AbstractOsteoarthritis (OA) is a degenerative joint disease that can be aggravated by synovitis and synovial immune disorders (SID). However, the role of synovial SID-related genes in OA synovium remains poorly understood. OA synovial and peripheral blood datasets were obtained from the GEO database (https://www.ncbi.nlm.nih.gov/). Immune-related genes (https://reactome.org/) showing differential expression in peripheral blood were identified as immune disorder genes. Subsequently, differentially expressed immune disorder genes in OA synovium were further identified as SID genes. The Venn diagram, random forest, SVM-RFE algorithm, and multivariate analysis were employed to determine SID-related hub genes in OA synovium. Using the identified hub genes, we constructed and validated a diagnostic model for predicting OA occurrence. The correlation between hub gene expression and immune-related modules was explored using CIBERSORT and MCP-counter analyses. We identified three SID-related hub genes (ACAT1, SPHK1, and ACACB) in OA synovium. The diagnostic model incorporating these hub genes demonstrated reliable predictive accuracy (AUC = 0.939). Through qPCR analysis, we quantitated the expression levels of the hub genes and confirmed that three hub genes could serve as novel biomarkers for OA patients (AUC = 0.960). Furthermore, we observed a significant correlation between the expression of these hub genes and immune cell infiltration, as well as inflammatory cytokine levels in OA synovium. Our findings suggest that three SID-related hub genes have the potential to serve as diagnostic biomarkers for OA patients. These genes are associated with immune disorder and contribute to immune alterations within the OA synovium.
IntroductionOsteoarthritis (OA) is a degenerative joint disease characterized by clinical manifestations, including joint pain, functional impairment, and structural alterations1. Various factors, such as age, dyslipidemia, synovitis, and genetic factors2,3,4, contribute to the progression of OA. With over 250 million estimated cases worldwide, OA poses a significant socioeconomic burden each year5,6,7. Currently, available treatment options for early-stage OA are limited to nonsteroidal anti-inflammatory drugs and physical therapy, aimed at slowing down disease progression8. However, as OA advances, joint replacement becomes inevitable9. Unfortunately, early symptoms of OA often go unnoticed, causing some patients to miss the optimal window for early intervention10. While previous studies have reported potential early diagnostic biomarkers for OA11, their testing may be challenging as they require samples from joint components (such as cartilage, synovium, meniscus, or subchondral bone) that are not suitable for routine screening purposes12. Therefore, the development of a reliable blood marker for early diagnosis of OA would hold great clinical utility.Mounting evidence suggests that OA is not solely a mechanical disease characterized by meniscus degeneration, subchondral sclerosis, and cartilage wear, but also involves immune system dysregulation and inflammation2,4,13. In OA, pathological changes in the synovium may precede those in the cartilage, with the severity of synovial lesions worsening as the disease progresses14,15,16. Synovial immune dysfunction has been identified as a crucial factor in OA pathogenesis17.Various immune cells within PBMCs (such as T cells, B cells, and macrophages) and inflammatory factors (including IL-2, IL-10, and IFN-γ) are recruited to the synovial tissue, triggering an inflammatory cascade that exacerbates synovitis and contributes to OA progression18,19. Thus, immune dysregulation within the synovial region is marked by substantial infiltration of immune cells and inflammatory mediators20. The affected synovial tissue can directly or indirectly drive changes in the polarization of blood-derived macrophages21, as well as alter the activation and differentiation potential of T cells22, B cells23, and other immune cells. While these observations highlight a close interplay between synovial immune dysregulation and peripheral blood cells, the expression levels of synovial immune-related genes in PBMCs and their clinical importance remain undefined.In this study, we investigated the SID-related genes that differentially expressed both in blood and synovium samples from OA patients. Furthermore, we explored whether SID related hub genes could serve as diagnostic markers for predicting the occurrence of OA. Our findings demonstrate that the three SID related hub genes can serve as reliable blood markers for diagnosing OA. Moreover, the expression of these hub genes may be implicated in the dysregulation of synovial immune responses in OA patients.MethodProcessing of raw dataThe workflow chart of this study is shown in (Fig. 1). Four synovium (GSE12021, GSE29746, GSE55235, GSE55457) and PBMC (GSE48556, GSE63359) datasets of patients with OA were downloaded from the GEO database (https://www.ncbi.nlm.nih.gov/geo/). The detailed information for these six GEO datasets is present in (Table 1). Gene annotation was performed according to the annotation files provided by the corresponding platforms. Probes without matching gene symbols were removed. For multiple probes matching the same gene symbol, the mean value was used. The SVA package was used to remove the batch effect and then obtained an integrated synovium-related database (training set, containing 40 OA and 40 healthy synovial samples). We obtained the immune-related genes from the Reactome website (https://reactome.org/) and these genes that are differential expressed in peripheral blood in OA patients were identified as SID genes.Fig. 1The work flow chart.Full size imageTable 1 Summarize GEO data information included in the study.Full size tableGO/KEGG enrichment and protein-protein interaction (PPI) network analysis of immune disorder genesGene ontology (GO) and Kyoto encyclopedia of genes and genomes (KEGG) pathway analysis was performed using the ‘clusterProfile’ package in R software. All pathways with statistical significance (p < 0.05) were retained. PPI network analysis was performed under the guidance of the String website, which is an open-access website exploring functional protein associations based on the genes we input.Identification of hubs genes of SID in OA synoviumDifferential expression analysis was performed between OA and control (CT) group, and the differential expressed genes of SID in OA synovium was used for next analysis. The ‘ggplots2’, ‘heatmap’, and ‘ggpubr’ packages were used to create volcano, heatmap clustering, and Box plots, respectively. Subsequently, a random forest classifier consisting of 1000 decision trees was constructed, and ten-fold cross-validation was performed using the ‘randomForest’ R package. The potential hub genes of SID were ranked using the random forest function and the forest plot was drawn using the ‘forestplot’ package. The support vector machine recursive feature elimination (SVM-RFE) was used to remove features with relatively low prediction values in each iteration. Therefore, potential hub genes were ranked from most to least important. The intersection of the top 10 potential hub genes retained by random forest and SVM-RFE analyses was identified as the potential hub genes.To verify the expression of potential hub genes of SID in OA synovium, we collected the synovial tissues from 39 OA patients who underwent knee joint replacement surgery and 7 normal individuals with acute anterior cruciate ligament (ACL) rupture who underwent ACL reconstruction (testing set). The study was approved by the Ethics Committee of the Third Affiliated Hospital of Anhui Medical University (Approval Number: 2023-022-01), and all patients signed the tissue collection informed consent form. The RT-qPCR was conducted as manufacturer’s instructions. Briefly, total RNA from synovial tissue was extracted using TRIzol reagent (Invitrogen, USA). The RNA concentration was detected with a NanoDrop spectrophotometer. The RNA samples were transcribed into cDNA using the PrimeScript RT reagent kit and then amplified using the TB Green Premix Ex Taq II reagent kit (TaKaRa, Japan). Primer sequences of mRNA were shown in (Table S1). The primers of all mRNAs were purchased from Sangon Biotech https://www.sangon.com/ (China). GAPDH was used as an endogenous control to estimate gene expression levels, and data analysis was performed by standardizing the relative expression levels. The 2−ΔΔCT method was used for data analysis. Each sample was checked three times.Construction of OA diagnostic modelBased on the expression of hub genes, we calculated the area under the receiver operator characteristic (ROC) curve to evaluate whether the identified hub genes of SID have diagnostic value in OA. Using ‘ROCR’ package, each point on the ROC curve was generated by training a classifier on all included samples. Then, a standard bootstrap procedure further verified the area under the ROC curve (AUC). Nomogram was conducted to predict the possibility of OA. The calibration curve was drawn to present the stability of the model, and the decision curve analysis was generated to demonstrate whether the constructed model is beneficial to OA patients. All the results were re-verified based on collected clinic data.Correlation between hub gene expression and immune microenvironment in OA synoviumThe CIBERSORT algorithm (https://cibersortx.stanford.edu/) was used to calculate the infiltrative degree of 22 types of immune-related cells in OA synovium24. Using the ‘relative’ and ‘absolute’ methods available on the analysis website CIBERSORT (https://cibersortx.stanford.edu/), the percentage of local immune cell infiltration in OA synovial tissue was analyzed by the CIBERSORT module. Gene expression deconvolution was performed on RNAseq TPM level data of clinical samples. Based on CIBERSORT analysis, we derive the correlation of verified hub genes and 22 immune-related cell infiltration. Then, the ‘MCP-counter’ package quantified the relationship between verified hub genes and the absolute abundance of 9 immune cells in the peripheral blood of OA.Consensus clustering and WGCNA analysisUnsupervised consensus clustering analysis was performed using the “ConsensusClusterPlus” software. The maximum cluster gene number in OA samples was set to 10. The top 5000 most variable genes represented by absolute deviation median were used for sample clustering. Unsupervised consensus clustering was used to cluster OA samples and the optimal number of clusters was selected. In each of the 1000 resampling iterations of the consensus clustering, ward linkage was used as the clustering method and Euclidean distance was used as the distance metric. Weighted gene co-expression network analysis (WGCNA) focuses on gene sets rather than individual gene expression and is a common method for understanding gene correlation patterns between different phenotype traits25,26. In the WGCNA analysis, OA expression data can be used to build a powerful scale-free network to identify hub biomarkers for mechanism evaluation and clinical diagnosis. First, the expression of OA synovial genes was standardized. The matrix was then converted into an adjacency matrix using the power adjacency function, which represented the strength of the connection between any two genes. The soft-thresholding parameter for the power adjacency function was selected based on the scale-free topology (SFT) criterion, which is a necessary condition for network construction. The best threshold parameter value with a model fitting saturation > 0.85 was accepted based on the SFT criterion recommendation. The “tree” method of “deepSplit” was used to determine the modules, which identified modules with at least 10 genes. Each module was characterized by feature genes, which were generated by WGCNA. Fisher’s exact test was used to detect enriched genes in each module of OA synovial tissue. Gene set enrichment analysis (GSEA) was used to evaluate the enrichment score of DEGs among the three clusters.Statistical analysisStatistical analysis was performed using SPSS software (version 23.0). All raw data processing and analysis were performed using R software (version 4.2.1). Kruskal-‐Walli’s test or Wilcoxon test was used to detect significant differences between the CT group and the OA group. P < 0.05 was considered statistically significant.ResultsIdentification and enrichment analysis of 32 immune disorder genes in OAWe performed differential gene expression analysis based on blood-related transcriptome profiles (GSE48556, GSE63359) in patients with osteoarthritis (OA). Utilizing Venn analysis, we successfully identified 32 immune disorder-related genes that exhibited differential expression in the peripheral blood of OA patients (Fig. 2A, Table S2). Subsequently, a comprehensive GO/KEGG analysis was conducted on these 32 immune disorder-related genes (Fig. 2B, Table S3), revealing their significant enrichment in bioprocesses associated with lipid metabolism, such as “fatty acid metabolic process” and “lipid modification”. Additionally, the molecular function of the immune disorder-related genes showed a significant enrichment in “acyltransferase activity”. Among the KEGG pathways, the most prominently enriched pathways were “Phosphatidylinositol signaling system” and “Inositol phosphate metabolism”. The protein–protein interaction (PPI) network of the 32 immune disorder-related genes is depicted in (Fig. 2C).Fig. 2Identification of synovial SID-related genes in OA. (A) Identification of SID-related genes both in two peripheral blood components and immune. (B) GO/KEGG analysis of identified SID-related genes. (C) Protein-protein interaction (PPI) network analysis of identified SID-related genes. The larger the yellow oval, the stronger the interaction of the node, with a corresponding increase in the number of edges connected to it. Six genes without interactions were removed. (D) Expression of SID-related genes between OA and healthy control. (E) Identification of synovial SID-related genes in OA. (F) Heat map of expression heterogeneity of synovial SID-related genes between OA and healthy control.Full size imageAfter addressing batch effects in four transcriptome profiles related to OA synovium (Fig. S1), we applied these 32 immune disorder-related genes for differential expression analysis in OA synovium. Our findings revealed a significant overall down-regulation in the expression levels of immune disorder-related genes (Fig. 2D). Furthermore, we identified 13 differentially expressed genes (DEGs) associated with immune disorders in OA synovium (Fig. 2E). Heatmap analysis based on an integrated dataset demonstrated heterogeneity in the expression patterns of these 32 genes between healthy (CT) and OA synovium samples (Fig. 2F).Three SID-related hub genes can predict OA occurrenceWe employed random forest analysis (Fig. S2A) and SVM-RFE algorithm (Fig. S2B) to rank the 32 immune disorder-related genes based on their importance in OA synovium. The overlapping genes obtained from the top 10 ranked genes by both algorithms were further subjected to multivariable logistic regression analysis (Fig. S2C). Through Venn analysis, we ultimately identified three hub genes of synovial immune disorder (SID): Acetyl-CoA Acetyltransferase 1 (ACAT1), Sphingosine Kinase (SPHK1), and Acetyl-CoA Carboxylase Beta (ACACB) (Fig. 3A). Notably, these three genes exhibited a significant correlation in their expression levels (Fig. 3B).Fig. 3Identification of hub genes of SID in OA. (A) Three hub genes were identified by random forest, SVM-RFE, and multivariate analysis. (B) Correlation between three expressed hub genes in OA synovium. (C) Expression levels of the three hub genes were quantified by qPCR analysis in health and OA synovium (n = 7, n = 39).Full size imageSubsequently, we quantified the expression levels of these three hub genes in OA synovium using RT-qPCR (Fig. 3C) in testing set. Among them, two genes were found to be down-regulated while one gene was up-regulated. Based on the expression levels of these three hub genes in the training and testing sets, we constructed a diagnostic model. The nomogram generated for the training set (Fig. 4A) and testing set (Fig. 4B) demonstrated that the probability of OA occurrence could be determined by the total score calculated using the expression levels of the identified hub genes. The AUC values indicated that the diagnostic model had excellent predictive performance for OA occurrence in both training set (Fig. 4C, AUC = 0.939) and testing set (Fig. 4D, AUC = 0.960). Additionally, the well-calibrated calibration curves further confirmed the reliability of the predictive model (Fig. 4E,F). Compared to individual hub genes, the diagnostic model incorporating these three SID-related hub genes provided improved benefits and diagnostic ability for OA patients, as evidenced by the higher AUC values (Fig. 4G,H, Fig. S2D). We validated the reliability of the ROC curve using bootstrap sampling from OA samples (n = 100 bootstraps) (Fig. S2E), and the distribution ranges of AUC, sensitivity, and specificity for the diagnostic model are presented in (Fig. S2F–H), respectively.Fig. 4Construction and validation of nomogram model. (A) Utilized three hub genes of SID, we generated nomogram for the training set and (B) testing set. (C) Receiver operating characteristic (ROC) curve of nomogram for OA diagnosis accuracy in training set and (D) testing set. (E) Calibration curve of the diagnostic model in training set and (F) testing set. (G) Decision curve analysis (DCA) in training set and (H) testing set reveals that nomogram model will be more benefits for patients with OA.Full size imageThe correlation between SID-related hub genes and immune pattern in OA synoviumWe explored the correlation between the expression levels of three hub genes immune cell infiltration, and inflammatory cytokines using CIBERSORT and mcp-counter analysis (Fig. 5). We found that ACAT1 was significantly correlated with the levels of 10 out of 22 immune cell infiltrations, 5 out of 24 inflammatory cytokines, and 2 out of 9 immune cell populations. For SPHK1, we observed significant correlations between its expression level and 5 out of 22 immune cell infiltrations, 9 out of 24 inflammatory cytokines, and 2 out of 9 immune cell populations. The expression level of ACACB was significantly correlated with 2 out of 22 immune cell infiltrations, 6 out of 24 inflammatory cytokines, and negatively correlated with the abundance of monocytes.Fig. 5The correlation between three hub genes expression and immune model. (A) Correlation between hub genes expression and 22 immune-related cell infiltration (CIBERSORT). (B) Correlation between hub genes expression and the abundance of 9 immune-related cells (MCP-counter). (C) Correlation between hub genes expression and the level of 24 inflammatory cytokines. *p < 0.05, **p < 0.01, ***p < 0.001.Full size imageComparison of age, gender, and immune characteristics among the three groupsUtilizing consensus cluster analysis based on the expression levels of the three hub genes, we successfully clustered the OA patients into three distinct groups (Fig. 6A). To examine the relationship between the hub genes and clinical factors, we performed analyses using the “ggpubr” package in R software. As shown in (Fig. 6B), there were significant heterogeneities observed in the expression of the three hub genes, as well as gender, age, and the likelihood of OA occurrence among the three clusters. Although no significant difference in age was observed across all three clusters (Fig. 6C), there was a variation in the likelihood of OA occurrence (Fig. 6D, p < 0.05). Notably, the number of female patients appeared to be higher than that of males in the younger group (Fig. 6E, cluster 1, p = 0.03). In the elderly group (cluster 2), the expression levels of ACAT1, SPHK1, ACACB, CSF1, and IFNG were upregulated, while CD4, IL3, IL1A, and TGFB3 exhibited downregulation (Fig. 7A,B). Additionally, T cells, B cells, monocytes, and myeloid dendritic cells were found to be decreased in cluster 2 (Fig. 7C).Fig. 6The results of unsupervised consensus clustering based on identified hub genes. (A) Consensus matrix plots depicting consensus values on a white-to-blue color scale ordered by consensus clustering when three clusters were selected. (B) The heterogeneity of gene expression is related to age and gender. Age (C) and age probability (D) difference between three clusters. (E) Gender difference between three clusters.Full size imageFig. 7(A) Expression differences of three hub genes between three clusters. (B)The difference of immune cells abundance between three clusters. (C) The difference of inflammatory factor levels between three clusters.Full size imageConstruction of co-expression networkA total of 40 osteoarthritis (OA) samples underwent quality control analysis, and all samples were included in subsequent analyses (Fig. S3A). Co-expression modules were constructed using the dynamic tree cutting methodology, and an unscaled network was generated with a soft threshold value of 5 (Fig. S3B). The identification of statistically significant co-expression modules was performed using weighted gene co-expression network analysis (WGCNA) based on optimal dynamic tree cutting and hierarchical clustering (Fig. 8A,B). Notably, the pink module showed the strongest positive correlation with cluster 1, while the greenyellow module exhibited a positive correlation with cluster 2, and the green module was positively correlated with cluster 3. Subsequently, GO/KEGG analysis was conducted for genes within these three modules (Fig. 8C). The genes in the pink module were mainly involved in the biological process of “positive regulation of cytosolic calcium ion concentration,” while the genes in the greenyellow module were significantly enriched in the KEGG pathway of “AMPK signaling pathway.” Within the green module, the genes primarily participated in the biological process of “glycosphingolipid metabolic process” (Fig. 8C). Furthermore, we explored the enriched pathways using gene set enrichment analysis (GSEA) based on differentially expressed genes (DEGs) among the three clusters (Fig. S4). Our findings revealed that the “phospholipase c activating G protein coupled receptor signaling pathway,” “adaptive immune response,” and “leukocyte mediated cytotoxicity” showed a positive correlation with cluster 1 and a negative correlation with cluster 2.Fig. 8Identification of key modules correlated with three clusters through WGCNA. (A) Clustering dendrogram of genes based on topological overlapping. (B) Heatmap of the correlation between module eigengenes and three clusters. (C) GO/KEGG analysis for three clusters. *p < 0.05, **p < 0.01, ***p < 0.001.Full size imageDiscussionOA is increasingly recognized as a low-grade inflammatory condition, primarily characterized by immune dysregulation and inflammatory responses in the synovium. While previous studies have identified immune-related genes associated with osteoarthritis27, the identification of SID-related genes in osteoarthritic synovial tissue and their corresponding roles remain unexplored. In this study, we characterized 32 SID genes, conducted functional enrichment and immune infiltration analyses, and identified three potential biomarkers that can distinguish osteoarthritis.Bioinformatics analysis is a widely utilized approach to explore key genes and pathways associated with diseases. For instance, Mahima et al. employed various bioinformatics methods to identify key biomolecules involved in bone metastasis in breast28 and prostate29 cancer. Van et al. demonstrated a difference in lipid metabolites between OA synovium and healthy control group30. In our study, we identify three SID-related hub genes that can predict OA occurrence. And 32 SID genes were primarily enriched in biological processes such as “fatty acid metabolic process” and “lipid modification”. Furthermore, the most significantly enriched KEGG pathway was the “Phosphatidylinositol signaling system”. This suggests a potential interaction between lipid metabolism and immune disorders in the peripheral blood of OA patients11. Acyltransferase activity, which regulates disease progression, has been proposed as a potential biomarker for OA31. Of interest, the “acyltransferase activity” was the most significantly enriched molecular function among the SID-related genes. Further studies are needed to explore the modulation of acyltransferase activity in OA31.Studies have reported potential novel biomarkers for diagnosing OA11,32. However, the constructed diagnostic models have been limited to identifying DEGs from a single articular component32, and reliable SID-related markers are rarely reported. In our study, we utilized bioinformatics and RT-qPCR to identify three hub genes of SID that were differentially expressed in both OA synovium and peripheral blood. Our constructed model suggests the possibility of using SID-related genes to screen for early-onset OA. SPHK1 enhances inflammatory response and increases the level of inflammatory mediators such as TGFB2 and IL633,34. Reduced SPHK1 expression is associated with attenuated joint pathology and synovitis progression in invasive arthritis35. Conversely, upregulation of SPHK1 can offer protection against ischemic heart disease and prevent cardiomyocyte death resulting from excessive production of reactive oxygen species36. In this study, the SPHK1 expression was down-regulated and positively correlated with expression level of IL-6 and the abundance of endothelial cells. These results suggest that down-regulated SPHK1 may reduce inflammation and synovial endothelial cells in OA synovium. The specific impact of SPHK1 expression in OA synovial immune needs further exploration. Previous studies have shown that IL1A is a main Pro-inflammatory factor which not only can aggravate the development of OA, but also arouse articular pain37. Meanwhile, the expression of CSF3 can reduce pain sensation induced by inflammatory factors38,39. In the current study, ACACB was found to be down-regulated in OA synovium and was positively correlated with IL1A, while being negatively correlated with CSF3. These results were consistent with previous studies. In the present investigation, we observed a down-regulation of ACAT1 in OA synovium, which was found to be associated with inflammatory factors. However, our current study reveals an up-regulation of ACAT1 expression in the synovium of patients with osteoarthritis, and a negative correlation with the levels of TGFB2 and IL-3. These factors have been implicated in ameliorating the degeneration of osteoarthritis cartilage and subchondral bone40,41. Conversely, ACAT1 expression exhibited a positive association with IL-2, a cytokine known to promote the progression of OA inflammation42. Collectively, these findings suggest that ACAT1 may serve as a potential risk factor in osteoarthritis.In this study, we distinguished the 40 OA patients into three clusters (cluster 1, cluster 2 and cluster 3). The cluster 2 (Older than the other two groups) exhibited downregulated relative scores of T cells, B cells, monocytes, and myeloid dendritic cells, indicating potential immune cell suppression in the elderly OA population. Additionally, we identified upregulation of CSF1 and IFNG, along with downregulation of CD4, IL3, IL1A, and TGFB3 in cluster 2. Previous studies have linked low levels of CSF1 to significant alleviation of OA progression43, while high levels of IL3 and TGFB3 have been associated with improved synovitis development in OA12,41. These results suggest that targeting downregulation of CSF1 or upregulation of IL3 and TGFB3 could be potential therapeutic strategies for elderly patients with OA.WGCNA analysis revealed that characteristic genes in the pink, greenyellow, and green modules were positively correlated with Cluster 1, Cluster 2, and Cluster 3, respectively. Specifically, the characteristic genes of the pink module primarily participate in the “positive regulation of cytosolic calcium ion concentration,” which is crucial for regulating immune and inflammatory responses44,45. Dysregulation of calcium homeostasis has been implicated in synovial inflammation, and calcium ions have shown potential protective and therapeutic effects in osteoarthritis46,47. Therefore, future treatment plans for Cluster 1 (younger OA patients) should focus on targeting the regulation of intracellular calcium ion concentration. AMPK, a key regulator of immune cell metabolism and function, plays a critical role in inhibiting inflammatory responses associated with OA48. Activating the AMPK signal has also been found to limit the development and progression of the degenerative diseases49. In our current study, the characteristic genes of the greenyellow module were enriched in the “AMPK signaling pathway.” Hence, in future treatment plans for Cluster 2 (elderly OA patients), particular attention should be given to targeting the AMPK-related signaling pathway. Lastly, the characteristic genes of the green module were primarily involved in the “glycosphingolipid metabolic process,” which has been shown to inhibit immune responses50,51. Degreed the level of glycosphingolipid metabolism has been proposed as a promising approach for treating OA52. Therefore, effective treatment strategies for OA patients in Cluster 3 should prioritize targeting the “glycosphingolipid metabolic process.”Several limitations exist in our study. Several limitations exist in our study that should be acknowledged. Firstly, we did not explore the influence of other variables such as disease duration and anti-OA medications (e.g., corticosteroids and NSAIDs) on inflammatory gene expression. Secondly, although we have identified three SID-related hub genes, whose expression levels can predict the onset of OA and are linked to synovial immunity, we have not investigated their translational levels or the underlying mechanisms associated with OA development. This represents a critical area for future investigation.ConclusionIn conclusion, our comprehensive analysis of SID-related genes in OA synovium has revealed three hub genes (ACAT1, ACACB and SPHK1), that hold potential ability for predicting OA occurrence. These hub genes exhibit significant involvement in immune dysregulation within the OA synovium. Furthermore, our findings highlight a strong correlation between the AMPK signaling pathway and elderly OA patients, with a notable decrease in the expression levels of ACACB and SPHK1 observed among older OA patients compared to their younger counterparts. This study provides valuable insights into the exploration of novel biomarkers for OA, contributing to the advancement of future research in this field.
Data availability
The datasets used and/or analyzed during the current study are available from the corresponding authors upon reasonable request.
ReferencesBarbour, K. E., Boring, M. & Brady, T. J. Vital signs: Prevalence of doctor-diagnosed arthritis and arthritis-attributable activity limitation—United States, 2013–2015. MMWR Morb. Mortal Wkly Rep. 66 (9), 246–253. https://doi.org/10.15585/mmwr.mm6609e1 (2017).Article
PubMed
PubMed Central
MATH
Google Scholar
Allen, K. D., Thoma, L. M. & Golightly, Y. M. Epidemiology of osteoarthritis. Osteoarthr. Cartil. 30 (2), 184–195. https://doi.org/10.1016/j.joca.2021.04.020 (2022).Article
CAS
MATH
Google Scholar
Parekh, S. M. et al. Risk factors for knee osteoarthritis in retired professional footballers: A cross-sectional study. Clin. J. Sport Med. 31 (3), 281–288. https://doi.org/10.1097/JSM.0000000000000742 (2021).Article
PubMed
Google Scholar
Ratneswaran, A. & Kapoor, M. Osteoarthritis year in review: genetics, genomics, epigenetics. Osteoarthr. Cartil. 29 (2), 151–160. https://doi.org/10.1016/j.joca.2020.11.003 (2021).Article
CAS
Google Scholar
Hunter, D. J. & Bierma-Zeinstra, S. Osteoarthritis. Lancet 393 (10182), 1745–1759. https://doi.org/10.1016/S0140-6736(19)30417-9 (2019).Article
CAS
PubMed
MATH
Google Scholar
Kloppenburg, M. & Berenbaum, F. Osteoarthritis year in review 2019: epidemiology and therapy. Osteoarthr. Cartil. 28 (3), 242–248. https://doi.org/10.1016/j.joca.2020.01.002 (2020).Article
CAS
Google Scholar
Safiri, S. et al. Global, regional and national burden of osteoarthritis 1990–2017: a systematic analysis of the Global Burden of Disease Study 2017. Ann. Rheum. Dis. 79 (6), 819–828. https://doi.org/10.1136/annrheumdis-2019-216515 (2020).Article
PubMed
Google Scholar
Quicke, J. G., Conaghan, P. G., Corp, N. & Peat, G. Osteoarthritis year in review 2021: epidemiology & therapy. Osteoarthr. Cartil. 30 (2), 196–206. https://doi.org/10.1016/j.joca.2021.10.003 (2022).Article
CAS
Google Scholar
Roos, E. M. & Arden, N. K. Strategies for the prevention of knee osteoarthritis. Nat. Rev. Rheumatol. 12 (2), 92–101. https://doi.org/10.1038/nrrheum.2015.135 (2016).Article
CAS
PubMed
MATH
Google Scholar
Sellam, J. & Berenbaum, F. The role of synovitis in pathophysiology and clinical symptoms of osteoarthritis. Nat. Rev. Rheumatol. 6 (11), 625–635. https://doi.org/10.1038/nrrheum.2010.159 (2010).Article
CAS
PubMed
Google Scholar
Tu, B. et al. Comprehensive analysis of arachidonic acid metabolism-related genes in diagnosis and synovial immune in osteoarthritis: based on bulk and single-cell RNA sequencing data. Inflamm. Res. https://doi.org/10.1007/s00011-023-01720-4 (2023).Article
PubMed
Google Scholar
Du, X., Cai, L., Xie, J. & Zhou, X. The role of TGF-beta3 in cartilage development and osteoarthritis. Bone Res. 11 (1), 2. https://doi.org/10.1038/s41413-022-00239-4 (2023).Article
CAS
PubMed
PubMed Central
Google Scholar
Woodell-May, J. E. & Sommerfeld, S. D. Role of Inflammation and the immune system in the progression of osteoarthritis. J. Orthop. Res. 38 (2), 253–257. https://doi.org/10.1002/jor.24457 (2020).Article
PubMed
Google Scholar
Guermazi, A. et al. Assessment of synovitis with contrast-enhanced MRI using a whole-joint semiquantitative scoring system in people with, or at high risk of, knee osteoarthritis: the MOST study. Ann. Rheum. Dis. 70 (5), 805–811. https://doi.org/10.1136/ard.2010.139618 (2011).Article
PubMed
Google Scholar
Roemer, F. W. et al. Presence of MRI-detected joint effusion and synovitis increases the risk of cartilage loss in knees without osteoarthritis at 30-month follow-up: the MOST study. Ann. Rheum. Dis. 70 (10), 1804–1809. https://doi.org/10.1136/ard.2011.150243 (2011).Article
PubMed
Google Scholar
Scanzello, C. R. et al. Synovial inflammation in patients undergoing arthroscopic meniscectomy: molecular characterization and relationship to symptoms. Arthr. Rheum. 63 (2), 391–400. https://doi.org/10.1002/art.30137 (2011).Article
Google Scholar
Zhang, Q. et al. Mechanism of immune infiltration in synovial tissue of osteoarthritis: a gene expression-based study. J. Orthop. Surg. Res. 18 (1), 58. https://doi.org/10.1186/s13018-023-03541-x (2023).Article
CAS
PubMed
PubMed Central
Google Scholar
Deligne, C. et al. Differential expression of interleukin-17 and interleukin-22 in inflamed and non-inflamed synovium from osteoarthritis patients. Osteoarthr. Cartil. 23 (11), 1843–1852. https://doi.org/10.1016/j.joca.2014.12.007 (2015).Article
CAS
Google Scholar
Miller, R. J., Malfait, A. M. & Miller, R. E. The innate immune response as a mediator of osteoarthritis pain. Osteoarthr. Cartil. 28 (5), 562–571. https://doi.org/10.1016/j.joca.2019.11.006 (2020).Article
CAS
MATH
Google Scholar
Sanchez-Lopez, E., Coras, R., Torres, A., Lane, N. E. & Guma, M. Synovial inflammation in osteoarthritis progression. Nat. Rev. Rheumatol. 18 (5), 258–275. https://doi.org/10.1038/s41584-022-00749-9 (2022).Article
PubMed
PubMed Central
Google Scholar
Del Sordo, L. et al. Impaired efferocytosis by synovial macrophages in patients with knee osteoarthritis. Arthr. Rheumatol. 75 (5), 685–696. https://doi.org/10.1002/art.42412 (2023).Article
CAS
Google Scholar
Murray-Brown, W. et al. Differential expansion of T peripheral helper cells in early rheumatoid arthritis and osteoarthritis synovium. RMD Open 8 (2), e002563. https://doi.org/10.1136/rmdopen-2022-002563 (2022).Article
PubMed
PubMed Central
Google Scholar
Xie, X., Doody, G. M., Shuweihdi, F., Conaghan, P. G. & Ponchel, F. B-cell capacity for expansion and differentiation into plasma cells are altered in osteoarthritis. Osteoarthr. Cartil. 31 (9), 1176–1188. https://doi.org/10.1016/j.joca.2023.03.017 (2023).Article
Google Scholar
Newman, A. M. et al. Robust enumeration of cell subsets from tissue expression profiles. Nat. Methods 12(5), 453–457. https://doi.org/10.1038/nmeth.3337 (2015).Article
CAS
PubMed
PubMed Central
MATH
Google Scholar
Presson, A. P. et al. Integrated weighted gene co-expression network analysis with an application to chronic fatigue syndrome. BMC Syst. Biol. 2, 95. https://doi.org/10.1186/1752-0509-2-95 (2008).Article
CAS
PubMed
PubMed Central
MATH
Google Scholar
Udyavar, A. R. et al. Co-expression network analysis identifies spleen tyrosine kinase (SYK) as a candidate oncogenic driver in a subset of small-cell lung cancer. BMC Syst. Biol. https://doi.org/10.1186/1752-0509-7-S5-S1 (2013).Article
PubMed
PubMed Central
MATH
Google Scholar
Hu, X., Ni, S., Zhao, K., Qian, J. & Duan, Y. Bioinformatics-led discovery of osteoarthritis biomarkers and inflammatory infiltrates. Front. Immunol. 13, 871008. https://doi.org/10.3389/fimmu.2022.871008 (2022).Article
CAS
PubMed
PubMed Central
Google Scholar
Bhardwaj, M. et al. Identification of biomarkers associated with Paget’s disease of bone and bone metastasis from breast cancer patients. Cancer Rep. (Hoboken) 7 (9), e70003. https://doi.org/10.1002/cnr2.70003 (2024).Article
CAS
PubMed
MATH
Google Scholar
Bhardwaj, M., Sri, T., Krupanidhi, S. & Singh, S. Identification of potential biomarkers for bone metastasis using human cancer metastasis database. Int. J. Health Sci. (Qassim) 18 (3), 6–14 (2024).PubMed
Google Scholar
Van de Vyver, A. et al. Synovial fluid fatty acid profiles differ between osteoarthritis and healthy patients. Cartilage 11 (4), 473–478. https://doi.org/10.1177/1947603518798891 (2020).Article
CAS
PubMed
Google Scholar
Tarantino, U., Ferlosio, A., Arcuri, G., Spagnoli, L. G. & Orlandi, A. Transglutaminase 2 as a biomarker of osteoarthritis: an update. Amino Acids 44 (1), 199–207. https://doi.org/10.1007/s00726-011-1181-y (2013).Article
CAS
PubMed
Google Scholar
Liu, Y. et al. Six macrophage-associated genes in synovium constitute a novel diagnostic signature for osteoarthritis. Front. Immunol. 13, 936606. https://doi.org/10.3389/fimmu.2022.936606 (2022).Article
CAS
PubMed
PubMed Central
Google Scholar
Tchetina, E. V., Antoniou, J., Tanzer, M., Zukor, D. J. & Poole, A. R. Transforming growth factor-beta2 suppresses collagen cleavage in cultured human osteoarthritic cartilage, reduces expression of genes associated with chondrocyte hypertrophy and degradation, and increases prostaglandin E(2) production. Am. J. Pathol. 168 (1), 131–140. https://doi.org/10.2353/ajpath.2006.050369 (2006).Article
CAS
PubMed
PubMed Central
Google Scholar
Wang, T. & He, C. Pro-inflammatory cytokines: The link between obesity and osteoarthritis. Cytokine Growth Fact. Rev. 44, 38–50. https://doi.org/10.1016/j.cytogfr.2018.10.002 (2018).Article
CAS
MATH
Google Scholar
Bu, Y., Wu, H., Deng, R. & Wang, Y. Therapeutic potential of SphK1 inhibitors based on abnormal expression of SphK1 in inflammatory immune related-diseases. Front. Pharmacol. 12, 733387. https://doi.org/10.3389/fphar.2021.733387 (2021).Article
CAS
PubMed
PubMed Central
Google Scholar
Shao, J. J., Peng, Y., Wang, L. M., Wang, J. K. & Chen, X. Activation of SphK1 by K6PC-5 inhibits oxygen-glucose deprivation/reoxygenation-induced myocardial cell death. DNA Cell Biol. 34 (11), 669–676. https://doi.org/10.1089/dna.2015.2959 (2015).Article
CAS
PubMed
PubMed Central
Google Scholar
Cai, H., Sun, H. J., Wang, Y. H. & Zhang, Z. Relationships of common polymorphisms in IL-6, IL-1A, and IL-1B genes with susceptibility to osteoarthritis: a meta-analysis. Clin. Rheumatol. 34 (8), 1443–1453. https://doi.org/10.1007/s10067-014-2708-x (2015).Article
PubMed
MATH
Google Scholar
Fan, Z. et al. Sympathetic nerve-mediated fellow eye pain during sequential cataract surgery by regulating granulocyte colony stimulating factor CSF3. Front. Cell Neurosci. 24 (16), 841733. https://doi.org/10.3389/fncel.2022.841733.eCollection2022 (2022).Article
Google Scholar
Son, E. S. et al. miRNA-6515-5p regulates particulate matter-induced inflammatory responses by targeting CSF3 in human bronchial epithelial cells. Toxicol. Vitro 84, 105428. https://doi.org/10.1016/j.tiv.2022.105428 (2022).Article
CAS
Google Scholar
Hou, C. et al. MiR-193b regulates early chondrogenesis by inhibiting the TGF-beta2 signaling pathway. FEBS Lett. 589 (9), 1040–1047. https://doi.org/10.1016/j.febslet.2015.02.017 (2015).Article
CAS
PubMed
Google Scholar
Kour, S. et al. IL-3 decreases cartilage degeneration by downregulating matrix metalloproteinases and reduces joint destruction in osteoarthritic mice. J. Immunol. 196 (12), 5024–5035. https://doi.org/10.4049/jimmunol.1500907 (2016).Article
CAS
PubMed
Google Scholar
Thornton, S., Boivin, G. P., Kim, K. N., Finkelman, F. D. & Hirsch, R. Heterogeneous effects of IL-2 on collagen-induced arthritis. J. Immunol. 165 (3), 1557–1563. https://doi.org/10.4049/jimmunol.165.3.1557 (2000).Article
CAS
PubMed
Google Scholar
Ota, T. et al. Expression of colony-stimulating factor 1 is associated with occurrence of osteochondral change in pigmented villonodular synovitis. Tumour Biol. 36 (7), 5361–5367. https://doi.org/10.1007/s13277-015-3197-5 (2015).Article
CAS
PubMed
MATH
Google Scholar
Ishikawa, H. & Barber, G. N. STING is an endoplasmic reticulum adaptor that facilitates innate immune signalling. Nature 455 (7213), 674–678. https://doi.org/10.1038/nature07317 (2008).Article
ADS
CAS
PubMed
PubMed Central
Google Scholar
Sun, L. J., Wu, J. X., Du, F. H., Chen, X. & Chen, Z. J. J. Cyclic GMP-AMP synthase is a cytosolic DNA sensor that activates the type I interferon pathway. Science 339 (6121), 786–791. https://doi.org/10.1126/science.1232458 (2013).Article
ADS
CAS
PubMed
MATH
Google Scholar
Zhang, Y. et al. ASIC1a induces synovial inflammation via the Ca(2+)/NFATc3/ RANTES pathway. Theranostics 10 (1), 247–264. https://doi.org/10.7150/thno.37200 (2020).Article
CAS
PubMed
PubMed Central
MATH
Google Scholar
Gao, W., Hasan, H., Anderson, D. E. & Lee, W. The role of mechanically-activated ion channels piezo1, Piezo2, and TRPV4 in chondrocyte mechanotransduction and mechano-therapeutics for osteoarthritis. Front. Cell Dev. Biol. 10, 885224. https://doi.org/10.3389/fcell.2022.885224 (2022).Article
PubMed
PubMed Central
Google Scholar
Salminen, A. & Kaarniranta, K. AMP-activated protein kinase (AMPK) controls the aging process via an integrated signaling network. Ageing Res. Rev. 11 (2), 230–241. https://doi.org/10.1016/j.arr.2011.12.005 (2012).Article
CAS
PubMed
MATH
Google Scholar
Li, J. et al. Metformin limits osteoarthritis development and progression through activation of AMPK signalling. Ann. Rheum. Dis. 79 (5), 635–645. https://doi.org/10.1136/annrheumdis-2019-216713 (2020).Article
CAS
PubMed
MATH
Google Scholar
Jongsma, M. L. M. et al. The SPPL3-defined glycosphingolipid repertoire orchestrates HLA class i-mediated immune responses. Immunity 54 (1), 132-150.e9. https://doi.org/10.1016/j.immuni.2020.11.003 (2021).Article
CAS
PubMed
Google Scholar
Zhang, T., de Waard, A. A., Wuhrer, M. & Spaapen, R. M. The role of glycosphingolipids in immune cell functions. Front. Immunol. 10, 90. https://doi.org/10.3389/fimmu.2019.00090 (2019).Article
CAS
PubMed
PubMed Central
Google Scholar
Seito, N. et al. Interruption of glycosphingolipid synthesis enhances osteoarthritis development in mice. Arthr. Rheum. 64 (8), 2579–2588. https://doi.org/10.1002/art.34463 (2012).Article
CAS
Google Scholar
Download referencesFundingThis study was supported by Grants from Anhui Key Clinical Speciality Construction Project.Author informationAuthor notesZheng Zhu, Bizhi Tu and Cheng Peng contributed equally to this work.Authors and AffiliationsDepartment of Orthopedics, Hefei First People’s Hospital, Anhui Medical University, 390 Huaihe Road, Hefei, 230061, Anhui, ChinaZheng Zhu, Bizhi Tu, Cheng Peng, Xun Xu, Peizhi Lu & Rende NingAuthorsZheng ZhuView author publicationsYou can also search for this author inPubMed Google ScholarBizhi TuView author publicationsYou can also search for this author inPubMed Google ScholarCheng PengView author publicationsYou can also search for this author inPubMed Google ScholarXun XuView author publicationsYou can also search for this author inPubMed Google ScholarPeizhi LuView author publicationsYou can also search for this author inPubMed Google ScholarRende NingView author publicationsYou can also search for this author inPubMed Google ScholarContributionsRende Ning conceived the study idea, revised the manuscript, and provided financial support. ZZ, BT and CP collected the data and wrote the initial draft. PZ and XX contributed to the data collection and analysis. All authors approved the final draft of the manuscript. All authors are accountable for all aspects of the work in ensuring related questions’ accuracy or integrity. Any parts of the work are appropriately investigated and resolved. NR is the guarantor. The corresponding author attests that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted.Corresponding authorCorrespondence to
Rende Ning.Ethics declarations
Competing interests
The authors declare no competing interests.
Ethics approval and consent to participate
Ethical approval for our study was granted by The Committee on Medical Ethics of The Third Affiliated Hospital of Anhui Medical University (Reference number 2023-022-01). Since all the data used in the current study was available online, and no individual patient was involved, it could be confirmed we have obtained all the written informed consent. All methods were implemented in accordance with relevant guidelines and regulations.
Additional informationPublisher’s noteSpringer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.Supplementary InformationSupplementary Information.Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Reprints and permissionsAbout this articleCite this articleZhu, Z., Tu, B., Peng, C. et al. Integrated bioinformatics and clinical data identify three novel biomarkers for osteoarthritis diagnosis and synovial immune.
Sci Rep 15, 10987 (2025). https://doi.org/10.1038/s41598-025-95837-7Download citationReceived: 11 June 2024Accepted: 24 March 2025Published: 31 March 2025DOI: https://doi.org/10.1038/s41598-025-95837-7Share this articleAnyone you share the following link with will be able to read this content:Get shareable linkSorry, a shareable link is not currently available for this article.Copy to clipboard
Provided by the Springer Nature SharedIt content-sharing initiative
KeywordsOsteoarthritisBiomarkerDiagnostic modelImmune infiltrationBioinformatics