Differentially Expressed Salivary miRNAs in Temporomandibular Disorders

ABSTRACT

Objective

Temporomandibular disorders (TMDs) are heterogeneous conditions of unclear aetiology involving the temporomandibular joint, masticatory muscles and neural tissues. Limited understanding of their pathogenesis hampers accurate diagnosis and targeted treatment. Therefore, this study aimed to identify salivary microRNA (miRNA) signatures associated with TMDs to support future diagnostic, therapeutic and prognostic applications.

Materials and Methods

Unstimulated cell‐free saliva (5 mL) was collected from 9 adult female TMD subjects (using Diagnostic Criteria/TMD) and eight healthy female controls of similar ages. Total RNA was extracted, small RNA libraries were prepared, and sequencing was performed using Illumina NovaSeq 6000. Reads were aligned to the human genome (GRCh38) via STAR. Differential expression analysis was conducted using DESeq2, followed by functional enrichment via miEAA 2.1.

Results

A total of 187 salivary miRNAs were significantly differentially expressed between TMD and control groups (adjusted p < 0.05; log2‐fold change > +1 or < −1), with 125 upregulated and 62 downregulated in TMD subjects. Several differentially expressed miRNAs were linked to the negative regulation of cadherin‐mediated cell–cell adhesion, neurogenesis and chemokine production. Some overlapped with miRNAs implicated in rheumatoid arthritis and osteoarthritis, suggesting shared mechanisms. While no clear association was found between miRNA and TMD phenotypes, 5 miRNAs were strongly (R = 0.67–0.77) and significantly (p < 0.05) correlated with pain intensity and chronic pain grade.

Conclusions

Salivary miRNA profiling offers promise as a non‐invasive diagnostic tool for TMDs, with the potential to uncover molecular endotypes and disease mechanisms not evident through clinical evaluation. Future studies with larger, more diverse cohorts are needed to validate these findings and assess their clinical utility.

1. Introduction

Temporomandibular disorders (TMDs) affect approximately 5%–10% of the U.S. population and are associated with an estimated $4 billion in annual healthcare costs [1, 2]. These conditions involve dysfunction of the temporomandibular joint (TMJ), masticatory muscles, neural tissues or a combination of these, and are frequently accompanied by pain, joint sounds and limited mandibular movement. While past research has provided insights into the epidemiological, genetic and phenotypic profiles of TMDs [3, 4], the initiating biological events and downstream effects remain poorly defined [1]. Thus, other than clinical phenotyping, no definitive diagnostic tests or disease‐modifying therapies exist, and current treatments remain largely palliative. Existing classification systems cannot adequately differentiate among TMD subtypes, largely due to overlapping clinical features and variable symptom presentation [2].

Identifying reliable biomarkers could improve diagnosis, disease monitoring and treatment evaluation in TMDs. Studies of synovial fluid, tissue biopsies and blood have identified candidate biomarkers [5, 6], including elevated pro‐inflammatory cytokines (TNF, IL‐1β, IL‐6, IL‐8), matrix metalloproteinases (MMP‐1, −3, −9, −13) [7, 8] and bone turnover markers such as CTX‐I/II, osteocalcin and bone‐specific alkaline phosphatase [9] and reduced interleukin‐10 [10, 11]. Additional evidence links autoimmune components (e.g., rheumatoid factor, anti‐citrullinated peptide antibodies) and neuropeptides (substance P, calcitonin gene‐related peptide) to chronic TMJ pain [12, 13]. While these findings provide windows of insight on TMD pathophysiology, they have not yet translated into clinically actionable biomarkers, possibly because assays focusing on a few analytes often lack sufficient sensitivity or specificity for TMD endotyping [5, 14]. The systematic application of contemporary approaches such as omics offers powerful and new yet untapped tools to discern molecular signatures of TMD towards clinical applications. These techniques, applied on more readily and less invasively accessible biospecimens than blood or synovial fluid that have been utilised in most studies, have the potential of expanding the routine clinical applicability of these potent approaches [2, 5].

Saliva is attractive to other biospecimens for its non‐invasive collection, repeatability and suitability for omics‐based profiling [14, 15]. Preliminary work has revealed alterations in salivary proteins such as alpha‐amylase and cystatins in patients with TMD [14]. Yet no individual salivary biomarker has demonstrated adequate diagnostic reliability. Among molecular biomarkers, salivary microRNAs (miRNAs) are especially promising due to their stability in biofluids, presence in saliva and known roles in regulating inflammation, cartilage homeostasis, pain and disease‐specific expression profiles [16, 17]. To date, no studies have comprehensively evaluated salivary miRNAs as diagnostic biomarkers for TMDs. The objective of this proof‐of‐concept study performed on a limited sample size was to identify differentially expressed salivary miRNAs in individuals with TMDs relative to those in asymptomatic controls towards the long‐term goal of developing reliable, non‐invasive molecular tools for diagnosis, monitoring and therapeutic targeting of TMDs.

2. Material and Methods

2.1. Subject Recruitment and Clinical Phenotyping

Subjects were recruited and consented following study approval by Institutional Review Boards at University of California Los Angeles (UCLA) (22‐001676, expiration 12/11/2025) and University of Michigan (HUM00225775, expiration 12/11/2025). The TMD findings were phenotyped using Diagnostic Criteria for Temporomandibular Disorders (DC/TMD) Axis I by calibrated examiners at the two sites.

2.2. Saliva Collection, Stabilisation and Processing

Unstimulated saliva samples (5 mL) were collected as described previously [15] from 17 consented adult subjects (9 with TMDs and 8 pain‐free healthy controls) satisfying the inclusion/exclusion criteria summarised in Table 1. Prior to saliva collection, subjects were asked to refrain from eating, drinking, smoking or oral hygiene procedures for at least 1 h prior to the collection. Following collection, saliva was centrifuged, 2600 × g for 15 min at 4°C, cell‐free saliva (CFS) supernatant collected, RNA inhibitor (SUPERase In, Ambion, Thermo Fisher Scientific, Waltham, MA, USA) added for RNA stabilisation and samples were frozen at −80°C until further analysis.

TABLE 1.

Study inclusion and exclusion criteria.

TMD group inclusion criteria

Presence of at least one cardinal TMD symptom (jaw pain, limited opening or TMJ noise)

Specific screening clinical findings must include one or more of the following: TMJ pain, internal derangement, limited mouth opening due to joint locking, jaw clicking, popping or crepitus

Control group inclusion criteria

No history or clinical signs of TMD (e.g., jaw pain, noise, locking crepitus)

No pain on TMJ exam

No systemic pain finding

Normal TMJ function by clinical standards

Exclusion criteria for all participants

Medical condition or physical exam finding that precludes participation, such as systemic arthritides, juvenile idiopathic arthritis, idiopathic condylar resorption, connective tissue disorders, mandibular osteonecrosis, mandibular and condylar fractures, salivary gland disorders, neoplasms, neurologic/neuropathic, endocrine or immune/autoimmune diseases and congenital or developmental disorders.

Pathologic TMJ found on imaging except disc displacements and osseous pathology including osteoarthritis/osteoarthrosis

TMJ or jaw surgery

Recent trauma (< 2 months), febrile illness delaying participation (> 90 days) or radiation to the head/neck

Pregnancy or lactation

Use of narcotic pain medication, muscle relaxants or steroid therapy unless discontinued for 60 days prior to examination

Use of antidepressant drugs unless the participants have been on a stable dose for 60 days.

Use of prescription or over‐the‐counter nonsteroidal anti‐inflammatory medications unless the medication(s) were discontinued for 60 days prior to the examination (use of acetaminophen allowed as a rescue drug)

Drug and/or alcohol abuse

Ongoing dental treatments requiring invasive procedures (implant placement) or prolonged mouth opening or that will be delayed by participation in the study

Presence of non‐TMD orofacial pain disorders on clinical examination

2.3. RNA Isolation, cDNA Library Construction and RNA‐Seq for Profiling miRNAs

RNA was isolated using the miRNeasy Micro Kit (Qiagen, Hilden, Germany), and the NEBNext Small RNA Library Prep Set for Illumina (New England Biolabs, Ipswich, MA, USA) used for the small exRNA library construction. Libraries were quantitated using the KAPA SYBR FAST Universal qPCR Kit (Roche, Basel, Switzerland) and diluted to 2 nM final stocks, pooled in equal molar amounts and sequenced on the Illumina NovaSeq 6000 platform (Illumina Inc., San Diego, CA, USA) using 8 nt unique dual indexes (forward and reverse PCR primer) to generate 30 M paired reads per sample at 100 nt read lengths with the 0.5 mL volumes of CFS and with sequencing to a depth of 20 M (1 × 50nt) reads.

2.4. Bioinformatic Analysis of RNA‐Seq Data

The salivary RNA‐ (sRNA‐) Seq data were first processed to remove adapter sequences and low‐quality reads. The bioinformatic analysis was performed using the pipelines we have previously developed, specifically for processing of sRNA‐Seq data of human saliva [15]. Briefly, raw sequencing data was aligned by STAR (v2.6+) to the human genome GCHr38 allowing at most 1 mismatch in the entire read [15], following which, the annotation was performed for miRNAs (miRBase, release 22.1) [18]. For human miRNAs, only uniquely mapped reads were retained. Count data from HTSeq were calculated [19], and differential expression (DE) analysis was performed using DESeq2 [20] to identify statistically significant salivary miRNAs associated with TMDs versus controls.

2.5. Assignment of Discovered miRNA Molecular Targets to Molecular Pathways Implicated in TMD Mechanisms

The set of the most statistically significant DE miRNAs over‐ and under‐expressed in TMD group (p adj < 0.05; log2‐fold change <−1 and > +1) were used for the pathway analysis. Bioinformatic analysis was performed using the miRNA Enrichment and Annotation analysis [21], ncRNA‐disease associations in mammals (MNDR) [22], KEGG [23] and miRWalk [24] pathways. The Benjamini‐Hochberg FDR method is used within those packages to control for multiple hypothesis testing in searching for enrichment of the genes associated with TMD for each gene ontology (GO) and pathway term.

2.6. Correlation Between Pain Intensity/Disability and Salivary miRNAs

Pain intensity was calculated as the mean of three 0–10 numeric rating items: current pain, worst pain and average pain in the last 30 days. This value was multiplied by 10 to yield a score ranging from 0 to 100. Similarly, disability score was calculated as the average of three 0–10 items assessing interference with daily, social and work‐related activities, also scaled from 0 to 100. Disability days were assessed by asking how many days the participant was kept from usual activities due to pain in the past 30 days. Participants were then classified into one of five chronic pain grades (CPG; 0–IV) as follows: Grade 0 = No pain (pain intensity = 0, disability score = 0 and disability days = 0); Grade I = Low intensity pain (< 50), low disability (< 30), with < 7 days of activity limitation; Grade II = High intensity pain (≥ 50), low disability (< 30), with < 7 days of activity limitation; Grade III = Pain intensity ≥ 30, disability score ≥ 30 and < 15 disability days; Grade IV = Pain intensity ≥ 30, disability score ≥ 30 and ≥ 15 disability days (Table 2). Spearman's correlations were conducted between all variables and Pain Intensity, Disability Score and CPG and reported as R values.

TABLE 2.

Summary of chronic pain grade (CPG) scale and DC/TMD findings for the TMD subjects (N = 9). Grade 0 = No pain (pain intensity = 0), disability score = 0 and 0 disability days; Grade I = Low intensity pain (< 50), low disability (< 30), with < 7 days of activity limitation; Grade II = High pain intensity (≥ 50), low disability (< 30), with < 7 days of activity limitation; Grade III = Pain intensity ≥ 30, disability score ≥ 30 and < 15 disability days; Grade IV = Pain intensity ≥ 30, disability score ≥ 30 and ≥ 15 disability days.

Summary of CPG Scale
Pain Grade, N (%)	Total (N = 9)
I—Low intensity, low disability	0 (0%)
II—High intensity, low disability	4 (44.4%)
III—High disability, moderately limiting	4 (44.4%)
IV—High disability, severely limiting	1 (11.1%)
Pain intensity
Mean (SD)	70.0 (12.91)
Median	66.7
Range	53.3–90.0
Disability Score
Mean (SD)	30.0 (25.11)
Median	33.3
Range	0.0–76.7

Summary of DC/TMD pain diagnosis	N (%)
None
No	0 (0.0%)
Myalgia
No	3 (33.3%)
Yes	6 (66.7%)
Myofascial pain with referral
No	3 (33.3%)
Yes	6 (66.7%)
Right Arthralgia, N
No	5 (55.6%)
Yes	4 (44.4%)
Left Arthralgia, N
No	2 (22.2%)
Yes	7 (77.8%)
Headache attributed to TMD
No	4 (44.4%)
Yes	5 (55.6%)

3. Results

3.1. Salivary miRNA Reflects TMD Biomarker Signature

Given the high female versus male predilection for TMD, unsurprisingly all 9 subjects recruited in the TMD group were females ranging in age from 22 to 65 years (Table S1). Thus, the eight control samples used were also from females aged between 25 and 63 years. The TMD subjects presented with variable combinations of clinical phenotypes including disc displacement (DD), osteoarthritis (OA), arthralgia and myalgia.

Bioinformatic pipelines were applied for analysis of sRNA‐Seq data. Gene expression data revealed 187 DE miRNAs in TMD versus TMD‐free controls (p adj < 0.05), with 125 over‐ and 62 underexpressed miRNAs in TMD subjects. The heatmap for top 50 DE miRNAs is shown in Figure 1. The results of the clustering analysis revealed two separate cluster groups, which segregates the clinical condition of TMDs from controls demonstrating that salivary miRNAs can differentiate between TMD and TMD‐free subjects. However, with the four clinical phenotypes having overlapping presentations the findings did not generate any discernible relationships between DE miRNAs and these TMD signs and symptoms.

FIGURE 1.

Heatmap of top 50 DE miRNAs between TMDs and non‐TMDs (p adj < 0.05). Bottom panel shows four primary TMD clinical phenotypes for each subject. A, arthralgia; DD, disc displacement; M, myalgia; N, No; OA, osteoarthritis; Y, Yes.

The Volcano plot (Figure 2) enable visualisation of miRNAs that are not only statistically significantly DE between the two groups (p adj < 0.05) but also display robust fold changes (log2‐fold change > +1 or < −1). The resulting plot for miRNAs shows highly underexpressed (left of Y axis, blue dots) or strongly overexpressed (right of Y axis, red dots) miRNAs in TMD compared to the TMD‐free group as well as statistical significance (p adj < 0.05; above red horizontal line). Black dots represent non‐significant miRNAs. The plot highlights the most strongly regulated and statistically significant miRNAs associated with TMD. Overall, our analysis revealed twice as many upregulated miRNAs (125) compared to those downregulated (62) with a remarkable hsa‐miR‐223‐3p that achieved the strongest statistical significance (p adj = 4.48E‐32, log2 fold change = 5.8116).

FIGURE 2.

Volcano plot for DE miRNAs with blue dots representing statistically significant downregulated miRNAs (p adj < 0.05; log2‐fold changes < −1) and red dots representing statistically significant upregulated miRNAs in TMD versus non‐TMD controls (p adj < 0.05; log2‐fold changes > +1).

3.2. Potential GO, Disease and Pathways Associated With Pathophysiology of TMDs

Functional enrichment analysis revealed several major GO signalling pathways for the 62 significantly decreased miRNAs in the TMD group compared to TMD‐free controls, including negative regulation pathways for cell–cell adhesion mediated by cadherin and neurogenesis and positive regulation of chemokine production (p adj < 0.05; log2‐fold change < −1) (Figure 3A). Among all miRNAs, down‐regulated miR‐122‐5p, miR‐26a‐5p, miR‐30a‐3p, miRNA‐22‐3 and miR‐125a‐5p are specifically associated with the majority of the enriched terms and with depleted diseases such as rheumatoid arthritis (RA), arthritis and OA (Figure 3B). In addition, the KEGG pathway analysis revealed several enriched categories associated with inflammation, pain, tissue remodelling and immune dysregulation, such as tryptophan metabolism (p value = 7.47e‐5, p adj = 0.006; 16 miRNAs including hsa‐miR‐5703, hsa‐miR‐6760‐3p, hsa‐miR‐1207‐5p, hsa‐miR‐1290, etc.); TGF‐beta signalling pathway (p value = 0.0016, p adj = 0.0177; 35 miRNAs including hsa‐miR‐6500‐5p, hsa‐miR‐6875‐3p, hsa‐miR‐5189‐5p, hsa‐miR‐127‐3p, etc.); extracellular matrix (ECM)—receptor interaction (p value = 6.83e‐4; p adj = 0.011; 22 miRNAs including hsa‐miR‐6875‐3p, hsa‐miR‐2861, hsa‐miR‐5189‐5p, hsa‐miR‐3911, etc.); or Notch signalling pathway (p value = 5.92e‐4, p adj = 0.010; 26 miRNAs including hsa‐miR‐6875‐3p, hsa‐miR‐6760‐3p, hsa‐miR‐6771, hsa‐miR‐3190‐3p, etc.) (Figure 3C).

FIGURE 3.

miRNA Enrichment and Annotation Analysis (miEAA) of 62 downregulated miRNAs in TMD group compared to TMD‐free controls (p adj < 0.05, log2‐fold change < −1). (A) Gene ontology—top 20 enriched functional categories for miRNAs sorted by p value. (B) Diseases (MNDR)—top 3 subcategories. (C) Pathway analysis—top 20 enriched categories sorted by significance (KEGG pathways).

Enriched categories for the 125 significantly increased miRNAs in TMD subjects compared to non‐TMD controls included those associated with various biological functions, molecular processes, pathways and diseases (p adj < 0.05, log2‐fold change > 1) (Figure 4A). Interestingly, the analysis revealed similar disease terms as for underexpressed DE miRNAs in TMD, but these disease categories were more highly represented for the upregulated than downregulated miRNAs including RA (miR‐24‐3p, miR‐27b‐3p, miR‐29a‐3p, miRNA‐361‐3p, miR‐423‐3p, etc.), arthritis (hsa‐miR‐122‐5p; hsa‐miR‐26a‐5p; hsa‐let‐7a‐5p; hsa‐miR‐125a‐5p; hsa‐miR‐340‐5p, etc.) and OA (hsa‐miR‐122‐5p, hsa‐miR‐26a‐5p, hsa‐miR‐30a‐5p, etc.) (Figure 4B). Although the KEGG pathway analysis did not reveal any significant results, the additional miRWalk pathway analysis showed compelling terms such as oestrogen receptor beta (Erb) signalling pathway (p value = 6.19e‐5, p adj = 0.014; 17 miRNAs including hsa‐miR‐193b‐3p, hsa‐miR‐31‐5p, hsa‐miR‐455‐3p, etc.); integrin signalling pathway (p value = 1.88e‐4, p adj = 0.015; 43 miRNAs including hsa‐miR‐223‐3p, hsa‐145‐5p, hsa‐miR‐18a‐3p, etc.); or MAPK signalling pathway (p value = 5.29e‐4, p adj = 0.018; 45 miRNAs including miR‐223‐3p, miR‐766‐3p, hsa‐miR‐145‐5p, etc.) (Figure 4C).

FIGURE 4.

miRNA Enrichment and Annotation Analysis (miEAA) of 125 upregulated miRNAs in TMD group compared to TMD‐free controls (p adj < 0.05, log2‐fold change > 1). (A) Top 20 enriched categories for miRNAs sorted by p value. (B) Diseases (MNDR)—top 3 subcategories. (C) Pathway analysis—top 20 enriched categories sorted by significance (miRWalk pathways).

3.3. Correlation Between miRNA Expression and Pain Intensity

The correlation analysis revealed some strong and significant (p < 0.05) correlations between the pain variables (intensity and CPG) and salivary miRNA expression. Specifically, pain intensity was highly correlated with hsa_miR_191‐5p (R = 0.71), hsa_miR_425‐5p (R = 0.73), hsa_miR_145‐5p (R = 0.77), and hsa_miR_4286 (R = 0.67), while CPG was significantly correlated with hsa_miR_205‐5p (0.72) (Table S2).

4. Discussion

Despite their high prevalence and negative impact on quality of life, the complex nature of TMDs involving multiple tissues and systems, their variable and overlapping presentations, and unpredictable course together make these disorders perplexing and challenging to understand and treat [1, 2]. Clinical symptoms often do not relate to disease severity or prognosis, and precise molecular mechanisms underlying TMD and the relatedness of the various clinical presentations remain poorly understood. Omics and computational biology combined with non‐invasive access to biospecimens offer powerful opportunities to identify biomarkers and pathologic signatures of TMDs towards the goal of precise diagnostics, therapeutics and monitoring of disease. Here, we investigated DE salivary miRNA in TMDs to identify pathways and disease relatedness of both up‐ and downregulated miRNAs in TMD subjects towards the longer‐term goal of unravelling specific biomarkers and disease endotypes.

In our study, pathway enrichment analyses of 125 statistically significantly upregulated salivary miRNAs in TMD patients compared to non‐TMD healthy controls (p adj < 0.05) identified several key biological processes potentially involved in the TMD pathogenesis. One prominent category is the negative regulation of cell–cell adhesion mediated by cadherin, which is crucial for maintaining tissue integrity and facilitating cellular communication [25]. Disruption of cadherin‐mediated adhesion can cause synovial membrane dysfunction, cartilage breakdown and tissue remodelling in the TMJ, processes that are also observed in joint‐degenerative diseases such as OA [26]. Another enriched process is the negative regulation of neurogenesis, which may be related to the TMD‐associated chronic pain because impaired neurogenesis or altered neural plasticity can affect peripheral and central sensitisation, thus exacerbating pain perception [27]. In addition, miRNAs involved in this pathway may modulate genes associated with neuronal growth, differentiation and synaptic remodelling, resulting in persistent and sometimes disproportionate pain experienced by TMD patients [28]. Finally, negative regulation of chemokine production can also serve as an important pathogenetic mechanism. Because chemokines are key mediators of immune cell recruitment to sites of inflammation, their dysregulation can lead to prolonged inflammation and tissue damage in the TMJ. Those miRNAs identified in saliva may restrain excessive immune responses in chronic inflammatory states like TMD [29]. Interestingly, many of the upregulated miRNAs in TMD patients overlap with those identified in autoimmune and inflammatory joint diseases such as RA and OA (overrepresented), suggesting potential involvement of similar molecular mechanisms for those disorders. These include miR‐24‐3p, miR‐27b‐3p, miR‐29a‐3p, miR‐361‐3p and miR‐423‐3p, which play an important role in RA [30]. Also, miR‐29a‐3p has been implicated in extracellular matrix composition via collagen degradation and expression [30], thus implicating its role in TMJ tissue remodelling. In addition, miR‐27b‐3p and miR‐24‐3p are shown to regulate inflammatory cytokine expression and cartilage metabolism, and their upregulation can reflect compensatory or pathological responses to ongoing joint stress in TMD [31, 32]. Similarly, elevated levels of miR‐27b‐3p have been observed in OA synovial tissues, where it causes synovial fibrosis and inflammation by modulating ECM‐related genes [31]. Finally, the miR‐361‐3p and miR‐423‐3p are involved in inflammatory pathways and may exacerbate joint inflammation and degeneration [32]. Thus, considering miRNA expression patterns in TMD that have been shown to play a role in RA, pro‐inflammatory and tissue turnover pathways further support the hypothesis that TMD may share autoimmune or inflammatory underpinnings with systemic joint and inflammatory disorders.

In addition, several KEGG pathways of under‐expressed miRNAs in TMDs from our study have been associated with TMDs, especially those involving pain mechanisms, inflammation, tissue remodelling and neural signalling. Based on current literature, tryptophan metabolism, linked to serotonin production and pain modulation, may be dysregulated in chronic pain, including TMD [33]. Another pathway linked to TMD pathophysiology is ECM‐receptor interaction that reflects ECM remodelling commonly encountered in TMJ cartilage pathology [34] or the Notch signalling pathway [35], which influences chondrocyte differentiation and joint development associated with cartilage damage in TMD (Figure 3C). Similarly, the miRWalk pathway analysis of overexpressed miRNAs in TMDs revealed pathogenic mechanisms associated with TMJ inflammation, cartilage remodelling, orofacial pain and immune dysregulation. For example, the main detected pathway—the Erb signalling pathway—is implicated in synovial inflammation, matrix remodelling and chondrocyte proliferation in TMJ disorders [36]. Another pathway, involved in cell‐ECM interactions, the integrin signalling pathway, is crucial for cartilage maintenance and TMJ degeneration [37]. Finally, the MAPK signalling pathway has been reported to be strongly related to TMD inflammation, pain sensitisation and joint tissue degeneration [38].

Since previous studies demonstrated significant relationships between salivary miRNA expression patterns and pain intensity scores in other chronic pain conditions [39], we utilised standardised pain assessment CPG scores to explore potential correlations between specific miRNA expression levels and pain severity in TMD subjects. Our findings show significant and strong correlations between specific miRNA and pain indices, providing potential insights into the molecular mechanisms underlying TMD‐related pain (Table S2).

Our study controlled for several variables that may impact miRNA expression levels. These include the age and sex of the subjects, and also the timing of sample collection to minimise the impact of circadian rhythm on miRNA expression levels and profiles. Other potential confounding variables such as BMI, oral hygiene status and stress levels which may impact miRNA expression were not systematically assessed in this study [40]. These factors could potentially influence the observed miRNA expression patterns and should be considered in future larger‐scale studies.

Because this was an exploratory study designed to provide a roadmap for future endeavours and generate new hypotheses, it incorporated a small number of individuals (9 TMD patients and 8 controls), limiting the statistical power for detecting additional significant findings and its broader generalisability. Thus, these results should be interpreted with caution as hypothesis‐generating rather than definitive. Ongoing larger prospective validation studies will be essential to confirm these findings. Also, salivary RNA is present in low abundance, can be influenced by local oral conditions such as gingival inflammation, oral microbiota, diet and oral hygiene, and is prone to enzymatic degradation, resulting in inherent limitations when used for RNA biomarker discovery. These biological and technical challenges may affect the reproducibility and generalisability of salivary RNA‐based assays [16]. Thus, while saliva offers clear advantages as a non‐invasive, repeatable and patient‐friendly medium, careful optimisation of collection, processing and analytical pipelines remains critical to advance salivary miRNAs towards clinical application. We strictly followed optimal procedures in collecting, processing and analysing our samples to minimise the impact of these variables on the results.

Taken together, our findings support the potential of salivary miRNAs as non‐invasive biomarkers for TMD disease endotypes while also providing novel insights into the molecular mechanisms of disease. Importantly, the overlap in miRNA expression between TMD and other inflammatory disorders such as RA or OA reinforces the need for the development of biomarkers that can distinguish TMD from related conditions with similar clinical characteristics. Additionally, longitudinal studies are needed to assess whether miRNA expression patterns correlate with disease progression, response to therapy or transition from acute to chronic states. Future directions should include validation of the miRNA candidates in larger and more diverse patient cohorts (prospective definitive validation), functional studies to determine causality and integration with deep clinical phenotyping and multi‐omic data. Ultimately, understanding the functional consequences of miRNA dysregulation may not only enable specific and sensitive diagnosis but also help identify novel therapeutic targets for specific TMD endotypes.

5. Conclusions

Our study identified DE salivary miRNAs, revealing that several upregulated and downregulated salivary miRNAs in TMD subjects converge on pathways implicated in immune dysregulation, pain sensitisation and joint tissue remodelling. The functional enrichment of these miRNAs in known inflammatory and neuroplasticity pathways reinforces their potential role in TMD pathophysiology. This molecular insight provides a foundational step towards refining the classification of TMD into biologically defined subtypes and lays the groundwork for future biomarker validation and functional studies aimed at enabling stratified diagnostics and targeted therapeutic strategies in TMD.

Author Contributions

K.E.K.‐U., M.P., S.K. and D.T.W.W. conceived and designed the study. S.A., E.R.H., T.A., ShA and M.H. contributed to patient recruitment and sample collection. M.A. and A.M. performed the experimental procedures. K.E.K.‐U. conducted bioinformatic analyses of RNA‐Sequencing data. H.W. and D.E. performed statistical analyses. K.E.K.‐U., H.W., M.P., D.E., S.K. and D.T.W.W. interpreted the data. K.E.K.‐U. drafted the manuscript. All authors (K.E.K.‐U., M.A., H.W., S.A., E.R.H., T.A., ShA, A.M., M.H., S.T., D.E., M.P., D.T.W.W., S.K.) critically revised the manuscript for important intellectual content and approved the final version for submission.

Ethics Statement

Subjects were recruited and consented following study approval by Institutional Review Boards at University of California Los Angeles (UCLA) (22–001676, expiration 12/11/2025) and University of Michigan (HUM00225775, expiration 12/11/2025). The TMD findings were phenotyped using Diagnostic Criteria for Temporomandibular Disorders (DC/TMD) Axis I by calibrated examiners at the two sites.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Table S1: Demographic and clinical characteristics of 9 TMD subjects and 8 non‐TMD healthy control subjects.

Table S2: Correlation analysis for top 30 differentially expressed salivary miRNAs with pain intensity, disability score and chronic pain grade (CPG) (*p < 0.05).

Data Availability Statement

The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.