ObjectiveTo assess the longitudinal stability of biomarker-based molecular endotypes of knee osteoarthritis (KOA) participants from APPROACH and to evaluate the consistency of findings in an independent KOA population. MethodsNineteen biomarkers were measured longitudinally in 295 KOA participants from the APPROACH cohort. K-means clustering was used to identify the structural damage, inflammation, and low tissue turnover endotypes at the six-, 12-, and 24-month follow-ups. Endotype stability was defined as having the same independent endotype assignment longitudinally for patients with complete data (n = 226). Clinical and biochemical characteristics were compared between participants with longitudinally stable and unstable endotypes. The presence and longitudinal stability of the endotypes were evaluated in a different KOA population from the placebo arm of the oral salmon calcitonin trials. ResultsAn average overall longitudinal endotype stability of 55% (Fleiss’ Kappa of 0.53; 95% CI: 0.46, 0.60) was demonstrated. An average stability of 59% (range: 54% - 59%) was observed for the structural damage endotype (Fleiss’ Kappa 0.52; 95% CI: 0.45, 0.60), 54% (52% - 56%) for the inflammatory (Fleiss’ Kappa 0.61; 95% CI: 0.53, 0.68), and 50% (49% - 52%) for the low tissue turnover endotype (Fleiss’ Kappa 0.46; 95% CI: 0.39, 0.54). Participants with longitudinally unstable endotypes exhibited molecular properties of more than one endotype, which were detectable already at the first visit. ConclusionsOur study showed for the first time that more than half of KOA participants exhibited a longitudinally stable endotype, highlighting the applicability of biomarker-based endotyping in a clinical trial setting.
Read full abstract