Protein biomarkers associated with enthesitis in psoriatic arthritis

Background: The presence of enthesitis in PsA may provide clues to disease pathogenesis, aid diagnosis and contribute to poor prognosis. Ultrasound is an invaluable tool in assessing entheseal disease, but it requires training and is time consuming. We investigated whether enthesitis may be measured by changes in serum proteins and as such allow us to better understand, diagnose and treat the condition.

Objectives: To use mass spectrometry (MS) based proteomics to identify protein biomarkers that may be associated with those PsA patients who suffer from enthesitis.

Methods: Serum was collected as part of a prospective observational study from PsA patients who fulfilled the CASPAR criteria and were ≥ 18 yrs. All patients were biologic naive and due to commence biologic treatment. We utilised theMAdrid Sonographic Enthesitis Index (MASEI) to assess US confirmed enthesitis with a score of ≥ 18 suggesting significant enthesitis and <18 determined to be non-significant. The samples were analysed using targeted MS multiple reaction monitoring (MRM) analysis with > 200 candidate proteins. The MRM data was analysed in 3 ways: 1) Raw with no normalisation 2) Normalisation to 7 stable isotopically labelled peptide spike-ins (SIL7) correcting for fluctuations in sample injections/mass spectrometry loading; and, 3) Normalisation to a set of endogenous peptides that represent total serum protein abundance (TSPA) correcting for different amounts of total serum across samples. Univariate analyses and multivariate machine learning Random Forest (RF) modelling were performed.

Results: 80 samples were analysed. 29 patients had a MASEI score of < 18[36.25%] and 51 had a score > 18 [63.75%]. We identified 35 candidate proteins from all data sets that are included in Figure 1. RF multivariate analysis of alldata revealed a set of peptide signatures with the ability to differentiate between MASEI scores < 18 vs ≥ 18. RF models generated from the peptide data had a testing and training AUC of 0.789 [95% CI 0.65, 0.93] 0.953 [95% CI 0.92, 0.98](Raw), 0.83 [CI 0.74, 0.94], 0.972 [95% CI 0.96, 0.99] (TSPA7) and 0.845 [95%CI 0.72, 0.97]0.966 [95% CI 0.94, 0.99] (SIL7) respectively.

Conclusion: We have identified serum proteins, within this small cohort,which are associated with enthesitis in PsA. Verification of these findings in a larger, independent dataset is the next required step. Proteins related to peptide sequences for univariate + multivariate analysis. Blue (all data sets) Green(all univariate data sets and at least one multivariate) Yellow (all multivariate data sets and at least one univariate) Grey (all multivariate data sets) Red (all Univariate data sets) Pink (2 Multivariate data sets) ANT3: Adenine nucleotide translocase 3 AGT:Angiotensinogen A2HSGP: Alpha 2 HS glycoprotein ALAI:Apolipoprotein A-I ARP3: Angiopoietin related protein 3 AT3: Antithrombin IIICNCC: Carboxypeptidase N catalytic chain CBG:Corticosteroid binding globulinGP3: Glutathione peroxidase 3 HSPB: Heat Shock Protein HSP 90-beta HRG:Histidine rich glycoprotein IGFBPC Insulin-like growth factor-binding protein complex acid labile subunit PEDF: Pigment epithelium derived factor SHBG: Sexhormone binding globulin PGSPD: Phosphatidylinositol-glycan-specific phospholipase D TX: Tenascin-X VDBP Vitamin D binding protein.