Article Text

PDF
SAT0499 Identification of biochemical phenotypes in knee osteoarthritis: longitudinal data from the fnih oa biomarker consortium
  1. B Jeremiasse1,
  2. PM Welsing1,
  3. C Fellows2,
  4. FP Lafeber1,
  5. WE Van Spil1
  1. 1Department of Rheumatology and Clinical Immunology, UMC Utrecht, Utrecht, Netherlands
  2. 2School of Veterinary Medicine, University of Surrey, Guildford, United Kingdom

Abstract

Background It is hypothesized that patients with knee osteoarthritis (OA) can be classified into different phenotypes. Knowledge of these phenotypes may contribute to developing effective targeted treatment strategies.

Objectives To identify different longitudinal phenotypes of knee OA using biochemical markers and to compare these phenotypes with regard to radiographic joint space loss (JSL) and/or pain progression.

Methods Baseline, 1-year, and 2-year biochemical marker data from the FNIH OA Biomarker Consortium were used. This consortium is a nested case-control study of 600 subjects with one symptomatic index knee showing radiographic OA changes of Kellgren and Lawrence grade 1 to 31. Subjects were classified as either JSL progressors, pain progressors, JSL+pain progressors, or non-progressors according to predefined criteria (pain progression=persistent increase in WOMAC pain≥9 points (0–100 scale) from baseline to 2, 3 or 4 years, JSL progression=decrease in JSW≥0.7mm from baseline to 2, 3 or 4 years).

Biochemical markers included in the current analysis were sCTX-I, uαCTX-I, uβCTX-I, sNTX-I, uCTX-II, sCPII, sC2C, sC1,2C, sColl2–1 NO2, sCOMP, sHA, and sMMP (u=urinary, s=serum). First, using principal component analysis (PCA), the individual markers were reduced into a number of clusters of markers (components) that may represent common underlying domains. Second, a hierarchical cluster analysis (HCA) was performed to differentiate between longitudinal courses (phenotypes) of these marker clusters. The optimum number was determined from the additive value of each newly identified phenotype as compared to already identified phenotypes. Third, these longitudinal phenotypes were compared with regard to percentages of patients in each of the JSL and/or pain progression categories.

Results PCA showed an optimal solution of three components. Looking at the markers that loaded maximally onto each of the components, they were interpreted as cluster of bone (sCTX-I, uαCTX-I, uβCTX-I, sNTX-I, uCTX-II), cartilage (sCPII, sC2C, sC1,2C, sColl2–1 NO2), and synovial (sCOMP, sHA, and sMMP) metabolism, respectively. HCA revealed an optimum of seven longitudinal phenotypes. Based on the relative predominance of the component(s) throughout follow-up, phenotypes were named “high bone”, “high cartilage”, “high synovium”, “low cartilage”, “low synovium”, “low bone, cartilage and synovium” and “low bone and high synovium” phenotype, respectively (Figure 1, dendrogram and heatmap). Phenotypes differed with regard to percentages of patients in JSL and/or pain progression categories (Figure 1, pie charts) as well as other demographic, clinical, and radiographic parameters (data not shown).

Conclusions Seven longitudinal phenotypes of knee OA could be identified based on biochemical markers representing bone, cartilage and synovial metabolism. These phenotypes showed relevant differences in other characteristics, such as JSL and/or pain progression.

References

  1. Kraus VB, Collins JE, Hargrove D, et al. Predictive validity of biochemical biomarkers in knee osteoarthritis: data from the FNIH OA Biomarkers Consortium. Ann Rheum Dis. 2016.

References

Disclosure of Interest None declared

Statistics from Altmetric.com

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.