Does Platelet Transcriptome Dysregulation Across the Lewy Body Continuum Mirror Neuronal Dysfunction?

Abstract

Platelets are increasingly recognized as multifunctional cells with roles extending beyond hemostasis to immune regulation, inflammation, and neurodegeneration. Here, we performed RNA-Seq profiling of platelets from patients with idiopathic REM sleep behavior disorder (IRBD), dementia with Lewy bodies (DLB), Parkinson disease (PD), Alzheimer disease (AD), and healthy controls (CTRLs) to explore disease-specific transcriptomic signatures. Across all groups, the RNA class distribution was similar, dominated by mRNAs (78–80%) and long non-coding RNAs (lncRNAs; 15–16%). DLB platelets displayed a reduced proportion of lncRNAs, suggesting an impaired RNA regulation, whereas IRBD concentrated the highest number of disease-specific lncRNAs, half of which were Y-linked, consistent with the male predominance observed in alpha-synucleinopathies. Differential expression analysis (DEA) revealed extensive transcriptomic remodeling in IRBD and DLB, particularly affecting RNA processing, cytoskeletal organization, and platelet activation pathways, while PD and AD showed minimal changes. These findings suggest a progressive impairment of platelet activation and signaling across the DLB continuum, potentially mirroring neuronal dysfunction. The limited transcriptional deregulation in PD may reflect its pronounced biological heterogeneity, consistent with recent multidimensional disease models. Overall, our study highlights platelets as accessible indicators of early and disease-stage-specific molecular alterations in α-synucleinopathies.

1. Introduction

Neurodegenerative diseases are a major and growing public health concern, with global cases projected to increase from 54.7 million in 2019 to 152.8 million by 2050, primarily due to aging populations [1]. Alzheimer’s disease (AD) is the most prevalent form, characterized by progressive cognitive decline and neuropathological hallmarks such as β-amyloid (Aβ) plaque accumulation and tau-related neurofibrillary tangles [2]. Dementia with Lewy bodies (DLB), the second-most common degenerative dementia, frequently overlaps with AD both clinically and pathologically, complicating early and accurate diagnosis [3,4]. Misdiagnosis of DLB remains high—up to 80% of cases are initially diagnosed as AD—often leading to suboptimal treatment. While cerebrospinal fluid (CSF) biomarkers, including reduced Aβ42 and elevated tau and neurofilament levels, are well established in AD, their specificity in distinguishing DLB remains controversial [5].

DLB and Parkinson’s disease (PD) are classified as Lewy body disorders (LBD), sharing a common pathophysiology involving the aggregation of α-synuclein [6]. Their underlying mechanisms include mitochondrial and lysosomal dysfunction, iron dysregulation, and neuroinflammation [7]. Importantly, idiopathic REM sleep behavior disorder (IRBD) has emerged as a prodromal phase of LBDs, since up to 91% of patients develop DLB or PD within 14 years [8,9]. Neuropathological and imaging findings in IRBD patients—such as reduced dopamine transporter (DAT) binding and aggregated α-synuclein species in CSF—support the assumption that IRBD is indeed an early synucleinopathy [10,11].

In recent years, blood and its different fractions have been explored as a peripheral biomarker source to support the differential diagnosis of neurodegenerative dementias. Platelets (PLTs), in addition to their crucial function in hemostasis [12], could be bridging blood and brain [13]. PLTs are a primary peripheral source of amyloid precursor protein (APP) in AD, contain α-synuclein, and express various neuronal receptors and enzymes. Additionally, they share biochemical pathways with dopaminergic neurons, including the capacity to store and release neurotransmitters [14]. Despite being anucleate, PLTs also contain diverse classes of RNAs, including mRNAs and miRNAs, along with the corresponding pathways that assure their functionality [15]. Therefore, miRNAs actively regulate mRNA levels [16] and might be responsible for changes in PLT activation capacity. These features highlight their potential role as systemic sensors linking environmental cues to internal physiological states. However, there still exists a significant research gap regarding how platelet transcriptomic alterations reflect central nervous system pathology and whether these molecular changes can reliably distinguish between neurodegenerative disorders. Given recent evidence supporting the role of platelets in neurodegenerative processes, analyzing their transcriptome rather than that of whole blood or PBMCs provides a more targeted and reliable approach, minimizing cellular heterogeneity while capturing biologically meaningful alterations linked to disease mechanisms.

In a recent study we profiled the platelet miRNome in DLB patients compared with controls and identified a DLB-specific biomarker signature [17]. This signature comprises seven miRNAs with decreased expression in DLB compared with controls and AD, discriminating efficiently between the two diseases. In an independent study, we evaluated the same miRNAs in IRBD and found that two of them were already downregulated at this prodromal stage [18].

Taking into account that miRNAs maintain their functionality in PLTs, we hypothesize that the observed reduction of miRNAs in DLB PLTs may lead to the overexpression of several mRNAs in these anucleate cells. Despite evidence supporting the role of platelets as peripheral models of LBD pathophysiology, and our previous findings on the platelet miRNome, the full platelet transcriptome remains unexplored in neurodegenerative diseases. The functional impact of miRNA dysregulation on platelet mRNA expression also remains unknown. Notably, while previous peripheral blood transcriptomic studies have focused on whole blood or PBMCs, this study is the first to apply Next-Generation Sequencing (NGS) to comprehensively profile the platelet transcriptome across these pathologies.

Therefore, the aim of our study was to analyze the PLT transcriptome in patients with DLB, AD, PD, IRBD and cognitively unaffected subjects in order to identify disease-specific expression changes and their associated pathways, providing molecular information required for the discovery of robust diagnostic and prognostic mRNA biomarkers in accessible peripheral blood cells. Additionally, we sought to compare our findings with previous NGS-based transcriptomic studies of PLTs and to explore potential age-related expression patterns.

2. Results

2.1. Demographic and Clinical Data

Demographic and clinical data of patients are summarized in

Table 1. Demographic and clinical data of the participants of the study.

	DLB (n = 12)	PD (n = 12)	IRBD (n = 12)	AD (n = 14)	CTRL (n = 14)	p1
Mean age, y 2 (age range, y)	74.1 (64–85)	66.9 (44–87)	74.5 (65–89)	68.8 (60–80)	71.3 (61–86)	0.017
Gender (male/female ratio)	7/5	8/4	9/3	7/7	7/7	0.069
Duration 3, years (range)	5.7 (2.1–10.6)	15.2 (4.9–23.7)	8.9 (2.5–18.2)	5.2 (0.8–8.0)
MMSE 4, mean (range)	15.3 (3–27)	n.a. 5		19.9 (5–28)	-	0.189
UPDRS-III 6, mean (range)	-	20.9 (5–39)		-	-	-
GDS fast 7, mean (range)	-	-		4.1 (3–6)	-	-
Parkinsonism, n (%)	10 (83.3%)	-	-	-		-
Positive DAT imaging, n (%)	11 (91.6%)	-	-	-		-

¹ p, p-value obtained by the Kruskal–Wallis test; ² y, years old; ³ duration, disease duration from disease onset to sample obtaining; ⁴ MMSE, Mini-Mental State Examination; ⁵ n.a., not applicable, since patients had no signs of cognitive impairment, thus MMSE was not carried out; ⁶ UPDRS-III, Unified Parkinson’s Disease Rating Scale Part III; ⁷ GDS, Global Deterioration Scale.

. Mean age was similar between DLB patients, IRBD patients and cognitively healthy controls (CTRLs). AD patients were mainly of early-onset, and PD patients were significantly younger. Due to the characteristics of the disease, the male-female ratio was higher in PD and IRBD than in AD and CTRLs. None of the PD patients had developed dementia at the time of sample collection and did not carry any LRRK2 or GBA variant.

2.2. Comparison of Gene Expression Profiles Across RNA-Seq Studies

In order to compare our RNA-Seq results with those from previous studies, we searched for RNA-seq studies in PLTs carried out between 2021 and 2025. In their review article, Thibord and Johnson provided a list of 60 RNA-Seq studies conducted between 2011 and 2023 [19]. Over the past two years, some additional PLT RNA-seq studies have been carried out. We retrieved the results from studies involving at least 25 control subjects aged 45 years or older, and three fulfilled the search criteria. Additionally, a study comparing the PLT transcriptome between 20 PD patients and 20 CTRL individuals was also included (

Table 2. Platelet RNA-Seq studies between 2021 and 2025, with more than 25 control individuals.

Study	Year	Accession Number	Samples (n)	Age 1	Expressed Genes 2
1a	2021 [20]	PRJNA732990	20 CTRLs	49.2 (21–75)	12,794
1b	2021 [20]	PRJNA732803	20 PD	67.1 (50–86)	13,747
2	2022 [21]	GSE183635	316 CTRLs	55.4 (18–86)	15,402
3	2022 [22]	PRJNA668820	56 CTRLs	47.8 (n/a)	12,401
4	2022 [23]	PRJNA737596	190 CTRLs	54.6 (31–72)	15,998
5a	2025	Our study	14 CTRLs	71.3 (61–86)	18,609
5b	2025	Our study	12 PD	66.9 (44–87)	18,132

¹ Mean age (age range). n/a—not available. ² Number of genes with more than 10 reads in all samples of each study.

).

2.2.1. CTRLs

Figure 1 shows the distribution of commonly expressed, overlapping and specific genes identified across the datasets. A total of 10,097 genes were shared among all five studies, corresponding to 78.9% of all genes detected in Study 1a, 65.6% in Study 2, 81.4% in Study 3, 63.1% in Study 4, and 54.3% in CTRLs of our study. In contrast, fewer than 1% of genes were uniquely detected in Studies 1a and 3, whereas 8.7%, 17.9%, and 22.9% of genes were specifically identified in Study 4, our study (Study 5a), and Study 2, respectively. The approximately 3500 genes uniquely expressed in Study 2 were primarily associated with ion channel processes, whereas the 3329 unique genes from our study (Study 5a) were predominantly involved in potassium channel functions.

2.2.2. PD

The overlap of expressed genes between the two studies that included PD patients (Studies 1b and 5b) is presented in Figure 2. A total of 12,873 genes were commonly expressed across both datasets, corresponding to 93.6% of all genes identified in the PD group of Study 1 and 79% in the PD group of our study (5b). The 5259 genes uniquely expressed in our PD patients were mainly related to ion channel and cytoskeletal motor activity.

2.3. Classification of Transcripts Expressed in PLTs

The distribution of five major RNA classes identified for each group is shown in Figure 3. Whereas 78–80% were protein-expressing genes, 15–16% were long non-coding RNAs (lncRNAs), 4% were pseudogenes and 1–2% were unknown transcripts. The group of minor RNAs, composed of small Cajal body-specific RNAs (scaRNAs), small nucleolar RNAs (snoRNAs), small nuclear RNAs (snRNAs), ribozymes and mitochondrial RNAs (mtRNAs), represented less than 1% of all RNAs.

2.3.1. Long-Non-Coding RNA (lncRNA)

The distribution of lncRNA differed among the five groups (p = 0.0064), and the relative amount of lncRNAs was smaller in DLB compared with CTRLs, AD and PD (p = 0.0063, 0.0023 and 0.0001, respectively). The relative amount of lncRNA was slightly, but not significantly lower in IRBD compared with CTRLs, AD and PD. However, the number of group-specific lncRNAs was similar in all groups, 13 in DLB, 20 in IRBD, 14 in PD, 12 in AD and 14 in CTRLs (Figure 4). Among those, whereas in DLB the majority (92.3%) were uncharacterized, only 50% were uncharacterized in IRBD. Group-specific lncRNA genes were distributed across all chromosomes, except for IRBD, where 50% of lncRNA genes were located on chromosome Y (Appendix A.1).

2.3.2. Minor RNAs

The analysis of minor RNAs revealed that most of them were snoRNAs (36–49%), followed by snRNA (19–31%) and scaRNAs (16–22%; Figure 1). Two ribozymes, RMRP and RPPH1, were present in all groups, representing between 3% and 5% of minor RNAs, and mtRNAs represented between 13% and 17% in DLB and IRBD, respectively, but only between 3% and 5% in PD, AD and CTRLs (Figure 5, Appendix A.2).

Whereas 61.5% of scaRNAs were found in all groups, only 24.5% of snoRNAs, 20% of snRNAs and 28.6% of mtRNAs were commonly expressed (Appendix A.3). When dividing the minor RNA groups into nuclear and mtRNAs, we found that IRBD expressed less specific nuclear RNAs compared with CTRLs and AD (p = 0.042 and p = 0.0038, respectively), and DLB compared with AD (p = 0.0071; Figure 6). On the contrary, both IRBD and DLB contained more specific mtRNA (66.8 and 71.4%, respectively) compared especially to PD and CTRLs, which expressed only common mtRNAs (Appendix A.3).

2.4. Differential Gene Expression, Gene Ontology (GO) Enrichment and KEGG Pathway Analysis

First, a post hoc power sensitivity analysis was conducted to determine the power of the study to detect a biologically relevant effect. The calculated statistical power was 96.8% and 97.8%, respectively.

In IRBD, 4690 differentially expressed genes (DEGs) were identified compared with CTRLs. Of these, 2568 (54.7%) showed increased and 1493 (45.3%) decreased expression. To determine which biological processes were positively or negatively affected, DEGs were subjected to Gene Ontology (GO) enrichment analysis. As a result, we observed an increase in the spliceosome and several mechanisms involved in RNA processing, and a decrease in processes related to PLT activation function. These specifically included the impairment of the actin cytoskeleton and PLT alpha-granules (Figure 7).

To further understand the functional consequences of gene overexpression or diminution, both gene lists were studied by Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis. Whereas the spliceosome was upregulated (Appendix B.1), an overall moderate downregulation of PLT activation-related processes was observed again (Appendix B.2). The latter included the diminution of junction proteins and impaired actin cytoskeleton regulation (Figure 8 and Appendix B.3).

In DLB, 6475 DEGs were found, of which 4036 were increased (73.5%) and 1778 decreased (26.5%). These DEGs were involved in 114 different biological processes identified by GO enrichment, with the most significantly enhanced processes related to ribosomal biogenesis, RNA modification, and the immune response, specifically involving adaptive immune response and lymphocyte-mediated immunity. In contrast, PLT activation (cytoskeleton organization, cell-to-cell signaling), adhesion (cell motility, regulation of localization) and aggregation (anatomical structure morphogenesis, tissue development) were markedly impaired, involving between 200 and 300 genes each (Figure 9).

KEGG pathway analysis revealed an enhancement of pathways related to the ribosome, and the impairment of pathways related to PLT activation, including Ca²⁺- and Rap-signaling pathways and the regulation of the actin cytoskeleton, involving approximately 40 to 50 genes per process (Figure 10, and Appendix B.2, Appendix B.3 and Appendix B.4, respectively).

In PD patients, only 23 genes were deregulated (1 gene (GPC6) increased and 22 decreased) and enriched GOs were related to peroxidase and oxygen activity, and enhanced pathways related to ribosomes. Finally, in AD patients, only 12 genes were deregulated (7 increased and 5 decreased). Most of these genes were enriched for GO terms related to microtubule minus-end binding and enoyl-CoA hydratase activity.

3. Discussion

In this study, we analyzed the PLT transcriptome in synucleinopathies including DLB, PD and one of their prodromal forms, IRBD, as well as in AD, and compared these groups with CTRLs. An increasing number of studies have begun to address PLT physiology, including their transcriptome, as PLTs are growingly recognized as multifunctional cells that extend far beyond their traditional role in hemostasis. They actively participate in inflammation, immune regulation, angiogenesis, and vascular integrity [24,25]. In recent years, this multifunctionality has linked PLTs to a wide spectrum of systemic and neurological diseases, since they contain abundant α-synuclein, APP and β-amyloid, and share key molecular and signaling machinery with neurons [26]. These shared features suggest that PLT function may mirror or even influence neuronal processes underlying neurodegeneration [27]. Thus, we sought to characterize the composition of the PLT transcriptome comprising five major RNA classes. Additionally, we wanted to identify altered molecular pathways in PLTs that could reflect disease-specific mechanisms of α-synucleinopathies and potentially serve as peripheral indicators of early or ongoing neurodegenerative changes.

3.1. Comparison of PLT RNA-Seq Studies

When comparing our RNA-Seq data with the other four studies, our dataset showed the highest number of expressed genes, particularly relative to Studies 1 and 3. This difference likely reflects greater sequencing depth in our study, as more reads per sample increase the detection of low-abundance transcripts. Our libraries were sequenced to ~50 million reads per sample, whereas this information was not reported for Studies 1–4. Correspondingly, genes uniquely detected in our PD samples but absent in Study 1b were predominantly low-expression genes.

Read length, sequencing strategy, and library preparation may also contribute to the detection of more genes. We used 150 bp paired-end reads, which improve alignment accuracy in repetitive regions such as pseudogenes and gene families. In contrast, three of the other studies used 100 bp single-end reads (except Study 4, which used paired-end reads), likely leading to more ambiguous alignments and fewer detected genes.

Additionally, we used the Illumina Stranded Total RNA Prep with Ribo-Zero Plus kit, which includes rRNA depletion and enhanced detection of low-abundance transcripts. The SMARTer Ultra Low RNA Kit used in Studies 2–4 lacks this depletion step, thus rRNA dominates the library, reducing sensitivity for other RNA species. Although the kit for library preparation was not reported for Study 1, the use of fragmented mRNA during library preparation could also explain the lower number of expressed genes compared to the other studies.

3.2. The Composition of the PLT Transcriptome

Across all five groups, IRBD, DLB, PD, AD and CTRLs the overall distribution of RNA classes was comparable, with mRNAs accounting for 78–80% of transcripts, lncRNAs for 15–16%, pseudogenes for approximately 4%, and 1–2% classified as unknown transcripts. This composition indicates that, in PLTs, protein-coding transcripts dominate the RNA landscape, while lncRNAs constitute a smaller but potentially functionally relevant subset. Additionally, it reflects the limited transcriptional activity of these anucleate cells and their reliance on mRNAs and regulatory RNAs inherited from megakaryocytes.

Interestingly, we observed a lower proportion of lncRNAs in DLB compared with the other groups. This reduction may indicate disease-specific alterations in the regulatory RNA repertoire of PLTs, potentially reflecting broader dysregulation in RNA processing, stability, or megakaryocyte-derived transcript packaging in DLB. Although lncRNAs are less abundant in PLTs than in nucleated cells, previous studies have shown that they exert key regulatory roles in formation, activation, and intercellular communication [28,29]. Large-scale transcriptomic analyses across human tissues have reported that lncRNAs can outnumber protein-coding genes and display high tissue specificity, whereas in PLTs an inverse trend is observed, with approximately fivefold more protein-coding genes than lncRNAs [30,31]. The reduced lncRNA fraction in DLB, therefore, may represent a loss of specific regulatory RNAs or an altered balance between coding and non-coding components of the PLT transcriptome, potentially mirroring disease-related changes in cellular homeostasis and RNA metabolism. Further studies integrating lncRNA expression with functional PLT phenotypes and disease severity could help to clarify whether these transcriptomic shifts have diagnostic or mechanistic significance in DLB and related synucleinopathies.

Among all groups, IRBD exhibited the highest number of disease-specific lncRNAs, with half (10 of 20) located on the Y chromosome. This striking enrichment may reflect sex-linked transcriptional regulation, consistent with the strong male predominance of IRBD and related synucleinopathies [32]. The presence of Y-linked lncRNAs in PLTs could therefore indicate early, sex-specific molecular alterations associated with prodromal stages of α-synuclein-related neurodegeneration.

3.3. Deregulation of Gene Expression in PLTs

When analyzing DEGs across the four disease groups compared with CTRLs, DLB exhibited the highest number of DEGs (>6400), followed by IRBD (4690). In contrast, both PD and AD showed fewer than 25 DEGs each. This distribution reveals two distinct transcriptomic response patterns: (1) a high-extent deregulation in IRBD and DLB, and (2) a low-extent response in PD and AD. These findings suggest two underlying biological behaviors—first, a progressive molecular impairment and shared pathway dysregulation within the LBD spectrum (IRBD–DLB continuum); and second, greater group heterogeneity in PD and AD, making it difficult to define disease-specific molecular signatures.

3.3.1. IRBD and DLB: Progressive Molecular Impairment and Shared Pathway Dysregulation

The extensive transcriptomic remodeling observed in IRBD, which is even more pronounced in DLB, suggests that IRBD may capture a prodromal phase marked by active molecular adaptation (4690 DEGs), whereas DLB reflects progression toward sustained inflammatory and translational activation (>6400 DEGs), consistent with advanced synucleinopathy. In IRBD, DEGs were strongly enriched in upregulated RNA processing pathways, particularly spliceosome-mediated pre-mRNA splicing, suggesting increased post-transcriptional activity in PLTs. This may reflect an adaptive response or early dysregulation of RNA maturation processes. Notably, specific nuclear RNAs were reduced compared with CTRLs and AD, indicating a possible imbalance between spliceosome assembly and RNA substrate availability. Such alterations could mirror compensatory or stress-induced changes in RNA metabolism, consistent with early molecular disturbances preceding over neurodegeneration.

Aberrant RNA processing has been increasingly linked to synucleinopathies, where α-synuclein interacts with RNA-binding proteins such as TDP-43, FUS, and hnRNPs, disrupting splicing, RNA stability, and translation [33,34]. The activation of spliceosomal pathways and concurrent nuclear RNA reduction in IRBD PLTs may therefore represent systemic manifestations of these neuronal processes, highlighting altered RNA metabolism as an early hallmark of disease [35].

Additionally, pathway analysis in IRBD revealed the impairment of PLT activation affecting the actin cytoskeleton, junction proteins and alpha-granule release, indicating early functional alterations that may compromise PLT responsiveness. Strikingly, in DLB, PLT activation mechanisms showed an overall impairment extending from the first activation steps to mechanisms related to adhesion and aggregation. At the same time, key intracellular signaling routes, including Ca2+- and Rap-dependent pathways, were also downregulated, indicating impaired signal transduction essential for PLT activation and integrin-mediated responses [36,37]. Taken together, these data point to a progressive loss of PLT activation capacity from IRBD to DLB, suggesting that cytoskeletal and signaling dysfunction in PLTs may mirror comparable alterations in neuronal and synaptic physiology characteristic of advancing synucleinopathy. However, whether these changes mirror impaired synaptic function needs to be further studied.

This progressive impairment of PLT activation and cytoskeletal regulation observed from IRBD to DLB could be linked to the physiological and pathological roles of α-synuclein. Under normal conditions, α-synuclein participates in vesicle trafficking, membrane curvature sensing, and actin cytoskeleton dynamics—processes essential for both neurotransmitter release in neurons and granule secretion in PLTs [38,39,40]. But, additional research should be carried out to determine whether the reduced activation, adhesion, and aggregation capacity, as found in our transcriptomic study, are secondary to α-synuclein dysregulation or aggregation.

Additionally, in our study, platelets from DLB patients showed a marked upregulation of immune-related pathways, particularly those linked to adaptive and lymphocyte-mediated responses. Similarly to findings in inflammatory disorders such as COVID-19, sepsis, and systemic lupus erythematosus, this suggests that platelets actively modulate immune processes through cytokine release and interactions with lymphocytes and endothelial cells [41]. The presence of comparable immune activation signatures in DLB indicates that platelet–immune crosstalk may contribute to neurodegenerative mechanisms, potentially mediated by platelet-derived α-synuclein or inflammatory signaling. Overall, these results indicate that platelets could represent dynamic immunomodulatory cells connecting peripheral immune alterations with central pathology in LBD, a hypothesis that needs further corroboration.

3.3.2. PD and AD: Disease Heterogeneity

In our PLT RNA-Seq data, the striking contrast between a lack of transcriptional signal in PD and AD versus the large signal in IRBD and DLB highlights the heterogeneity within neurodegenerative diseases and underscores the potential for masked subtype-specific signatures. Particularly, this limited transcriptional response in PD may reflect the substantial biological heterogeneity of the disease. As shown by the recently proposed SynNeurGe framework, PD encompasses multiple interacting dimensions—α-synuclein pathology (S), neurodegeneration (N), and genetic predisposition (G)—which combine to produce diverse clinical phenotypes (C) [42]. This multidimensional model underscores that individuals clinically diagnosed with PD may represent distinct molecular subtypes with variable synuclein burden, neurodegenerative progression, and genetic background. Consequently, pooled transcriptomic analyses may obscure subtype-specific signatures, masking expression changes that are more evident in biologically homogeneous groups such as IRBD or DLB.

Comparable heterogeneity has been demonstrated in AD. In a recent large-scale study, five molecular AD subtypes based on CSF proteomic profiles have been identified. These were defined, respectively, by hyperplasticity, innate immune activation, RNA dysregulation, choroid plexus dysfunction, and blood–brain barrier impairment [43]. These subtypes differ in genetic risk factors, cortical atrophy patterns, and clinical trajectories. Similarly, a meta-analysis integrating neuropathological and neuroimaging data delineated four biological subtypes—typical, limbic-predominant, hippocampal-sparing, and minimal atrophy AD—each with distinct regional tau distribution, demographic associations, and disease progression patterns [44]. All together, these findings reinforce that both PD and AD encompass multiple mechanistic trajectories rather than a single linear disease continuum.

Within this framework, the absence of major transcriptomic deregulation in AD PLTs may indicate that PLTs do not capture all biological pathways involved across AD subtypes. Given our limited sample size, subtype-specific alterations may have been masked in the current analysis. Correspondingly, it could be expected that PLTs of patients with the AD subtype characterized by blood–brain barrier impairment would indeed show important transcriptomic changes [45,46].

3.4. Future Biomarker Development

Our findings—demonstrating altered gene expression profiles in platelets, an easily accessible biofluid—in DLB and IRBD provide strong evidence for their potential translation into clinically useful biomarkers. To advance this translational pathway, we are currently finalizing differential expression analyses among disease groups, including both synucleinopathies (DLB and PD), IRBD, and DLB versus AD, to delineate pathology-specific transcriptional signatures. Building on the RNA-seq data, which revealed extensive dysregulation in pathways such as the spliceosome, ribosome, platelet activation, and immune response, we are now focusing on defining a core, clinically actionable gene signature. This will involve prioritizing top-ranking DEGs with the highest effect sizes and lowest variability, particularly those belonging to the most discriminative pathways, followed by orthogonal validation using qPCR to ensure robustness and technical reproducibility.

To ensure clinical reliability and generalizability, we plan to validate this core signature in an independent, large, and clinically well-characterized cohort. The validated qPCR data will then be used to develop predictive models—such as logistic regression or machine learning approaches—optimized for diagnostic and prognostic applications. Specifically, these models aim to distinguish DLB from AD and to predict phenoconversion in IRBD patients who progress to DLB or PD. Translating the platelet RNA signature into a routine clinical assay will also require addressing key optimization procedures and establishment of clinically relevant cut-off values based on model sensitivity and specificity. Together, these efforts are expected to facilitate the development of a robust and clinically implementable diagnostic and prognostic assay for synucleinopathies.

4. Materials and Methods

4.1. Source of PLT Samples

In total, 64 individuals were prospectively recruited at the Neurodegenerative Disease Unit of the Neurology Department from the Hospital Universitari Germans Trias i Pujol (HUGTP; Badalona, Barcelona, Spain) and Sleep Unit of the Neurology department from the Hospital Clinic de Barcelona (Barcelona, Spain). The cohort included individuals divided into five groups. The four patient groups were: 12 DLB patients who fulfilled criteria for probable DLB [3], 12 PD patients diagnosed according to the UK PD Society Brain Bank criteria [47], 12 IRBD patients who fulfilled diagnostic criteria in subjects reporting nightmares and dream-enacting behaviors in whom nocturnal video-polysomnography showed increased electromyographic activity in REM sleep [48], and 14 AD patients fulfilling criteria for probable AD (National Institute on Aging–Alzheimer’s Association criteria) [49]. Group five was composed of 14 cognitively unaffected controls (CTRLs), mainly non-blood relatives of the patients without a family history of neurological disorders, suggestive symptoms of IRBD or any symptoms or signs indicating parkinsonism or cognitive impairment.

The study was approved by the HUGTP Ethical Committee for Clinical Investigation (PI-22-024). All participants or their legal guardians signed written informed consent according to the Declaration of Helsinki [50].

4.2. PLT Obtaining and RNA Purification

Na-citrate Vacutainer tubes (BD, Plymouth, UK) were used to collect blood samples. They were processed by two consecutive centrifugation steps within the first four hours after blood extraction to avoid expression changes due to PLT activation. First centrifugation was performed at 200× g for 15 min to obtain PLT-rich plasma, and the following centrifugation was carried out at 2500× g for 15 min to obtain PLT-rich pellets (PRP). PRP was stored at −80 °C until RNA purification and thawed on ice before processing. Total RNA isolation was performed using the mirVana^TM miRNA Isolation Kit (ThermoFisher, Waltham, MA, USA), and isolated RNA samples were stored at −80 °C until expression analysis.

4.3. Total RNA Discovery by Next-Generation Sequencing (NGS)

The concentration of total RNA samples was measured by Qubit and 15 ng from each sample were adjusted to a final volume of 6.5 µL. Samples were pooled in pairs. Pooling was done for samples with similar concentrations, to assure similar RNA proportions and avoid confounding enrichment. Quality control and size distribution of the pools was assessed with the 4200 TapeStation System (Agilent Technologies, Santa Clara, CA, USA) using High Sensitivity RNA ScreenTape (Agilent, Santa Clara, CA, USA). Of the pooled sample pairs, only 5 had RIN values between 5.5 and 6.5; the rest of them presented RIN values higher than 6.5.

From each pool, 11 µL was used for library preparation by Illumina Stranded Total RNA Prep, Ligation with Ribo-Zero Plus for Illumina (Illumina, San Diego, CA, USA) following the manufacturer’s instructions. Amplification cycles were adjusted according to the estimated RNA concentration. Individual libraries were subjected to quality analysis and quantification using a D1000 ScreenTape (Agilent Technologies, Santa Clara, CA, USA). If the library profile was not acceptable, a repurification step was performed. All libraries were adjusted to 8 nM in a final volume of 5 µL and were all pooled together. The library pool was finally assessed by High Sensitivity D1000 ScreenTape (Agilent Technologies, Santa Clara, CA, USA).

Clustering and sequencing were performed in an Illumina Sequencer (NovaSeq6000, Illumina, San Diego, CA, USA). Samples were sequenced to ~50 million (M) 2 × 150 bp paired-end. Sample quality was checked using FastQC and MultiQC tools (version 0.12.0 and 1.14, respectively) [51,52]. The obtained FastQ raw data was analyzed as follows: (1) the TrimGalore tool (version 0.6.10) was used to remove the adapter sequences from the reads and the base pairs with a Phred score under 20 [53]; (2) paired RNAseq reads were aligned to the Gencode GRCh38.p109 human reference genome using STAR (version 2.7.10b) [54]; (3) the Salmon tool (version 1.4.0) was used for quantifying the expression of each transcript [55]; (4) the matrix count was generated with tximport (version 1.30.0) [56]; (5) the total count of reads was normalized using the median of ratios method from DESeq2 package (version 1.45.1) [57]. Steps 1 to 3 were performed from the Ubuntu terminal (Linux), while steps 4, 5, and the following analysis were conducted using R software (version 4.3.2).

4.4. Sequencing Data Analysis

For NGS expression analysis, transcripts with a minimum of 10 reads per sample were considered expressed. Gene biotype classification (mRNA, pseudogene, lncRNA, scaRNA, snRNA, snoRNA, rRNA, and mitochondrial RNA) of the genes that were expressed in the different groups (AD, IRBD, DLB, PD, and CTRLs) was obtained using the getBM function from the biomaRt package (version 2.58.2).

Differential expression analysis (DEA) was performed using the Wald test, and p-values were corrected by the Benjamini–Hochberg method with DESeq2, establishing significance as an adjusted p-value lower than 0.05. A post hoc power sensitivity analysis was performed assuming a Fold-Change of 2 and using the RNASeqPower package (version 1.42.0).

4.5. Comparison of RNA Expression Between RNA-Seq Studies

RNA-seq data from four different studies analyzing the whole PLT transcriptome was retrieved from the NCBI Gene Expression Omnibus (GEO) and NCBI Sequence Read Archive (SRA) databases: PRJNA732990, PRJNA732803, GSE183635, PRJNA668820 and PRJNA737596, (defined in results as studies 1a, 1b, 2, 3 and 4, respectively). For the GSE183635 study, raw data counts were obtained directly from its repository. For PRJNA732990, PRJNA732803, PRJNA668820 and PRJNA737596 studies, FastQ files were acquired. As these studies used paired-end sequencing, FastQ files of all samples were processed similarly to those in our study, using the Salmon tool for quantifying the expression of each transcript and the tximport function for generating the matrix count. Only the expression of control individuals from each study was considered in the comparative analysis, except for Study 1 where PD samples were also obtained. In all samples, only transcripts with at least 10 reads were kept. A Venn diagram (VennDiagram package, version 1.7.3) was obtained to visualize the expression of overlapping genes among studies.

4.6. LncRNA and Minor RNA Distribution Analysis

The distribution of RNA biotypes and four groups of minor RNAs (scaRNAs, snoRNAs, snRNAs and mtRNAs) was analyzed by comparing all five groups. Pairwise comparison between groups was carried out using chi-square and Fisher’s exact test. The results were corrected for multiple testing by the Bonferroni method. Significance was set at 0.05.

4.7. Gene Ontology (GO) Enrichment and KEGG Pathway Analysis

Gene Set Enrichment Analysis (GSEA) of transcripts differentially expressed between the four diseases (IRBD, AD, DLB and PD) and CTRLs was performed using the enrichGO and gseDO function from clusterProfiler (verison 4.10.1) and DOSE packages (version 3.28.2), respectively; and KEGG pathway analysis using gseKEGG function from DOSE package. Dotplot function from DOSE package was used to generate dotplots. A p-value below 0.05 was considered a significant enrichment.

GO enrichment was also analyzed for the genes specifically expressed in CTRL individuals from each of the five different RNAseq studies and in PD patients from Study 1b and our study.

5. Conclusions

In summary, our study provides the first comprehensive comparison of PLT transcriptomes across major neurodegenerative disorders, revealing disease- and stage-specific molecular alterations. PLTs from IRBD and DLB patients displayed extensive transcriptomic remodeling, encompassing dysregulation of RNA processing, cytoskeletal organization, and activation pathways. These findings suggest that systemic RNA metabolism and PLT signaling mechanisms may mirror early and progressive stages of α-synucleinopathy. In contrast, the limited number of DEGs in PD supports the growing view that PD represents a highly heterogeneous disorder, in which diverse molecular subtypes may obscure shared transcriptomic signatures at the group level.

The absence of significant deregulation in AD PLTs further underscores that not all central pathological processes are reflected peripherally, particularly given the molecular heterogeneity of AD subtypes. Nonetheless, our results highlight the potential of PLT transcriptomics to reveal peripheral molecular correlates of neurodegeneration and identify early systemic changes in prodromal synucleinopathies such as IRBD. Future studies integrating multi-omic PLT profiling with clinical, imaging, and genetic data will be essential to validate these findings and to determine whether specific PLT transcriptomic signatures could serve as accessible biomarkers of disease onset, subtype, or progression across neurodegenerative disorders.