POS1349 SYNOVIAL-FLUID BASED PROTEOMIC SIGNATURES OF BIOLOGICAL AGING IN KNEE OSTEOARTHRITIS: IDENTIFICATION OF AGE-ASSOCIATED PATHWAYS AND PREDICTIVE MARKERS USING STEpUP OA

Background: Osteoarthritis (OA) is a chronic, age-related disease characterised by joint degeneration. The synovial fluid (SF) proteome in OA changes significantly with disease severity (manuscript in revision). However, the specific contribution of ageing biology to the proteomic alterations in OA remain unknown. Identifying age-associated protein signatures within the diseased joint may help us understand how the molecular mechanisms in OA change with age and how therapeutic strategies might need to be modified accordingly in different age groups. Objectives: Our aim was to identify proteins associated with participant age in knee OA SF using STEpUP OA, a large highly phenotyped cohort of individuals with OA, to explore molecular pathways modulated by age. We also aimed to develop a predictive model based on SF proteomics to estimate the ‘biological age' in a given OA patient and to explore other contributing factors. Methods: Through STEpUP OA, we used the SomaScan® v4.1 Assay (SomaLogic, Inc, Boulder, Colorado) to analyse the SF proteome (quantifying 7,289 proteins per sample) from 1361 participants with knee OA (by x-ray and/or joint symptoms) divided into Discovery (N=708, mean age (SD): 64.4 yrs (11.5)) and Replication (N=653, mean age (SD): 65.5 yrs (10.3)) datasets. Linear regression models, adjusted for biological sex, with and without additional adjustment for advanced radiographic disease status (Kellgren-Lawrence (KL) grades: 0-2 vs 3-4) and cohort (random effect), were used to examine associations between SF proteomes and participant age (continuous, years). To identify proteins associated with age, data were first analysed in Discovery and Replication datasets separately, followed by verification in the Combined dataset. Pathway enrichment analyses (using Hallmark, KEGG, and Gene Ontology gene datasets) of all proteins identified biological pathways influenced by age. Elastic net regression was used to create a model predicting participant age with model accuracy assessed by Pearson correlation analysis between predicted and actual participant age. The prediction model was built by randomly splitting the Combined OA dataset into two equally sized datasets; one used to train the model, and the second to evaluate the model. P-values were corrected for multiple testing by Benjamini-Hochberg. Results: In the Combined dataset, we identified 1360 age-associated proteins (adjusted p-value ≤0.05), many of which had been previously linked to aging processes such as oxidative stress and cellular aging. Notably, Growth differentiation factor 15 (MIC-1 also known as GDF15), R-spondin-1 (RSPO1), and Chordin-like protein 1 (CRDL1) emerged as top associating proteins, with MIC-1 known to have a crucial role in stress responses, mitochondrial dysfunction, and oxidative damage, which are described in both aging and OA. 279 of these protein associations were replicated across Discovery and Replication datasets. Associations were stable after additional adjustment for advanced radiographic disease status (Figure 1A). The correlation of corresponding protein associations before and after radiographic disease status adjustment was high when considering all proteins (r=0.97, p<2.2e -16 ) (Figure 1B). Pathway enrichment analysis of the Combined dataset identified several age-related biological processes, including epithelial-mesenchymal transition (EMT), inflammatory response, and coagulation, although these pathways did not replicate (Figure 1C). The age prediction model, constructed using 209 proteins identified by elastic net regression, showed a strong correlation (r=0.80) between predicted and actual participant age (Figure 1D). Of the 209 protein predictors, 94 of these proteins remained significantly associated with age in the model adjusting for advanced radiographic disease status, indicating that these proteins appeared to reflect biological aging independently of OA severity. Prediction using a parsimonious panel of the 5 most highly associated proteins (based on size of coefficient, largest to smallest), which included CRDL1, PCD10:ECD, sFRP-3, SCF sR and MIC-1, was slightly weaker than using all 209 proteins (r=0.69 vs 0.80). Conclusion: This study provides valuable insights into the relationship between biological age and the SF proteome in OA, revealing key proteins and pathways even after adjustment for radiographic disease severity. Although some biological pathways did not replicate, the identified proteins show robust associations with age related processes. The successful development of a biological age prediction model using SF proteomics identifies a potential tool for stratifying patients with high biological age. Although anti-ageing therapeutics are in development across a number of disease areas, specific mechanistic studies in OA will be required to understand whether such pathways can be targeted to modulate OA biology for clinical benefit. REFERENCES: NIL . Figure 1 (A ) Protein abundance was measured in 1,321 samples, adjusted for biological sex, radiographic severity, and spin-status (by Combat). Volcano plot shows beta estimates vs. p-values for proteins associated with age in the Combined dataset. Red/blue indicate positive/negative associations (adjusted p-value ≤ 0.05). Top 30 proteins (15 positive/negative) are labelled, with replicating proteins shown in orange. (B) Scatter plot of beta estimates with/without radiographic adjustment. (C ) Bubble plot of differentially expressed Hallmark pathways (adjusted p-value < 0.05). (D ) Scatter plot of predicted vs. actual age using 209 proteins (correlation = 0.80, p-value < 2.2e −16 ). Acknowledgements: We would like to express our gratitude and thanks to all cohort participants who contributed samples to STEpUP OA. We are grateful for the support from Floris Lafeber and Simon Mastbergen (Utrecht Medical Centre) for provision of samples. We thank the Oxford Knee Surgery Team. We thank Gretchen Brewer for her administrative support of the consortium. The STEpUP OA Consortium author block includes: University of Nottingham: Ana M. Valdes, David A. Walsh, Michael Doherty, Vasileios Georgopoulos; Lund University: Staffan Larsson, L. Stefan Lohmander, André Struglics; University of Cambridge: Brian D.M. Tom, Laura Bondi; University of Toronto: Mohit Kapoor, Rajiv Gandhi, Anthony Perruccio, Y. Raja Rampersaud, Kim Perry; University of Manchester: Tim Hardingham, David Felson; University of Oxford: Tonia L. Vincent, Thomas A. Perry, Luke Jostins-Dean, Yun Deng, Vicky Batchelor, Jennifer Mackay-Alderson, Gretchen Brewer, Rose M. Maciewicz, Brian Marsden, Nigel K. Arden, Philippa A. Hulley, Andrew J. Price, Stefan Kluzek, Megan Goff, Vinod Kumar, James Tey, Tamas Szommer; Imperial College London: Fiona E. Watt, Andrew Williams, Artemis Papadaki; University College Maastricht: Tim J. Welting, Pieter Emans, Tim Boymans, Liesbeth Jutten, Marjolein Caron, Guus van den Akker; University of Western Ontario: C. Thomas Appleton, Trevor B. Birmingham, J. Daniel Klapak; Biosplice: Sarah Kennedy, Jeymi Tambiah; Fidia: Devis Galesso, Nicola Giordan; SomaLogic: Joe Gogain, Darryl Perry, Anna Mitchel, Ela Zepko; Novartis: Sophie Brachat, Joanna Mitchelmore, Juerg Gasser, Lori Jennings; UCB: Waqar Ali. Disclosure of Interests: Thomas A Perry: None declared , Philippa Hulley: None declared , Rose Maciewicz Owns shares in AstraZeneca, Previously worked for AstraZeneca, Joanna Mitchelmore Owns shares in Novartis, Works for Novartis, Staffan Larsson: None declared , Joe GoGain Owns shares in SomaLogic, Works for SomaLogic, Sophie Brachat Owns shares in Novartis, Works for Novartis, André Struglics: None declared , C. Thomas Appleton Has received honoraria for educational purposes from Novartis, Has received consultancy fees from Novartis, Stefan Kluzek: None declared , David T. Felson: None declared , Mohit Kapoor: None declared , Stefan Lohmander Has received consultancy fees from Arthro Therapeutics AB, and was an advisory board member of AstraZeneca, Tim Welting Owns shares in Chondropeptix BV, David A. Walsh Has received honoraria for educational purposes from Pfizer Ltd and AbbVie Inc, Has received consultancy fees from GlaxoSmithKline plc, AKL Research & Development Limited, Pfizer Ltd, Eli Lilly and Company, Contura International, and AbbVie Inc, Ana Valdes: None declared , Fiona E. Watt Has received consultancy fees from Pfizer, Brian Tom: None declared , Tonia Vincent Has received grant income for STEpUP OA from industry partners including: Galapagos, Biosplice, Novartis, Fidia, UCB, Pfizer (non-consortium member) and Somalogic (in kind contributions). © The Authors 2025. This abstract is an open access article published in Annals of Rheumatic Diseases under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Neither EULAR nor the publisher make any representation as to the accuracy of the content. The authors are solely responsible for the content in their abstract including accuracy of the facts, statements, results, conclusion, citing resources etc.

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.