u003c/span\u003e\u003cspan address=\"10.18129/B9.BIOC.ANNOTATIONHUB\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMoquin-Beaudry, G. et al. Mapping the peripheral immune landscape of Parkinson's disease patients with single-cell sequencing. Brain 148, 2847\u0026ndash;2860 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMount, S. M., Pettersson, I., Hinterberger, M., Karmas, A. \u0026amp; Steitz, J. A. The U1 small nuclear RNA-protein complex selectively binds a 5\u0026prime; splice site in vitro. Cell 33, 509\u0026ndash;518 (1983).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eMutez, E. et al. Transcriptional profile of Parkinson blood mononuclear cells with LRRK2 mutation. Neurobiology of Aging 32, 1839\u0026ndash;1848 (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNavarrete, F. et al. Whole transcriptome analysis of peripheral blood mononuclear cells from de novo and drug-naive Parkinson's disease patients. Neurotherapeutics 22, e00762 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNikom, D. \u0026amp; Zheng, S. Alternative splicing in neurodegenerative disease and the promise of RNA therapies. Nat. Rev. Neurosci. 24, 457\u0026ndash;473 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eOh, S., Aggarwal, A., Riester, M. \u0026amp; Waldron, L. HGNChelper: Identify and Correct Invalid HGNC Human Gene Symbols and MGI Mouse Gene Symbols. 0.8.15 \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.32614/CRAN.package.HGNChelper\u003c/span\u003e\u003cspan address=\"10.32614/CRAN.package.HGNChelper\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePaul, S. et al. Cerebellar Micro-RNA Profile in a Mouse Model of Spinocerebellar Ataxia Type 2. Neurol Genet 10, e200144 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePickrell, A. M. \u0026amp; Youle, R. J. The roles of PINK1, parkin, and mitochondrial fidelity in Parkinson's disease. Neuron 85, 257\u0026ndash;273 (2015).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003ePosavi, M. et al. Characterization of Parkinson's disease using blood-based biomarkers: A multicohort proteomic analysis. PLoS Med 16, e1002931 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRobin, X. et al. pROC: Display and Analyze ROC Curves. 1.19.0.1 \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.32614/CRAN.package.pROC\u003c/span\u003e\u003cspan address=\"10.32614/CRAN.package.pROC\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2010).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRonchi, J. \u0026amp; Foti, M. MIRit: an integrative R framework for the identification of impaired miRNA-mRNA regulatory networks in complex diseases. Preprint at \u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttps://doi.org/10.1101/2023.11.24.568528\u003c/span\u003e\u003cspan address=\"10.1101/2023.11.24.568528\" targettype=\"DOI\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScherzer, C. R. et al. Molecular markers of early Parkinson's disease based on gene expression in blood. Proc. Natl. Acad. Sci. U.S.A. 104, 955\u0026ndash;960 (2007).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eScott, H. et al. MiR-3120 Is a Mirror MicroRNA That Targets Heat Shock Cognate Protein 70 and Auxilin Messenger RNAs and Regulates Clathrin Vesicle Uncoating. Journal of Biological Chemistry 287, 14726\u0026ndash;14733 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShaheen, N. et al. MicroRNAs regulation in Parkinson's disease, and their potential role as diagnostic and therapeutic targets. NPJ Parkinsons Dis 10, 186 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSobhifar, P. \u0026amp; Brown, D. R. Interaction Between α-Synuclein and DJ-1 in Parkinson's Disease. Brain Sci 15, 899 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSoreq, L. et al. Small RNA sequencing-microarray analyses in Parkinson leukocytes reveal deep brain stimulation-induced splicing changes that classify brain region transcriptomes. Front. Mol. Neurosci. 6, (2013).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSpillantini, M. G. et al. Alpha-synuclein in Lewy bodies. Nature 388, 839\u0026ndash;840 (1997).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSu, D. et al. Projections for prevalence of Parkinson's disease and its driving factors in 195 countries and territories to 2050: modelling study of Global Burden of Disease Study 2021. BMJ 388, e080952 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSurmeier, D. J., Obeso, J. A. \u0026amp; Halliday, G. M. Selective neuronal vulnerability in Parkinson disease. Nat Rev Neurosci 18, 101\u0026ndash;113 (2017).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTai, Y. et al. miR-34a-5p regulates PINK1-mediated mitophagy via multiple modes. Life Sciences 276, 119415 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTansey, M. G. et al. Inflammation and immune dysfunction in Parkinson disease. Nat Rev Immunol 22, 657\u0026ndash;673 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eTng, T. J. W. et al. Systematic analysis of multi-omics data reveals component-specific blood-based biomarkers for Parkinson's disease. transl med commun 9, 12 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eT\u0026ouml;nges, L. et al. Blood-based biomarker in Parkinson's disease: potential for future applications in clinical research and practice. J Neural Transm (Vienna) 129, 1201\u0026ndash;1217 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eValente, D. et al. Distinct microRNA signatures define sporadic PSP-RS and PD in patient-derived midbrain organoids. iScience 28, 113162 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eVallelunga, A. et al. Identification of circulating microRNAs for the differential diagnosis of Parkinson's disease and Multiple System Atrophy. Front. Cell. Neurosci. 8, (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhang, W. T. et al. Circulating microRNAs as potential biomarkers for the diagnosis of Parkinson's disease: A meta-analysis. Neurolog\u0026iacute;a (English Edition) 39, 573\u0026ndash;583 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhu, W. et al. Effects of U1 Small Nuclear Ribonucleoprotein Inhibition on the Expression of Genes Involved in Alzheimer's Disease. ACS Omega 5, 25306\u0026ndash;25311 (2020).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"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":"npj-parkinsons-disease","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjparkd","sideBox":"Learn more about [npj Parkinson's Disease](http://www.nature.com/npjparkd/)","snPcode":"41531","submissionUrl":"https://submission.springernature.com/new-submission/41531/3","title":"npj Parkinson's Disease","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"transcriptomics, microRNA, parkinson's disease, biomarkers, multi-omics, blood-based diagnostics","lastPublishedDoi":"10.21203/rs.3.rs-9074426/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9074426/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eParkinson\u0026rsquo;s disease (PD) is a progressive neurodegenerative disorder characterised by the loss of dopaminergic neurons and the accumulation of intraneuronal misfolded α-synuclein aggregates. Despite advances in understanding the underlying molecular mechanisms of PD, diagnosis still relies largely on clinical symptoms that appear relatively late in disease progression. To advance both the mechanistic understanding of PD and the development of new diagnostic tools, we here performed an integrative analysis of matched whole-blood mRNA-seq and miRNA-seq data from 2,604 PPMI participants (Healthy n\u0026thinsp;=\u0026thinsp;658; PD n\u0026thinsp;=\u0026thinsp;1,946), stratified into pre-clinical (n\u0026thinsp;=\u0026thinsp;561), mild, moderate, and severe stages. Differential expression (adjusted p\u0026thinsp;\u0026lt;\u0026thinsp;0.05; |fold change| \u0026gt; 1.5) revealed a U-shaped trajectory, with maximal alterations in pre-clinical (7,427 DEGs; 157 DE-miRNAs) and severe PD. Pre-clinical signatures were dominated by widespread downregulation, notably involving multiple U1 snRNA genes, indicating early repression of U1-snRNP spliceosomal components. Integration of predicted targets with Spearman anti-correlation identified 252 miRNA\u0026ndash;mRNA pairs, driven by hub miRNAs including miR-34a-5p, miR-1299, miR-3120-5p, and miR-412-3p. Using pre-clinical DE-miRNAs for classification (n\u0026thinsp;=\u0026thinsp;1,219), Naive Bayes and GLMNet achieved AUCs of 0.886 and 0.883 with minimal overfitting. Taken together, these findings highlight the possible employment of early spliceosomal dysfunction and miRNA-mediated regulation as novel blood-based biomarkers of pre-clinical PD.\u003c/p\u003e","manuscriptTitle":"Integrative blood transcriptomics identifies U1-snRNP gene repression and miRNA–mRNA regulatory hubs in pre-clinical Parkinson’s Disease","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-04-01 10:13:10","doi":"10.21203/rs.3.rs-9074426/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"decision","content":"Revision requested","date":"2026-04-27T12:12:34+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-09T03:32:45+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-04-04T12:17:18+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"74537666212968715139354120220372082738","date":"2026-03-31T02:12:47+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"194173278825034441608082875604425300690","date":"2026-03-30T06:06:24+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2026-03-30T03:25:11+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2026-03-11T04:39:18+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2026-03-11T04:37:23+00:00","index":"","fulltext":""},{"type":"submitted","content":"npj Parkinson's Disease","date":"2026-03-09T14:41:45+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"npj-parkinsons-disease","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"npjparkd","sideBox":"Learn more about [npj Parkinson's Disease](http://www.nature.com/npjparkd/)","snPcode":"41531","submissionUrl":"https://submission.springernature.com/new-submission/41531/3","title":"npj Parkinson's Disease","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"NPJ","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"ed405563-5ac0-4f5b-b956-ec904357f0bf","owner":[],"postedDate":"April 1st, 2026","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"in-revision","subjectAreas":[{"id":65430125,"name":"Health sciences/Biomarkers"},{"id":65430126,"name":"Biological sciences/Computational biology and bioinformatics"},{"id":65430127,"name":"Health sciences/Diseases"},{"id":65430128,"name":"Biological sciences/Genetics"},{"id":65430129,"name":"Biological sciences/Molecular biology"},{"id":65430130,"name":"Health sciences/Neurology"},{"id":65430131,"name":"Biological sciences/Neuroscience"}],"tags":[],"updatedAt":"2026-04-27T12:25:39+00:00","versionOfRecord":[],"versionCreatedAt":"2026-04-01 10:13:10","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9074426","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9074426","identity":"rs-9074426","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}