Skip to main content

Quantitative proteomic screening uncovers candidate diagnostic and monitoring serum biomarkers of ankylosing spondylitis



We sought to discover serum biomarkers of ankylosing spondylitis (AS) for diagnosis and monitoring disease activity.


We studied biologic-treatment-naïve AS and healthy control (HC) patients’ sera. Eighty samples matched by age, gender, and race (1:1:1 ratio) for AS patients with active disease, inactive disease, and HC were analyzed with SOMAscan™, an aptamer-based discovery platform. T-tests tests were performed for high/low-disease activity AS patients versus HCs (diagnosis) and high versus low disease activity (Monitoring) in a 2:1 and 1:1 ratio, respectively, to identify differentially expressed proteins (DEPs).

We used the Cytoscape Molecular Complex Detection (MCODE) plugin to find clusters in protein–protein interaction networks and Ingenuity Pathway Analysis (IPA) for upstream regulators. Lasso regression analysis was performed for diagnosis.


Of the 1317 proteins detected in our diagnosis and monitoring analyses, 367 and 167 (317 and 59, FDR-corrected q < .05) DEPs, respectively, were detected. MCODE identified complement, IL-10 signaling, and immune/interleukin signaling as the top 3 diagnosis PPI clusters. Complement, extracellular matrix organization/proteoglycans, and MAPK/RAS signaling were the top 3 monitoring PPI clusters. IPA showed interleukin 23/17 (interleukin 22, interleukin 23A), TNF (TNF receptor-associated factor 3), cGAS-STING (cyclic GMP-AMP synthase, Stimulator of Interferon Gene 1), and Jak/Stat (Signal transducer and activator of transcription 1), signaling in predicted upstream regulators. Lasso regression identified a Diagnostic 13-protein model predictive of AS. This model had a sensitivity of 0.75, specificity of 0.90, a kappa of 0.59, and overall accuracy of 0.80 (95% CI: 0.61–0.92). The AS vs HC ROC curve was 0.79 (95% CI: 0.61–0.96).


We identified multiple candidate AS diagnostic and disease activity monitoring serum biomarkers using a comprehensive proteomic screen. Enrichment analysis identified key pathways in AS diagnosis and monitoring. Lasso regression identified a multi-protein panel with modest predictive ability.


Ankylosing spondylitis (AS) is a chronic, systemic inflammatory disease with cardinal features of inflammatory back pain, sacroiliitis, and spinal fusion that lead to significant functional impairment [1]. AS is also associated with extra-articular features such as uveitis, psoriasis, and inflammatory bowel disease, which can precede or follow characteristic spinal involvement [2]. AS is a subset of axial spondyloarthritis (AxSpA), estimated to affect up to 2–3 million people in the USA [3]. Genetic studies in AS have identified over 113 susceptibility loci that have suggested pathologic mechanisms for this complex disease [4,5,6]. Gene expression studies of peripheral blood at both the bulk and single-cell level have recently provided insight into disease pathogenesis [7,8,9].

Large-scale examination of proteins, unlike DNA and RNA, remains scarce and limited in AS. Recent mass spectrometry-based studies of AS patients compared to healthy controls have revealed dysregulation of serum proteins—including complement, metalloproteinases, and serum amyloid A1 (SAA1) [10,11,12]. Dynamic range limitations of this lab technique however limit proteome coverage, which may restrict molecular characterization of AS [13].

Biomarkers have been defined as “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention” [14]. Many rheumatic diseases now have incorporated serologic and proteomic markers into their classification criteria and disease activity measures. Biomarkers that can aid in the diagnosis and monitoring of disease activity of AS could improve clinical care and advance our pathophysiology understanding. The two most utilized in AS clinical practice and trials include C-reactive protein (CRP) and erythrocyte sedimentation rate (ESR). These lack sensitivity with two thirds of those with established AS and clinically active disease having within normal levels [15]. To this end, a great need remains for the identification of better diagnostic and monitoring AS biomarkers.

In this study, we investigated an extended aptamer-based panel of over 1300 proteins in biologic pharmacotherapy naïve AS patients to (1) discriminate AS patients versus healthy controls (HC) and (2) compare AS patients with active versus inactive disease.

Methods and datasets

Patient and specimens

Serum samples from the Prospective Study of Outcome in Ankylosing Spondylitis (PSOAS) cohort UTHealth biorepository were examined [16]. This multicenter cohort was initiated in 2003 and includes patients from UTHealth Houston, University of California San Francisco, the NIH Clinical Center, Cedars-Sinai Medical Center, and the Queensland University of Technology (Australia). The research carried out followed the Helsinki Declaration, each institution had the study approved by their respective institutional review boards (IRB), and each participating patient reviewed and signed an informed consent form. We studied age-, gender-, and race-matched patient and control sera in a 1:1:1 ratio for active, inactive, and healthy controls. All AS patients met modified New York classification criteria. Active disease was defined by an Ankylosing Spondylitis Disease Activity Score C-reactive Protein (ASDAS-CRP) of ≥ 2.1 and inactive disease as ASDAS-CRP of < 1.3 [17]. All patients were biologic and synthetic disease-modifying-anti-rheumatic drug (DMARD) naïve at the time of sera draw. Serum samples were immediately stored at a temperature lower than − 70 °C and had not been previously thawed.

Serum protein determination

Serum samples were analyzed using the SOMAscan™ assay (SomaLogic; Boulder, CO), which is a sensitive and quantitative protein biomarker discovery platform. SOMAmers (Slow Off-rate Modified Aptamers), single-stranded DNA aptamers with modified nucleotides, bind to specific proteins in the serum that are then quantified as DNA [18, 19]. The SOMAscan assay quantified a total of 1320 proteins in each patient sample. Sample data was first normalized to remove hybridization variation within a run followed by median normalization across all samples to remove plate effects between runs. All plates were matched between active disease, inactive disease, and healthy controls in a 1:1:1 ratio, respectively. The median lower limit of quantitation for all measured proteins was 0.3 picomolar (pM), with a dynamic range of > 5 logs, and a median coefficient of variation (%CV) of 5%.

Statistical analysis

T-tests and Mann–Whitney U tests were performed for active and inactive AS patients compared to healthy controls (diagnosis analysis) and active compared to inactive disease activity (Monitoring Analysis) in a 2:1 and 1:1 ratio, respectively. In our analyses, p-values and q-values (false discovery rate [FDR] corrected p-value) were calculated for all proteins using the Benjamini and Hochberg method in the R environment for statistical computing ( All analyzed proteins with a p < 0.05 were considered differentially expressed proteins (DEPs).

We used the Search Tool for the Retrieval of Interacting Gene/Proteins (STRING, database to analyze DEPs in our diagnosis and monitoring analyses for protein–protein interactions (PPI) with Cytoscape software. The Cytoscape Molecular Complex Detection (MCODE) plugin was used to find clusters in PPI networks, confidence cutoff of 0.4.

DEPs were also analyzed using Ingenuity Pathway Analysis (IPA) software (Qiagen to identify active molecular targets. The goal of Upstream Regulator Analysis is to identify upstream molecular regulators and to predict whether they are active or inhibited. This analysis is based on expected causal effects between upstream molecular targets. A Z-score algorithm is used to make predictions with Z-scores > 2 and <  − 2 considered significant.

A multi-biomarker panel among our DEPs for diagnosis were selected using a L1-penalized logistic regression using the least absolute shrinkage and selection operator (lasso) classifier, over a range of lambda, a tuning parameter that determines how many biomarkers are selected. Lasso regression analysis was performed of our diagnosis DEPs to determine optimal protein combination in a 2:1 training/test split. The models were evaluated using cross-validation and inspecting plots receiver operating characteristic (ROC) curves. R packages, glmnet and caret, were used for these statistical analyses.


Demographic and clinical characteristics

We studied 80 study patients’ sera, n = 26:26:28 for active, inactive, and HC, respectively. Study patients’ sex, age, and other important clinical characteristics are summarized in Table 1. Our AS patients were a mean 46 ± 14 years of age with a symptom duration of average of 25 ± 13 years. All patients and controls were White.

Table 1 Demographic and clinical characteristics of the PSOAS participants and control subjects

Serum protein levels

Of the 1317 proteins detected in our diagnosis and monitoring analyses, 367 and 167 DEPs were detected, respectively (p < 0.05) (Fig. 1). When corrected for multiple comparison (FDR-corrected q < 0.05), 247 and 27 DEPs remained for diagnosis and monitoring, respectively. Thirteen DEPs overlapped between our diagnosis and monitoring analyses (Table 2). Eleven of the 13 overlapping DEPs had associations in concordant direction for diagnosis and monitoring. The top 10 upregulated and downregulated DEPs based on fold change for diagnostic and monitoring biomarkers are presented in Table 3. The complete list of differentially expressed proteins is available in Supplementary Table 1.

Fig. 1
figure 1

Volcano plot of all serum proteins for A diagnosis (AS patients vs. healthy controls) and B monitoring (AS high vs. low disease activity patients). Fold change (FC) of AS patients compared to controls < 0.5, 0.5–2.0, and > 2 is presented in yellow, black, and red, respectively. 367 and 157 proteins were differentially expressed for diagnosis and monitoring, respectively

Table 2 Shared AS diagnostic and monitoring serum proteins
Table 3 Top upregulated and downregulated serum proteins detected by a 1320-plex proteomic screen

Pathway analyses

MCODE analyses identified complement regulation/signal transduction, interleukin (IL)-10 signaling/immune system, and immune system/interleukin signaling as the top 3 overrepresented pathways. Complement, extracellular matrix organization/proteoglycans, and mitogen-activated protein kinase(MAPK)/rat sarcoma virus protein (RAS) signaling were the top 3 overrepresented pathways among the monitoring biomarkers (Fig. 2).

Fig. 2
figure 2

Protein–protein interaction networks. A Top 367 diagnosis DEPs and B top 157 monitoring DEPs (T-test p < 0.05) through the CytoscapeSTRING App with a confidence cutoff of 0.4. Molecular Complex Detection (MCODE) clustering was performed and displayed are the top three clusters. The color of each MCODE node corresponds to the fold change. Nodes with a fold change less than one range in color from blue to purple while those with a fold change greater than one range from pink to red. The confidence score of each interaction is displayed as the edge thickness and opacity. The top two reactome pathways associated with each cluster are displayed below each cluster

The IPA upstream analysis predicted, activated upstream regulators included IL 23/17 (IL-22, IL-23A), tumor necrosis factor (TNF receptor-associated factor 3), cGAS-STING (cyclic GMP-AMP synthase, Stimulator of Interferon Gene 1), and Jak/Stat (Janus Kinase/Signal transducer and activator of transcription 1), signaling in AS. Predicted inhibited upstream regulators included those involved in lipid metabolism (Nuclear Receptor Subfamily 5 Group A Member 2, Peroxisome proliferator-activated receptor alpha) and protein folding (Clusterin, Presenilin-2) (Fig. 3.)

Fig. 3
figure 3

Top predicted upstream molecular regulators based on the Ingenuity Knowledge Base. X-axis shows the activation Z score calculated based on the Ingenuity Pathway Analysis for identifying upstream regulators. Proteins that were differentially expressed with ankylosing spondylitis compared to healthy control subjects used were input into Ingenuity

Lasso regression

All samples were randomly assigned in a 2:1 ratio between discovery (N = 54) and validation sets (N = 26). Lasso regression identified a Diagnostic 13-DEP model predictive of AS: Immunoglobulin A (IgA), Complement component 5a (C5a), Secreted frizzled-related protein 1 (SARP-2), Secretory leukocyte peptidase inhibitor (SLPI), Cathepsin A (CTSA), Neurexophilin 1 (NXPH1), C-X-C motif chemokine ligand 16 (CXCL16), Interleukin 6 signal transducer (gp130), complement component 4b (C4b), Cofilin-1 (CFL1), Cell adhesion molecule L1 like (CHL1), Signaling lymphocytic activation molecule family member 6 (SLAF6), and Macrophage mannose receptor (MRC1). This model had a sensitivity of 0.75, specificity of 0.90, kappa of 0.59, and an overall accuracy of 0.80 (95% CI: 0.61–0.92). The predictive probability of our model to discriminate AS patients vs controls based on ROC curve (95% CI) was 0.79 (0.61–0.96) (Fig. 4). The McNemar’s test of our model was non-significant (p > 0.05), suggesting a similar proportion of misclassification for diagnosis (e.g., false negatives and false positives). We did not perform Lasso regression for monitoring as this would have not been meaningful due to the modest sample size.

Fig. 4
figure 4

Receiver operator curve of diagnostic 13-protein LASSO discrimination of AS patients vs. controls


The goal of this study was to identify candidate biomarkers for AS diagnosis and monitoring disease activity. Our proteomic screen of ankylosing spondylitis identified 317 and 59, FDR-corrected, DEPs for diagnosing and monitoring, respectively. Further bioinformatics tools highlighted signaling pathways and potential in silica upstream regulators. In our PPI analyses, we elucidated complement and immune signaling importance in diagnosis and MAPK/RAS for monitoring disease activity. IPA upstream analysis predicted activated TNF, IL-23/17, and cGAS-STING signaling pathway targets as well as inhibited lipid metabolism and protein folding targets. Finally, our machine-learning model showed that a 13-DEP model had ability to discriminate AS patients from controls with modest discrimination, AUC 0.79.

These results add to our knowledge of ankylosing spondylitis, showing that serum proteins can molecularly distinguish AS from healthy controls. We also elucidated additional proteins that differentiate between disease activity states that may be useful in monitoring disease in clinic. Bioinformatics tools highlighted important inflammatory and immunologic pathways important for AS pathogenesis. This including well-known AS association of TNF, IL 23/17, and JAK/Stat signaling that contrasted with cGAS-STING, MAPK/RAS signaling that were identified as well.

Previous biomarker studies in AS have largely focused on individual proteins based on a priori knowledge of biomarkers in associated conditions [20]. Comprehensive discovery studies of AS biomarker susceptibility have only used mass spectrometry to date. In contrast, our study was based on broad screens of aptamer-based libraries of > 1300 proteins in an unbiased fashion. Our study is the first to use aptamer-based high-throughput technology to capture low-abundance serum proteins in AS patients and controls, identifying DEP new potential biomarkers for diagnosis and disease activity in addition to confirming previous findings. For example, among our top diagnosis DEPs, SAA1 has now been shown in multiple studies to be elevated in AS while Trefoil Factor (TFF)1 has not been reported in AS/AxSpA [10,11,12].

While the pathophysiologic function of TFF1 is unknown, TFF peptides modulate cell junctional complexes thus contributing to the gastrointestinal epithelial barrier function [21]. Impairment of the mucosal epithelial barrier is a hallmark of inflammatory bowel disease (IBD) [22]. TFF levels increase upon epithelial injury, presumably to prevent further damage and disease progression in IBD [23,24,25]. Beyond IBD, TFF1 is also upregulated in the intestine in response to injury [26, 27]. Given TFF1 long known association with inflammatory bowel disease, a condition seen in overlap with AS, and the large prevalence of colitis in AS, our newly identified DEPs may be useful AS biomarkers [24].

Among our 11 concordant biomarkers for diagnosis and monitoring, they can be grouped into three categories: acute phase reactants (e.g., SAA1, Haptoglobin, CRP), complement (e.g., C5a, C5b-C6 complex, C6, C9, Factor B, Factor I), and lipopolysaccharide-related proteins (LPS) (e.g., FUT5, LBP). SAA1 had greater fold difference than CRP, suggesting it may be a more useful biomarker than CRP for diagnosis and disease activity. Complement system proteins, most notably C4, have been reported to be upregulated in ankylosing spondylitis [28, 29]. It has been suggested that complement activation may be a key pathway involved in AS pathogenesis through murine studies [30]. We identified additional complement proteins in our study that may elucidate complement involvement in AS. Gram-negative bacteria are long thought to be an environmental trigger for ankylosing spondylitis, which relates to our LPS protein findings [31]. Furthermore, prior peripheral blood gene expression studies in AS have identified an upregulation of Toll-like receptor 4 (TLR4) AS which is the ligand of LBP and LPS [7, 32]. Our results support the involvement of LPS-LBP/TLR4 axis AS pathogenesis and disease activity.

We also piloted lasso regression, a machine-learning technique, to find a protein panel that can best identify AS patients. This led to a 13-protein panel with an overall AUC superior to that of C-reactive protein, the commonly used AS serum biomarker [33]. Our protein panel performance was comparable to previous reports of advanced imaging including MRI and low-dose CT [34]. These proteins included complement and interleukin-6 signaling proteins, already highlighted as important pathways as well as other inflammation-associated proteins. For example, SLAF6, also known as SLAMF6, has been shown to lead to increased IL-17 production in autoimmune conditions [35,36,37,38,39,40,41,42,43]. These candidate proteins require further investigation to determine their clinical utility singularly and in combination.

This project had limitations. While all patients studied were not exposed to biologic or synthetic DMARDs, non-steroidal anti-inflammatory drugs (NSAIDs) were not accounted for in our analyses. This treatment modality might have some impact on the serum protein profile; however, we expect that NSAIDs would lead to a decrease in the serum inflammatory markers. We instead observed an increase in the inflammatory serum proteins in the AS versus control comparison, suggesting that the observed molecular profile is disease-related rather than secondary to NSAID treatment. Most of our AS patients had long-standing disease; serum biomarkers may differ between early and established AS. The large number of potential biomarkers and modest sample size of patients may make our findings susceptible for overfitting. We chose Lasso modeling over other machine-learning techniques due to the penalization limiting this potential error. We also chose to study ankylosing spondylitis patients compared to healthy controls, two distinct conditions. Biomarkers that could distinguish non-radiographic axial spondyloarthritis compared to non-specific chronic lower back pain would have greater clinical utility. Our results thus can only be taken in an indirect context for the aforementioned clinical scenario. Biomarkers identified thus in our study require further testing in patients with the full AxSpA spectrum compared to various chronic lower back pain conditions that mirror clinical symptoms of AxSpA.

The clinical utility of our identified biomarkers requires replication. Our findings indicate that the required sample size will differ for each DEP. For example, serum amyloid A requires only 35 patients per group (diagnosis) while Complement Factor I would require 116 patient samples (diagnosis) for replication as diagnostic biomarker based on their delta/standard deviation from our proteomic screen at a power of 80% and significance of 5%. Our study however adds to the literature of potential candidate biomarkers that can aid in AS diagnosis and monitoring of disease activity.


In summary, the current methods for the diagnosis of AS patients have relied on a combination of patient-reported symptoms, imaging, and non-specific acute phase reactants [17]. These have been useful for the classification to study disease; however, among the nearly 20% of the US population with chronic back pain, there remains a large portion of potentially undiagnosed AxSpA. Better biomarkers would address a significant unmet need both in clinic, where earlier diagnosis and referral might be improved, as well as in bettering our understanding of disease mechanisms [44]. Our study adds to the current literature by highlighting inflammatory pathways involved in AS diagnosis and disease activity as well as using machine learning to identify potential diagnostic and monitoring biomarkers in AS. This project may serve as a useful adjunct to earlier diagnosis, more accurate disease activity monitoring, and development of a predictive prognosis model that can ultimately lead to earlier AS diagnosis and treatment.

Availability of data and materials

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.



Ankylosing spondylitis


Axial spondyloarthritis


Complement component 4b


Complement component 5a




Cyclic GMP-AMP synthase, Stimulator of Interferon Gene 1


Cell adhesion molecular L1 like


C reactive protein

CXCL 16:

C-X-C motif chemokine ligand 16


Cathepsin A


Differentially expressed proteins


Disease-modifying anti-rheumatic drug


Erythrocyte sedimentation rate


False discovery rate


Interleukin 6 signal transducer


Inflammatory bowel disease


Immunoglobulin A




Ingenuity Pathway Analysis


Janus Kinase-Signal transducer and activator of transcription 1


Least absolute shrinkage and selection operator




Mitogen-activated protein kinase


Cytoscape’s Molecular Complex Detection plugin


Macrophage mannose receptor


Neurexophilin 1


Non-steroidal anti-inflammatory drugs


Protein-protein interactions


Rat Sarcoma Virus Proteins


Receiver operating characteristic


Serum amyloid A1


Secreted frizzled-related protein 1


Secretory leukocyte peptidase inhibitor


Slow Off-rate Modified Aptamers


Search Tool for the Retrieval of Interacting Genes/Proteins


Toll-like receptor 4


Trefoil Factor


  1. Taurog JD, Chhabra A, Colbert RA. Ankylosing spondylitis and axial spondyloarthritis. N Engl J Med. 2016;374(26):2563–74.

    Article  PubMed  Google Scholar 

  2. Firestein GS, Gabriel SE, McInnes IB, O’Dell JR. Kelley and Firestein’s textbook of rheumatology. 2017.

    Google Scholar 

  3. Reveille JD, Witter JP, Weisman MH. Prevalence of axial spondylarthritis in the United States: estimates from a cross-sectional survey. Arthritis Care Res. 2012;64(6):905–10.

    Article  Google Scholar 

  4. Ellinghaus D, Jostins L, Spain SL, Cortes A, Bethune J, Han B, Park YR, Raychaudhuri S, Pouget JG, Hubenthal M, et al. Analysis of five chronic inflammatory diseases identifies 27 new associations and highlights disease-specific patterns at shared loci. Nat Genet. 2016;48(5):510–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  5. Burton PR, Clayton DG, Cardon LR, Craddock N, Deloukas P, Duncanson A, Kwiatkowski DP, McCarthy MI, Ouwehand WH, Samani NJ, et al. Association scan of 14,500 nonsynonymous SNPs in four diseases identifies autoimmunity variants. Nat Genet. 2007;39(11):1329–37.

    Article  CAS  PubMed  Google Scholar 

  6. Li Z, Wu X, Leo PJ, De Guzman E, Akkoc N, Breban M, Macfarlane GJ, Mahmoudi M, Marzo-Ortega H, Anderson LK, et al. Polygenic risk scores have high diagnostic capacity in ankylosing spondylitis. Ann Rheum Dis. 2021;80(9):1168–74.

    Article  CAS  PubMed  Google Scholar 

  7. Assassi S, Reveille JD, Arnett FC, Weisman MH, Ward MM, Agarwal SK, Gourh P, Bhula J, Sharif R, Sampat K, et al. Whole-blood gene expression profiling in ankylosing spondylitis shows upregulation of toll-like receptor 4 and 5. J Rheumatol. 2011;38(1):87–98.

    Article  CAS  PubMed  Google Scholar 

  8. Alber S, Kumar S, Liu J, Huang ZM, Paez D, Hong J, Chang HW, Bhutani T, Gensler LS, Liao W. Single cell transcriptome and surface epitope analysis of ankylosing spondylitis facilitates disease classification by machine learning. Front Immunol. 2022;13:838636.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Karow F, Smiljanovic B, Grün JR, Poddubnyy D, Proft F, Talpin A, Hue C, Boland A, Deleuze JF, Garchon HJ, et al. Monocyte transcriptomes from patients with axial spondyloarthritis reveal dysregulated monocytopoiesis and a distinct inflammatory imprint. Arthritis Res Ther. 2021;23(1):246.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Liu S, Ji W, Lu J, Tang X, Guo Y, Ji M, Xu T, Gu W, Kong D, Shen Q, et al. Discovery of potential serum protein biomarkers in ankylosing spondylitis using tandem mass tag-based quantitative proteomics. J Proteome Res. 2020;19(2):864–72.

    Article  CAS  PubMed  Google Scholar 

  11. Lu YP, Zhang XL, Zheng F, Yun C, Zhu C, Cai W, Liu D, Hong X, Li Q, Hu B, et al. Quantitative proteomic analyses to reveal the key features of proteins in new onset ankylosing spondylitis patients. ACS Omega. 2020;5(32):20153–61.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  12. Yu Z, Hong X, Zhang X, Zheng F, Liu F, Xu H, Zhu C, Cai W, Liu D, Yin L, et al. Global proteomic analyses reveals abnormal immune regulation in patients with new onset ankylosing spondylitis. Front Immunol. 2022;13:838891.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  13. Billing AM, Ben Hamidane H, Bhagwat AM, Cotton RJ, Dib SS, Kumar P, Hayat S, Goswami N, Suhre K, Rafii A, et al. Complementarity of SOMAscan to LC-MS/MS and RNA-seq for quantitative profiling of human embryonic and mesenchymal stem cells. J Proteomics. 2017;150:86–97.

    Article  CAS  PubMed  Google Scholar 

  14. Biomarkers and surrogate endpoints. preferred definitions and conceptual framework. Clin Pharmacol Ther. 2001;69(3):89–95.

    Article  Google Scholar 

  15. Poddubnyy DA, Rudwaleit M, Listing J, Braun J, Sieper J. Comparison of a high sensitivity and standard C reactive protein measurement in patients with ankylosing spondylitis and non-radiographic axial spondyloarthritis. Ann Rheum Dis. 2010;69(7):1338–41.

    Article  CAS  PubMed  Google Scholar 

  16. Rahbar MH, Lee M, Hessabi M, Tahanan A, Brown MA, Learch TJ, Diekman LA, Weisman MH, Reveille JD. Harmonization, data management, and statistical issues related to prospective multicenter studies in Ankylosing spondylitis (AS): experience from the Prospective Study Of Ankylosing Spondylitis (PSOAS) cohort. Contemp Clin Trials Commun. 2018;11:127–35.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Lukas C, Landewe R, Sieper J, Dougados M, Davis J, Braun J, van der Linden S, van der Heijde D. Development of an ASAS-endorsed disease activity score (ASDAS) in patients with ankylosing spondylitis. Ann Rheum Dis. 2009;68(1):18–24.

    Article  CAS  PubMed  Google Scholar 

  18. Gold L, Ayers D, Bertino J, Bock C, Bock A, Brody EN, Carter J, Dalby AB, Eaton BE, Fitzwater T, et al. Aptamer-based multiplexed proteomic technology for biomarker discovery. PLoS ONE. 2010;5(12):e15004.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. SOMAscanTM Technical White Paper []

  20. Fischer R, Trudgian DC, Wright C, Thomas G, Bradbury LA, Brown MA, Bowness P, Kessler BM. Discovery of candidate serum proteomic and metabolomic biomarkers in ankylosing spondylitis. Mol Cell Proteomics. 2012;11(2):M111.013904.

    Article  PubMed  Google Scholar 

  21. Braga Emidio N, Brierley SM, Schroeder CI, Muttenthaler M. Structure, function, and therapeutic potential of the trefoil factor family in the gastrointestinal tract. ACS Pharmacol Transl Sci. 2020;3(4):583–97.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  22. Jäger S, Stange EF, Wehkamp J. Inflammatory bowel disease: an impaired barrier disease. Langenbecks Arch Surg. 2013;398(1):1–12.

    Article  PubMed  Google Scholar 

  23. Wright NA, Poulsom R, Stamp G, Van Noorden S, Sarraf C, Elia G, Ahnen D, Jeffery R, Longcroft J, Pike C, et al. Trefoil peptide gene expression in gastrointestinal epithelial cells in inflammatory bowel disease. Gastroenterology. 1993;104(1):12–20.

    Article  CAS  PubMed  Google Scholar 

  24. Aamann L, Vestergaard EM, Grønbæk H. Trefoil factors in inflammatory bowel disease. World J Gastroenterol. 2014;20(12):3223–30.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Shaoul R, Okada Y, Cutz E, Marcon MA. Colonic expression of MUC2, MUC5AC, and TFF1 in inflammatory bowel disease in children. J Pediatr Gastroenterol Nutr. 2004;38(5):488–93.

    Article  CAS  PubMed  Google Scholar 

  26. Furuta GT, Turner JR, Taylor CT, Hershberg RM, Comerford K, Narravula S, Podolsky DK, Colgan SP. Hypoxia-inducible factor 1-dependent induction of intestinal trefoil factor protects barrier function during hypoxia. J Exp Med. 2001;193(9):1027–34.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  27. Alison MR, Chinery R, Poulsom R, Ashwood P, Longcroft JM, Wright NA. Experimental ulceration leads to sequential expression of spasmolytic polypeptide, intestinal trefoil factor, epidermal growth factor and transforming growth factor alpha mRNAs in rat stomach. J Pathol. 1995;175(4):405–14.

    Article  CAS  PubMed  Google Scholar 

  28. Kinsella TD, Espinoza L, Vasey FB. Serum complement and immunoglobulin levels in sporadic and familial ankylosing spondylitis. J Rheumatol. 1975;2(3):308–13.

    CAS  PubMed  Google Scholar 

  29. Hou S, Qi J, Liao D, Zhang Q, Fang J, Zhou Y, Liu Y, Bai L, Zhang M, Kijlstra A, et al. Copy number variations of complement component C4 are associated with Behçet’s disease but not with ankylosing spondylitis associated with acute anterior uveitis. Arthritis Rheum. 2013;65(11):2963–70.

    Article  CAS  PubMed  Google Scholar 

  30. Yang C, Ding P, Wang Q, Zhang L, Zhang X, Zhao J, Xu E, Wang N, Chen J, Yang G, et al. Inhibition of complement retards ankylosing spondylitis progression. Sci Rep. 2016;6:34643.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  31. Hwang MC, Ridley L, Reveille JD. Ankylosing spondylitis risk factors: a systematic literature review. Clin Rheumatol. 2021;40(8):3079–93.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Zhang J, Xu R, Wu L, Jiang J. Expression and function of Toll-like receptors in peripheral blood mononuclear cells in patients with ankylosing spondylitis. Mol Med Rep. 2019;20(4):3565–72.

    CAS  PubMed  PubMed Central  Google Scholar 

  33. Li Z, Wu X, Leo PJ, De Guzman E, Akkoc N, Breban M, Macfarlane GJ, Mahmoudi M, Marzo-Ortega H, Anderson LK, et al. Polygenic Risk Scores have high diagnostic capacity in ankylosing spondylitis. Ann Rheum Dis. 2021;80(9):1168–74.

    Article  CAS  PubMed  Google Scholar 

  34. Ye L, Liu Y, Xiao Q, Dong L, Wen C, Zhang Z, Jin M, Brown MA, Chen D. MRI compared with low-dose CT scanning in the diagnosis of axial spondyloarthritis. Clin Rheumatol. 2020;39(4):1295–303.

    Article  PubMed  Google Scholar 

  35. Klingberg E, Nurkkala M, Carlsten H, Forsblad-d’Elia H. Biomarkers of bone metabolism in ankylosing spondylitis in relation to osteoproliferation and osteoporosis. J Rheumatol. 2014;41(7):1349–56.

    Article  PubMed  Google Scholar 

  36. Spoorenberg A, van der Heijde D, de Klerk E, Dougados M, de Vlam K, Mielants H, van der Tempel H, van der Linden S. Relative value of erythrocyte sedimentation rate and C-reactive protein in assessment of disease activity in ankylosing spondylitis. J Rheumatol. 1999;26(4):980–4.

    CAS  PubMed  Google Scholar 

  37. Gratacos J, Collado A, Filella X, Sanmarti R, Canete J, Llena J, Molina R, Ballesta A, Munoz-Gomez J. Serum cytokines (IL-6, TNF-alpha, IL-1 beta and IFN-gamma) in ankylosing spondylitis: a close correlation between serum IL-6 and disease activity and severity. Br J Rheumatol. 1994;33(10):927–31.

    Article  CAS  PubMed  Google Scholar 

  38. Li XL, Lin TT, Qi CY, Yuan L, Xia LP, Shen H, Lu J. Elevated serum level of IL-33 and sST2 in patients with ankylosing spondylitis: associated with disease activity and vascular endothelial growth factor. J Investig Med. 2013;61(5):848–51.

    Article  CAS  PubMed  Google Scholar 

  39. Mattey DL, Packham JC, Nixon NB, Coates L, Creamer P, Hailwood S, Taylor GJ, Bhalla AK. Association of cytokine and matrix metalloproteinase profiles with disease activity and function in ankylosing spondylitis. Arthritis Res Ther. 2012;14(3):R127.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  40. Turina MC, Yeremenko N, Paramarta JE, De Rycke L, Baeten D. Calprotectin (S100A8/9) as serum biomarker for clinical response in proof-of-concept trials in axial and peripheral spondyloarthritis. Arthritis Res Ther. 2014;16(4):413.

    Article  PubMed  PubMed Central  Google Scholar 

  41. Arends S, van der Veer E, Groen H, Houtman PM, Jansen TL, Leijsma MK, Bijzet J, Limburg PC, Kallenberg CG, Spoorenberg A, et al. Serum MMP-3 level as a biomarker for monitoring and predicting response to etanercept treatment in ankylosing spondylitis. J Rheumatol. 2011;38(8):1644–50.

    Article  CAS  PubMed  Google Scholar 

  42. Genre F, Lopez-Mejias R, Miranda-Filloy JA, Ubilla B, Carnero-Lopez B, Palmou-Fontana N, Gomez-Acebo I, Blanco R, Rueda-Gotor J, Pina T, et al. Osteoprotegerin correlates with disease activity and endothelial activation in non-diabetic ankylosing spondylitis patients undergoing TNF-alpha antagonist therapy. Clin Exp Rheumatol. 2014;32(5):640–6.

    PubMed  Google Scholar 

  43. Pedersen SJ, Sorensen IJ, Garnero P, Johansen JS, Madsen OR, Tvede N, Hansen MS, Thamsborg G, Andersen LS, Majgaard O, et al. ASDAS, BASDAI and different treatment responses and their relation to biomarkers of inflammation, cartilage and bone turnover in patients with axial spondyloarthritis treated with TNFalpha inhibitors. Ann Rheum Dis. 2011;70(8):1375–81.

    Article  CAS  PubMed  Google Scholar 

  44. Weisman MH, Witter JP, Reveille JD. The prevalence of inflammatory back pain: population-based estimates from the US National Health and Nutrition Examination Survey, 2009–10. Ann Rheum Dis. 2013;72(3):369–73.

    Article  PubMed  Google Scholar 

  45. Hwang MC, Zheng K, J. Assassi, S. Mohan, C. Reveille, J. Quantitative proteomic screening uncovers diagnostic and monitoring serum biomarkers of ankylosing spondylitis [abstract]. Arthritis Rheumatol. 2022;74(Supplement 9).

Download references


The authors would like to thank the staff and participants of the UTHealth Spondyloarthritis Program. An earlier version, in part, was presented at the American College of Rheumatology Convergence 2022 [45].


This study was supported by grants from the United States Department of Health and Human Services, National Institutes of Health (NIH), National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS), P01-052915–06, and from the Spondylitis Association of America.

Author information

Authors and Affiliations



MH, SA, JZ, CM, and JR made substantial contributions to the conception, design, and data interpretation of these proteomic studies. MH, JC, RC, and KV made substantial contributions to the acquisition and analysis of the serum proteomic data. All authors read and approved the final manuscript.

Corresponding author

Correspondence to Mark Hwang.

Ethics declarations

Ethics approval and consent to participate

All procedures performed in studies involving human participants were in accordance with the ethical standards of the University of Texas Institutional Review Board (HSC-MS-07–0022) and with the 1964 Helsinki Declaration and its later amendments or comparable ethical standards.

Consent for publication

Informed consent was obtained from all individual participants included in the study.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information


Additional file 1: Supplementary Table 1.

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 The Creative Commons Public Domain Dedication waiver ( applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and Permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Hwang, M., Assassi, S., Zheng, J. et al. Quantitative proteomic screening uncovers candidate diagnostic and monitoring serum biomarkers of ankylosing spondylitis. Arthritis Res Ther 25, 57 (2023).

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI:


  • Ankylosing spondylitis
  • Axial spondyloarthritis
  • Proteomics
  • Biomarkers
  • Diagnosis
  • Disease activity