Background Sleep disturbance has been shown to contribute to systemic inflammation, as well as increased pain sensitization. The role of sleep disturbance in knee osteoarthritis (OA) has not been established. Sleep disturbance may be an aggravating contributor to and/or a consequence of knee OA symptoms, with inflammation serving as a potential mediator.
Objectives To compare knee symptoms over nine years of follow-up by frequency of sleep disturbance at baseline, among participants with or at risk for radiographic knee OA (ROA), and to estimate the association between the presence of inflammation on knee MRI and pain that disturbs sleep among knees that developed ROA.
Methods Knees from the Osteoarthritis Initiative (OAI) with or at risk for ROA were included in this longitudinal analysis. Self-reported frequency of restless sleep in the past week was assessed using a CES-D question. Knee symptoms were assessed annually using the WOMAC. Knees that developed incident ROA (i.e., ≥2 Kellgren-Lawrence grade, KLG) through four years of follow-up were assessed for effusion-synovitis on non-contrast-enhanced 3T MRI using the MRI Osteoarthritis Knee Score (MOAKS) from annual clinic visits between four years prior to incident ROA and up to one year after ROA detection. Effusion-synovitis represents a combination of joint effusion and synovial thickening on fluid-sensitive sequences and scored as 0 (normal), 1 (mild), 2 (moderate), or 3 (severe). Annual mean WOMAC total score was estimated using a mixed model with participant and knee treated as random effects, in 5,028 knees at risk for ROA at baseline, and in 3,893 knees with ROA at baseline. Log-binomial regression with generalized estimating equations was used to estimate the association between effusion-synovitis and knee pain in bed that disturbs sleep, adjusted for age, sex, and BMI in a sample of 355 knees with an average of 3.8 MRI assessments.
Results There was dose-dependent effect, with participants reporting restless sleep 1–2 days, 3–4 days, and 5–7 days in the past week having higher mean WOMAC total scores compared to those who reported <1 day of restless sleep (i.e., difference in means: 2.5 [95% CI: 1.3 to 3.8], 5.1 [95% CI: 3.2 to 7.1], and 10.1 [95% CI: 7.6 to 12.6], respectively) among knees with ROA (KLG≥2) at baseline. Differences in average WOMAC total score between groups were relatively persistent over nine years (Figure 1). A similar dose-dependent effect of restless sleep was observed among knees at risk of ROA (i.e., KLG 0 or 1). Among knees that developed incident ROA, those with mild effusion-synovitis had a 52% higher risk of knee pain in bed that disturbs sleep at the same visit (RR=1.52; 95% CI: 1.13 to 2.04), while knees that had moderate/severe effusion-synovitis had more than double the risk of knee pain that disturbs sleep (RR=2.55; 95% CI: 1.87 to 3.47), compared to knees with no MRI-detected effusion-synovitis.
Conclusions Restless sleep was associated with knee symptoms and disability in a dose-dependent manner, with average levels persistent over nine years of follow-up among knees with and at risk of ROA. Effusion-synovitis was associated with pain that disturbs sleep among knees that developed incident ROA. Sleep disturbance and knee inflammation may be important targets for interventions in knee osteoarthritis.
Disclosure of Interest C. K. Kwoh Grant/research support from: Abbie, EMD Serono, E. Ashbeck: None declared, A. Guermazi Grant/research support from: Boston Imaging Core Laboratory, Consultant for: MerckSerono, TissueGene, Ortho Trophix, AstraZeneca, Genzyme, F. Roemer Grant/research support from: Boston Imaging Core Laboratory, S. Parthasarathy: None declared