Osteoarthritis and injury in the knee and hip

Context: Osteoarthritis (OA) is one of the most common causes of disability in the United States, but effective interventions for this disease are limited. Difficulty with defining OA, including the lack of sensitive measures of incident OA, hinders the development of successful treatments. Biomarkers of cartilage turnover and inflammation may be important, sensitive indicators of OA, but biomarker levels may be influenced by joint injury. Additionally, previous efforts to examine injury as a risk factor for OA, either with cross-sectional or longitudinal study designs, have resulted in a wide range of estimates. Understanding the association between injury and OA and exploring new, sensitive measures of OA have important clinical implications for patients who have sustained joint injuries, for those seeking treatment for OA, and for identifying participants for clinical trials of OA. Objectives: This research was conducted in a large, community based sample to: 1) determine whether the biomarkers of cartilage oligomeric matrix protein (COMP), hyaluronan (HA), high-sensitivity C-reactive protein (hsCRP), and keratan sulfate (KS) predict incident radiographic OA at the knee and whether their associations vary by lower extremity injury history, 2) examine the hazard of incident OA of the knee and hip by history of injury, and 3) compare prevalence and hazard estimates from cross-sectional and longitudinal designs. Methods: A longitudinal community-based cohort completed baseline (1991-1997) clinical evaluation and identical follow-up assessment (1999-2003; median follow-up time = 5.6 years, range=3.0-13.1 years) for OA. The OA outcomes examined for Objectives 1 were: radiographic OA (Kellgren-Lawrence [K-L] grade of 2 or greater), osteophyte (OST) formation, and joint space narrowing (JSN), and for Objectives 2 and 3 were: radiographic OA (based on K-L grade), chronic joint symptoms, radiographic OA or symptoms, and radiographic OA with symptoms. History of knee injury or hip injury was based on self-report. Baseline serum COMP, HA, hsCRP, and KS were measured for 803 participants. The study groups with injury data comprised 1,570 participants with linked baseline and follow-up knee radiographs and 1,446 participants with linked hip radiographs. Among participants with biomarker data, linked baseline and follow-up radiographs were available for 542 knees at risk for incident OA, 349 knees at risk for incident osteophyte (OST) formation, and 440 knees at risk for incident joint space narrowing (JSN). For biomarker analyses, Cox regression models were used to estimate the hazard ratio (HR) for a 1-unit increase in the natural log of each biomarker, adjusting for age, race, gender, and body mass index and exploring history of lower extremity injury and chronic knee symptoms as potential modifiers. For injury and OA analyses, Cox regression models were used to determine the hazard of incident knee and hip OA outcomes and logistic regression models were use to estimate the prevalence of knee and hip OA outcomes. Results: The hazard of incident knee OA and incident knee OST increased with higher baseline lnCOMP levels, and the hazard of incident knee JSN also increased with higher lnCOMP levels and higher lnHA levels. Higher levels of lnhsCRP and lnKS did not predict the incident knee outcomes. Neither history of lower extremity injury nor chronic knee symptoms was a strong modifier of these associations. The hazard of incident knee radiographic OA and OA or symptoms was higher among participants with knee injury at baseline compared to those without injury. Injury to the knee was associated with incident ipsilateral and contralateral radiographic knee OA. Hip injury also predicted a higher hazard of hip radiographic OA outcomes, although precision was lower for the hip analyses due to small numbers of outcomes. HR and odds ratio (OR) patterns were similar for some knee injury-OA associations, but the OR estimates tended to be higher than the HR estimates. The pattern of HRs and ORs is less clear for the hip due to low numbers for the longitudinal analysis. Conclusions: Higher baseline lnCOMP and lnHA levels predicted incident knee OA, OST, and JSN, but history of lower extremity injury did not modify the association. The hazard of incident radiographic knee and hip OA was higher in those with a history of injury compared to those without injury. Knee injury predicted an increased hazard of incident radiographic knee OA in the ipsilateral joint and the contralateral joint. Estimates calculated from a cross-sectional analysis may provide a fair approximation of the incidence of radiographic knee OA by knee injury status, but further research over a longer follow-up period, particularly for the hip, is needed to help clarify these comparisons