The study procedures and protocols were reviewed and approved by the institutional review board of the College of Medicine, King Saud University. Written informed consent was obtained from all the participants (IRB number: E-21-6341). The primary treating physician determined the diagnosis for RA based on the American College of Rheumatology and the European League Against Rheumatism (ACR/EULAR) 2010 criteria. The patients in the present study were recruited from those attending the rheumatology outpatient clinics at King Khalid University Hospital, College of Medicine, King Saud University, in the age group of 36–65 years. A total of 60 patients with known RA for a mean duration of 20 years mostly women, were recruited. The patients at the time of the clinical visit did not have any symptoms of early morning stiffness, tender or swollen joints and were considered to be in clinical remission with a (Disease Activity Score-28 (DAS28) <2.6). Control plasma from age-matched volunteers who were otherwise healthy was collected after informed consent was obtained. The baseline demographic information and clinical laboratory parameters, including complete blood count with an ESR, lipid markers, liver function tests, fasting glucose, glycated hemoglobin, and C-reactive protein (CRP), were collected from the patient’s medical records (Table 1). Patients with RA were next stratified according to their ESR, levels >35 mm/h were considered as high and below that considered as low. Additional stratification was carried out using CRP, levels > 10 mg/dL being considered as high and below that considered as low. The study procedures were conducted according to the ethical standards of the Declaration of Helsinki and the International Conference on Harmonization Good Clinical Practice guidelines.

The blood samples were collected by venipuncture into EDTA-containing tubes from all participants and processed (centrifuged at 1,500 × g for 10 min at 4 °C) to obtain clear plasma. The plasma was aliquoted and stored at −80 °C until the day of analysis. Metabolites were extracted from plasma using our previously established protocol (Alfadda et al., 2024). Briefly, cold methanol and chloroform were added to 50 μL of plasma, followed by water and shaking. Equal volumes of chloroform and water were added before centrifugation at 10,000 × g for 5 min. All samples were dried using a vacuum centrifugal evaporator and stored at − 80 °C until further analysis.

Metabolomics profiling was carried out utilizing untargeted metabolomics analysis via LC-HRMS, as previously reported (Sebaa et al., 2023). Metabolites were analyzed using a Waters ACQUITY UPLC system and an Xevo G2-S QTOF mass spectrometer with an electrospray ionization source (ESI) in positive and negative modes (ESI+, ESI−). The metabolites were chromatographed using an ACQUITY UPLC XSelect column (100 mm × 2.1 mm, 2.5 μm) from Waters Ltd., Elstree, United Kingdom. The mobile phases A and B (A: 0.1% formic acid in dH2O—B: 0.1% formic acid in 1:1 v/v MeOH and ACN) were pumped to the column in a gradient mode (0–16 min 95%–5% A, 16–19 min 5% A, 19–20 min 5%–95% A, 20–22 min 5%–95% A) at a flow rate of 300 μL per minute. The MS settings were as follows: the source temperature was 150 °C, the desolvation temperature was 500 °C for both modes, the capillary voltages were 3.20 kV (ESI+) or 3 kV (ESI−), the cone voltage was 40 V, the desolvation gas flow was 800.0 L/h, and the cone gas flow was 50 L/h. In MS^E mode, the collision energy of low and high functions was set to 10–50 V, respectively. As indicated by the seller, the mass spectrometer was calibrated with sodium formate in the 100–1,200 Da range in both ionization modes. Data-independent acquisition (DIA) was performed in continuum mode using a Masslynx™ V4.1 workstation (Waters Inc., Milford, MA, United States).

The raw mass spectrometry data underwent processing through a standard pipeline involving m/z and retention time alignment, peak selection, and quality-based signal filtering, using Progenesis QI v.3.0 (Waters). Multivariate analysis was performed with MetaboAnalyst v.5.0 https://www.metaboanalyst.ca/ accessed on 10 April 2025, applying median, Pareto scaling, and log transformation to normalize the data. Models such as PLS-DA and OPLS-DA were constructed, with model fitness evaluated by R2Y and Q2 metrics. Univariate differences were assessed via (unpaired t-test, FDR, q value <0.05, Fold change (FC) 1.5) using Mass Profiler Professional (MPP) v.15.0. Venn diagrams were generated with MPP, and key features were annotated against the Human Metabolome Database with emphasis on precursor mass and fragmentation; the tolerance was 5 ppm. The workflow used targeted-untargeted annotation to combine precise mass measurements (≤5 ppm) with complete isotopic pattern analysis and diagnostic MS/MS fragmentation data. The internal RT library provided additional retention-time constraints for more than 200 recurring metabolites. The system used multiple evaluation criteria to generate a single identification score which selected the simplest molecular structure without making excessive assumptions. The identification confidence followed Metabolomics Standards Initiative and Schymanski frameworks at Level 2 for high-quality spectral matches and Level 3 for tentative candidate/class identification. The LIPID MAPS shorthand system was used to identify lipids while maintaining reporting at the lowest level possible based on head-group and neutral-loss evidence (class/species without acyl positional or double-bond localization claims). Exogenous substances, including drugs and food additives, were manually excluded from the final feature list.

Two independent and complementary bioinformatics approaches, MetaboAnalyst 5.0 https://www.metaboanalyst.ca/ accessed on 10 April 2025, and Ingenuity Pathway Analysis (IPA, QIAGEN), were used to analyze the metabolomic data to ensure robust interpretation. MetaboAnalyst v5 was employed for statistical analysis, visualization, and pathway-based interpretation of the metabolite data using the Metabolomics Pathway Analysis (MetPA) module based on the KEGG database. Meanwhile, IPA network pathway analysis was applied to identify connections between different metabolites and provide a biological and mechanistic perspective by integrating the metabolomic data with curated biological knowledge bases.

The demographic and biochemical characteristics of the study participants, in the discovery and validation phases, are presented in Table 1. The mean ages of the study participants in the RA and control groups were 49.88 ± 12.05 and 47.52 ± 13.86 years, respectively, and were matched in body weight, BMI, and other laboratory markers. Majority of the patients were taking methotrexate followed by sulfasalazine and only five patients were taking prednisolone as adjunct therapy. Significant differences were noted in the ESR and CRP levels, which were higher in the RA group than the controls.

An initial untargeted metabolomics analysis detected a total of 18,065 mass ion features, including 11,000 in positive ion mode and 7,065 in negative ion mode. Following data preprocessing steps—such as alignment, peak picking, and removal of missing values—a filtering criterion requiring a cut-off percentage of 50% of all samples in at least one condition was applied, reducing the dataset to 12,212 features. To ensure data normality and minimize technical variation, the dataset was median-centered, log-transformed, normalized, and processed using Pareto scaling. The metabolites that distinguished between RA and control are displayed in Figure 1. The OPLS-DA, a supervised multivariate approach between the two groups, RA patients and controls, is shown in Figure 1A. The distinct separation of the RA group from the control suggests that plasma metabolites may be useful for identifying RA. The PLSDA plot also shows separation between study groups, as shown in Figure 1B.

The statistical evaluation between two groups (RA and controls) was performed on 12,212 features using a volcano plot (unpaired T-test, FDR p < 0.05, FC 1.5), which revealed 300 significantly dysregulated metabolites between the study groups, with 147 and 153 metabolites up- and downregulated, respectively (Figure 2). Among these, these 182 were successfully annotated using Human Metabolome Database (HMDB), Metlin MS/MS, LipidBlast, lipidMap, and KEGG. After excluding the exogenous molecules (i.e., drugs, drug metabolites, environmental exposures, etc.), 60 endogenous metabolites were identified between the RA and control groups (Supplementary Table S1). The identified endogenous metabolites in the data set were cross-checked independently with our in-house library, and inosine, cytidine, l-isoleucine, mandelic acid, L-Acetylcarnitine, and 4-Hydroxyphenylpyruvic acid were confirmed. The binary subgroup analysis based on the stratification within the RA group between the high ESR compared to low ESR groups, revealed significant dysregulation in 509 metabolites with 199 and 310 metabolites being up- and downregulated, respectively. 67 endogenous metabolites were identified between patients when stratified with high and low ESR (Supplementary Table S2; Supplementary Figure S1). Of these, three endogenous metabolites were observed to overlap with the significantly dysregulated metabolites that were identified between the RA and control groups (Figure 3). Additionally, a binary comparison based on stratification using CRP levels as high vs. low revealed 6 significantly dysregulated metabolites between the groups. No overlap between these metabolites was noted with the metabolites that were significantly dysregulated in the comparison between RA and controls (60 endogenous).

To elucidate metabolic biomarkers and pathways related to RA patients, an unpaired t-test with FDR (q value < 0.1, FC cut off 1.5) was conducted and revealed 410 significantly dysregulated metabolites, where 178 and 232 metabolites were up (red) and down (blue)-regulated between the two groups. Among these, only 77 endogenous metabolites were identified. The potential biomarkers that differed between RA and controls are shown in Figure 4. OPLS-DA, a multivariate supervised method, is displayed in Figure 1A. A clear separation in the identified metabolites was observed between the groups, indicating that plasma metabolomics can be used to identify RA-related metabolites that serve as biomarkers for distinguishing RA.

A multivariate exploratory ROC analysis was conducted using OPLS-DA as a feature-ranking and classification approach based on the identified common and significantly dysregulated metabolites between the RA and control groups. The AUC of the exploratory ROC curve for the top 3 metabolites was 0.981, as shown in Figure 4A. The frequency plot shows the significantly dysregulated endogenous metabolites between the RA and control groups. OPLS-DA was used as a classification and feature ranking approach for the multivariate exploratory ROC analysis based on the common and significantly dysregulated metabolites identified between the two groups (Figure 4B). From the top 15 metabolites, two metabolites, PGP (i-24:0/PGD2) (AUC = 0.912 and 0.953), and succinyladenosine were upregulated, while expression of L-isoleucine (AUC = 0.931) was downregulated in the RA group, as shown in Figure 4C, respectively.

Pathway topology analysis using MetaboAnalyst revealed several significantly altered metabolic pathways in RA patients compared to controls (Figure 5A). The most enriched pathways included phenylalanine, tyrosine, and tryptophan biosynthesis, pyrimidine metabolism, and tyrosine metabolism, which are all relevant to the chronic inflammatory and immune activation state seen in RA. Additionally, the IPA generated a molecular interaction network between the metabolites and identified humoral immune response, inflammatory response, hematological system development, and function as the network pathways affected with the highest score between the RA and control groups (score of 19 and 8 focus molecules) (Figure 5B; Supplementary Table S3). The pathway suggests that RA-related metabolic changes are tightly integrated with immune and inflammatory networks, highlighting IL6, IL2, IL1, MAPK, and kininogen as central hubs, connecting metabolomics to inflammation, vascular response, and immune regulation in RA. The top canonical pathways included G alpha (q) signaling events (1.39 E^-05, 1.8%), Class A/1 (Rhodopsin-like receptors) that were significantly enriched and predicted to be inhibited in RA patients compared to controls (z-score < 0, –log (B-H p-value) > 1.3) (Figure 5C).

The plasma lipidome in RA showed changes across multiple classes, including glycerophospholipids, ceramides, fatty acids, and acylcarnitines. Lipids are known not only as structural components but also as bioactive signaling molecules, and their systemic imbalance contributes to both local joint damage and systemic comorbidities in RA (Park et al., 2021). It is well known that inflammation underlies the pathophysiology of RA and involves changes in innate and adaptive immune systems and their cells. Multiple cell types, including T cells, B cells, dendritic cells, neutrophils, macrophages, and fibroblast-like synoviocytes, actively contribute to disease activity and progression. Previous studies indicate that lipid metabolism is significantly altered in RA by regulating the activity and function of these cells (Lei et al., 2023). In our analysis, we found notable dysregulation in lipid metabolites, with 19 metabolites decreased and 20 increased. The most substantial changes occurred in glycerophosphoglycerols (PGP), phosphatidic acid (PA), monogalactosyl diacylglycerols (MGDG), diglycerides (DG), phosphatidylethanolamine (PE), ceramides, acylcarnitines, and various fatty acids and sterol derivatives.

Phospholipids, especially glycerophospholipids, are integral to membrane architecture and function as precursors for potent lipid mediators that orchestrate immune activation, inflammation, and tissue remodeling in RA. They are also active precursors and carriers for inflammatory mediators. In our study, levels of several phosphatidylglycerol-containing prostanoid species, PGP (i-24:0/PGD₂), and PGP(PGE₂/i-21:0), were decreased, and the level of PGP(PGF₁α/20:1 (11Z)) was increased. In RA, PGD2 has been shown to suppress Th1-mediated inflammation and also contributes to eosinophilic inflammation and mast cell activation. The conjugation of PGD2 to phospholipids such as PGP may enhance its bioavailability and stability in synovial fluid. Elevated levels of PGP (24:0/PGD2) in RA patients may reflect active lipid mediator signaling and remodeling of membrane phospholipids—a common feature in chronically inflamed joints (Kim et al., 2023). The dysregulation of these lipid species indicates alterations in arachidonic acid metabolism, which is known to be involved in inflammation and RA (Wang et al., 2021; Sano et al., 2020). Similarly, phosphatidic acid, PA (8:0/20:3 (8,11,14)-2OH(5,6)) was dysregulated, highlighting their role as intermediates in both glycerophospholipid biosynthesis and intracellular signaling, which regulate immune cell activation and cytokine production (Fang et al., 2001). We also observed an increase in phosphatidylethanolamine PE (PGE₂/22:6 (4Z,7Z,10Z,13Z,16Z,19Z)), suggesting coupling between inflammatory prostaglandin synthesis and omega-3 fatty acid–derived lipid mediators, potentially reflecting a mixed pro- and resolution-phase lipid signaling environment. Collectively, these lipidomic changes support the concept that in RA, glycerophospholipid remodeling not only reflects heightened immune activation but also directly contributes to the propagation and persistence of synovial inflammation.

PAs are central intermediates in glycerophospholipid biosynthesis and serve as precursors for DG and CDP-diacylglycerols (CDP-DGs). Their elevation in plasma suggests enhanced membrane turnover and lipid signaling activity in circulating immune cells, which may contribute to the chronic inflammatory state characteristic of RA. Notably, CDP-DG species exhibited mixed expression patterns. In contrast, CDP-DG species, such as CDP-DG (22:6-OH/i-20:0) and CDP-DG (20:5-3OH/i-12:0), were significantly downregulated, suggesting impaired systemic availability or utilization of anti-inflammatory ω-3 polyunsaturated fatty acids (Duchez et al., 2019). These shifts point to a pro-inflammatory plasma lipid profile favoring eicosanoid synthesis while limiting resolution mediators. In contrast, phosphatidylinositols and platelet-activating factor (PAF) were decreased, indicative of reduced anti-inflammatory lipid signaling. These shifts collectively represent an eicosanoid-dominant, pro-inflammatory lipid remodeling in the plasma compartment of RA.

In addition to these, we identified multiple carnitine species that exhibited differential expression, with significant upregulation of octadecenoylcarnitine, and L-acetylcarnitine and a downregulation of 13-(3,4-dimethyl-5-pentylfuran-2-yl)tridecanoylcarnitine, revealing distinct perturbations in fatty acid transport and mitochondrial metabolism in RA. Alterations in carnitine metabolism in RA reflect a broader disruption in systemic energy homeostasis and immune cell metabolism. Carnitines are essential for the mitochondrial β-oxidation of long-chain fatty acids, serving as shuttles that transport acyl groups across the mitochondrial membrane. They play a central role in mitochondrial fatty acid oxidation, energy homeostasis, and immunometabolic regulation, and their dysregulation in the plasma of RA patients indicates the presence of a systemic pathophysiology (Xiang et al., 2025). The striking elevation of long-chain acylcarnitines, particularly octadecenoylcarnitine, likely reflects an imbalance between lipid influx and mitochondrial processing capacity. Long-chain acylcarnitines have been noted to increase Th17-associated cytokine production in T cells due to compromised β oxidation, which in turn promotes inflammation (Nicholas et al., 2019). Accumulation of long-chain acylcarnitines may indicate incomplete fatty acid oxidation, a metabolic phenotype associated with inflamed immune cells and mitochondrial stress. This metabolic inflexibility is consistent with reports of impaired oxidative phosphorylation and elevated glycolytic dependency in RA T cells and macrophages (Weyand et al., 2017). An increase in the levels of circulating long-chain acylcarnitines has also been found in other chronic inflammatory diseases, including chronic heart failure patients (Ahmad et al., 2016). Mitochondrial inefficiency and impaired fatty acid oxidation may contribute to systemic energy deficits. At the same time, elevations in L-acetylcarnitine, a medium-chain acylcarnitine, could signal a shift towards glycolysis and acetyl-CoA overflow (Izzo et al., 2023), consistent with the metabolic reprogramming seen in activated immune cells (Andonian et al., 2022). In animal models, L-acetyl carnitine was found to inhibit the expression of inflammatory factors and antioxidants to suppress the development of atherosclerosis (Wang et al., 2020). Our findings are in line with et al., who identified increase in seven acylcarnitine metabolites with lower disease activity (Hur et al., 2021).

The significant changes in lipid metabolites in our study underscore inflammation-driven metabolic reprogramming, oxidative stress, and immune dysregulation linked to RA. These findings demonstrate the profound reorganization of lipid metabolism in RA, aligning with previous multi-omics studies that connect lipid disturbances to immune dysregulation, synovial inflammation, and joint destruction.

Amino acids serve essential roles in immune cell metabolism, inflammation, and tissue remodeling. In RA patients, we observed significant perturbations in crucial amino acids, bioactive peptides, post-translationally modified residues, and oxidative derivatives, pointing toward interconnected biochemical pathways that sustain inflammation and joint damage. Changes in the levels of amino acids have been linked to abnormal immune response in RA. Previous metabolomic studies suggest a direct link between amino acid metabolism and T cell and macrophage responses by promoting and modulating inflammation, which could be involved in RA pathogenesis (Ren et al., 2017). Notably, in our study, branched-chain amino acids (BCAAs), such as L-isoleucine, were markedly decreased. BCAAs are critical energy substrates for immune cells, and their depletion suggests increased consumption by inflammatory leukocytes, especially activated macrophages and lymphocytes, which rely on these amino acids to support anabolic pathways necessary for cell proliferation and inflammatory mediator synthesis (Dimou et al., 2022). Statistically significant increases in N2-acetyl, N6-methyllysine, a double-modified product of lysine acetylation and methylation, were observed between patients with RA and controls. Lysine is an essential amino acid that has been implicated in RA pathogenesis. Post-translational modifications of lysine are abundant in histones and transcriptional regulators and are key determinants of gene expression. Elevated circulating levels likely arise from increased turnover of nuclear proteins in activated immune and synovial cells. This is in concert with the fact that RA is characterized by persistent epigenetic reprogramming, particularly histone acetylation and methylation that locks fibroblast-like synoviocytes into a pro-inflammatory, tissue-invasive phenotype (Yang et al., 2022).

We additionally identified elevated levels of 2-(β-D-mannopyranosyl)-L-tryptophan (C-mannosyl-tryptophan), a C-glycosylated post-translational modification of tryptophan. C-mannosylation in the endoplasmic reticulum is known to play critical roles in the folding, sorting, and/or secretion of substrate proteins. It is produced in part by degradation of C-mannosylated proteins through the autophagic pathway in cells and is considered a product of protein C-mannosylation turnover (Minakata et al., 2021). Previous studies have shown that it is a circulating biomarker for kidney function, peripheral artery disease, and ovarian cancer. An interesting finding is that the C-mannosyl-tryptophan modification is important for, be required for the intracellular transport of IL-21R (Siupka et al., 2015). IL-21 is a key cytokine involved in the activation and proliferation of B cells and T cells, including Th17 cells and follicular helper T cells, both of which are implicated in RA pathogenesis. IL-21 can promote inflammation by increasing the production of other pro-inflammatory cytokines and by impairing the function of regulatory T cells, and contributes to the inflammation and tissue damage characteristic of the disease (Shbeer and Ahmed, 2024). Studies have shown that IL-21 levels correlate with disease activity and radiographic damage in early RA (Dinesh and Rasool, 2018; Rasmussen et al., 2010). Levels of cystathionine sulfoxide and S-(3-oxo-3-carboxy-n-propyl) cysteine, which are both derivatives of cystathionine, an intermediate compound in the transsulfuration pathway, were decreased in patients with RA compared to controls. These are sulfur-containing amino acids formed through the oxidation of cystathionine by ROS. These compounds are intermediates in the methionine–cysteine–glutathione pathway, a central axis for redox balance (Mateen et al., 2016; Pajares et al., 2015). The decrease in their levels points to increased evidence of oxidative stress in patients with RA.

Beyond metabolic and oxidative stress pathways, our findings also implicate the neuroimmune interface in RA. We observed decreased levels of O-acetylcholine, a key neurotransmitter in the cholinergic anti-inflammatory pathway. Acetylcholine inhibits NF-κB activation and reduces pro-inflammatory cytokine release from immune cells (Tracey, 2002). Its suppression may remove an important control on systemic inflammation. On the other hand, we also observed an elevation in the levels of dynorphin B (10–13), a breakdown metabolite from the dynorphin metabolic pathway. An increase in its levels points to activation of the endogenous opioid system that may provide short-term analgesia in chronic pain syndromes and RA (Millan, 2002). Dynorphin acts on opioid receptors to produce analgesia, but it also has non-opioid effects, including activating bradykinin receptors. Bradykinin receptors (B1 and B2) are involved in inflammatory pain and hyperalgesia. Interestingly, our metabolomic analysis also identified increased levels of bradykinin and bradykinin fragments in our data set, in line with previous studies (Vanarsa et al., 2020).

Plasma levels of the dipeptides cysteinyl-serine, cysteinyl-alanine, phenylalanylvaline, and tyrosyl-arginine were increased while Arginylarginine levels were lowered in patients with RA compared to controls. RA is known to increase levels of multiple cysteine/serine proteases, which lead to heightened extracellular proteolysis, in turn generating low-molecular-weight circulating peptides. This aligns with RA peptidomics studies showing an expanded low-molecular-weight peptide/dipeptide signature in serum/plasma compared with controls. Tyrosylarginine is a dipeptide which was identified as an opiod with analgesic properties. A decrease in its levels in RA patients suggests the heightened pain response in these patients (Ueda, 2021). A decrease in arginylarginine may reflect increased consumption of L-arginine by inducible nitric-oxide synthase in activated myeloid and stromal cells, in line with previous studies that documented robust NO/iNOS signaling in RA.

We found several nucleotide- and nucleoside-related metabolites altered in the plasma of patients with RA. Persistent activation of immune cells due to hypoxia and chronic inflammation in RA leads to accelerated DNA and RNA turnover in metabolizing cells, influencing circulating nucleotides in plasma (Kishikawa et al., 2021). In our study, we identified intermediates of purine/pyrimidine metabolism inosine, cytidine, and AICAR, and modified nucleosides succinyladenosine, N6-Methyladenosine, and 1-methylguanine. Levels of succinyladenosine and 1-methylguanine were elevated, while the others were down. Succinyladenosine, a modified adenosine bound to succinate, sits on the adenylosuccinate lyase arm of de novo purine biosynthesis and accumulates when flux through this pathway is high. Adenosine and succinate are known to act as signaling molecules and modulate immune responses, with elevated levels linked to inflammatory conditions. In the setting of chronic immune activation, a rise in circulating succinyladenosine is therefore consistent with heightened purine synthesis/turnover in proliferating immune and stromal cells (Huang et al., 2024; Magni and Ceruti, 2020). Inosine, an adenosine-pathway metabolite with anti-inflammatory actions via adenosine receptors, was also reduced; lower inosine levels fit enhanced catabolism during immune activation and the loss of an adenosinergic anti-inflammatory tone. RA and other inflammatory models document inosine’s capacity to limit cytokine output through A2A/A3 signaling (Kim and Jo, 2022).

Conversely, depletion of certain nucleosides, such as inosine and cytidine, may reflect their rapid utilization during immune-cell proliferation or altered enzymatic processing, such as increased cytidine deaminase activity. Several intermediates with immunoregulatory potential were lower. AICAR, an intermediate of de novo purine synthesis, was decreased, compatible with brisk flux toward IMP and a relative loss of AMPK control that otherwise restrains inflammatory programs (Daignan-Fornier and Pinson, 2012). We also saw evidence of RNA modification/turnover changes. N⁶-methyladenosine (m⁶A), the dominant internal mRNA modification in plasma, was decreased. It is known to modulate stability/translation of inflammatory transcripts; emerging rheumatology work links m⁶A machinery to RA pathogenesis, so altered circulating m⁶A nucleosides are compatible with shifted RNA turnover during immune activation (Wu et al., 2021).

In contrast, 1-methylguanine was elevated. Modified purine bases/nucleosides (including 1-methylguanine/1-methylguanosine) are released during RNA turnover and are not efficiently salvaged; increases in biofluids are widely used as systemic markers of heightened RNA metabolism. The levels of the pyrimidine nucleoside, cytidine, were also observed to be reduced in patients with RA compared to controls. Beyond serving RNA synthesis, cytidine is consumed by activated cells and is regulated by cytidine deaminase. Clinical RA cohorts have previously shown elevated serum CDA activity, providing a plausible enzymatic route to lower circulating cytidine during active immune remodeling (Fatima et al., 2025; Gatarek et al., 2020). Taken together, an increase in succinyladenosine and 1-methylguanine coupled with a decrease in AICAR, inosine, m⁶A, and cytidine supports accelerated de novo purine synthesis and increased RNA turnover in RA. The relationship between m6A enzymes, immune cells, and RA suggests that m6A modification offers evidence for the pathogenesis of RA.

Patients with established disease or in remission are generally monitored using markers of inflammation, namely, ESR and CRP. An elevated ESR ≥ 28 mm/h and/or positive rheumatoid factor (RF) test are used to support the diagnosis of RA, although the levels for remission are lower (ESR < 20 mm/h for men and < 30 mm/h for women) (Wolfe and Michaud, 1994). As persistent systemic inflammation is expected to drive alterations in energy metabolism, amino acid turnover, lipid remodeling, and oxidative stress–related pathways observed in RA plasma profiles, stratification using ESR provides a biologically relevant variable for understanding the changes in metabolomic analysis. The subgroup analysis based on the stratification by ESR levels revealed significant dysregulation in a number of metabolites. From among these, levels of three plasma metabolites, namely, cysteinyl-serine, tyrosyl-arginine, and N2-acetyl, N6-methyllysine levels were observed to overlap with the significantly dysregulated metabolites identified in RA vs. control group. We observed that levels of cysteinyl-serine were increased both in patients with RA compared to controls and in patients with high ESR, suggesting a consistent change with both disease state and systemic inflammation. On the other hand, levels of tyrosyl-arginine and N2-acetyl, N6-methyllysine showed an opposite regulation, being increased in patients with RA compared to controls and were decreased in patients with high ESR levels. The dysregulation in the dipeptides cysteinyl-serine and tyrosyl-arginine point to alterations in systemic proteolysis consistent with the upregulation of cysteine and serine proteases in RA that increases chronic inflammation, which in turn is measured by ESR (Solau-Gervais et al., 2007). N2-acetyl, N6-methyllysine a doubly modified lysine is part of the lysine metabolic pathway, which has been reported to be associated with early RA patients who achieved remission. The increase in its levels in low ESR patients, points to a potential ongoing systemic inflammation, increased transcriptional activity and histone acetylation/methylation dynamics, activating an epigenetic activation/turnover signature in RA (Lu-Culligan et al., 2023). Similar to our results, previous studies also demonstrated an association between N2-Acetyl, N6-methyllysine with lower RA disease activity (Teitsma et al., 2018; Pinals et al., 1981). These metabolites together indicate a possible composite read of mitochondrial stress, proteolysis, and epigenetic reprogramming that converge on IL-6 and the hepatic acute-phase axis that influences chronic inflammation and ESR in patients with RA. On the other-hand, stratification with CRP did not reveal any metabolites that overlapped with the metabolites identified between RA and controls.

ROC analysis identified a panel of ten metabolites that showed the highest discriminatory power between RA and controls with an AUC of 0.993. These metabolites included succinyladenosine, PGP (24:0/PGD2), cysteinyl-serine, N-(1-Deoxy-1-fructosyl) phenylalanine, 2-(beta-D-Mannopyranosyl)-L-tryptophan, cysteinyl-alanine, CDP-DG (PGE2/i-19:0), CDP-DG (i-19:0/18:1 (12Z)-2OH(9,10)), L-Isoleucine and 4-Amino-1H-imidazole-2-carboxamide (Figure 4A). Bioinformatics and network Pathway analysis of the differentially regulated metabolites using the Ingenuity Pathway Analysis (IPA, QIAGEN), identified that the dysregulated the humoral immune response, inflammatory response, hematological system development and function pathways. The actions of these metabolites centered around regulating MAPK, PKC, IL6, IL1, and IL2, which are key inflammatory and signaling molecules, suggesting involvement in immune signaling, vascular function, and cytokine regulation consistent with previous studies (Ding et al., 2023). These findings are consistent with established RA pathophysiology, which is driven by interconnected TNF-α/NF-κB, IL-6/JAK–STAT, and MAPK signaling cascades that sustain chronic inflammation and immunometabolic dysregulation (Ucci et al., 2023). Recent studies have also identified Wnt signaling as an important immuno-inflammatory and metabolic regulator in RA, potentially intersecting with the dysregulated signaling networks identified in our metabolomic analysis (Riitano et al., 2025). The network pathway also showed the involvement of IL-21, a cytokine involved in stimulating follicular T helper cells, which was identified in our previous proteomic analysis (Masood et al., 2025). Canonical pathway analysis of the significantly altered metabolites revealed several biologically relevant processes associated with rheumatoid arthritis. Notably, G alpha (q) signaling events and Class A/1 (Rhodopsin-like receptors) pathways were significantly enriched and predicted to be inhibited in RA patients compared to controls. Aside from these pathways immune–metabolic dysregulation in RA has also been noted to be tightly coupled to epigenetic remodeling, dysregulated protein turnover, and sustained immune activation, which may contribute to the metabolite changes observed in the context of cellular stress and tissue remodeling (Riitano et al., 2023). Our results similarly show that these pathways are primarily involved in G-protein–coupled receptor signalling, which plays a critical role in immune cell activation, cytokine release, and inflammation. The observed inhibition suggests a possible disruption or downregulation of GPCR-mediated signalling in the RA metabolic landscape.

The findings from our study need to be interpreted in the context of our cohort that comprised of patients with long-standing rheumatoid arthritis receiving chronic therapy as part of their standard of care management that may confound the interpretation. Moreover, treatment variables were not included as covariates based on the heterogeneity within our cohort. Lastly our study employed an untargeted metabolomics approach aimed at exploratory and hypothesis generating allowing the identification of broad metabolic and pathway-level alterations in established RA. Future studies using larger cohorts and targeted analysis will be required to validate these findings.