The nutritional epidemiology of osteoarthritis and musculoskeletal pain : an analysis of data from two UK population cohorts

Osteoarthritis (OA) is a leading cause of pain and disability in the UK for which there is, at present, no effective pharmacological treatment and no cure. Given the prevalence of the disease, there is increasing interest in the impact of nutrition and its potential to contribute both to prevention and treatment. Existing data in this area is both inconsistent and limited. This research investigates associations in two large population based cohorts, in cross-sectional and longitudinal settings, with a particular focus on biological pathways known to contribute to disease development. Data available contained radiographic OA and pain variables in the TWINS-UK cohort, and clinician classified OA and pain variables in the EPIC-Norfolk cohort. Both groups had data on serum lipids and information on dietary exposure obtained through the use of food frequency questionnaires. Genetic data were also available in the EPIC-Norfolk cohort. A cross-sectional analysis was performed in 20517 adults from the EPIC-Norfolk cohort and in 8351 females from the TWINS-UK cohort. Longitudinal analysis investigated incidence and change in hip and knee pain over time in EPIC-Norfolk. Key associations were explored further in a Mendelian randomisation analysis in an attempt to clarify the potential causal relationships between putative exposures and outcomes. The main finding from the FFQ cross-sectional analysis in TWINS-UK is that a ‘Traditional English’ dietary pattern showed a 28% increase in the risk of hand OA. An unexpected result was that the ‘Fruit and Vegetable’ dietary pattern was suggested to increase the risk of knee OA by 19%. Further analysis into the Mediterranean dietary pattern revealed a protective association with OA decreasing the risk of radiographic hand OA by 12%. Analysis of serum lipid levels showed that triglycerides increased the risk of hip OA, hand OA, elbow and forearm pain, and chronic widespread pain. Cholesterol and LDL were also consistently found to increase the risk of OA/pain and the LDL component apoB1 was found to significantly increase the risk of hand OA. Findings from the case-control and longitudinal analysis in EPIC-Norfolk, revealed that increased BMI is strongly associated with hip and knee OA (p<0.0001) and with onset of hip pain (p<0.0001) and knee pain (p=0.0001). Triglycerides were also found to increase knee OA (p=0.023) and knee pain (p<0.0001) in the case-control analysis and were found to increase knee pain in females (p=0.0013). Findings also showed that cholesterol levels increased the risk of future knee pain (p=0.016) and increased vegetable consumption was inversely associated with change in hip pain. The findings from the Mendelian Randomisation study, using SNPs from FTO and APOA5 to investigate reverse causality between BMI and triglycerides and OA, showed no causal associations with OA. In conclusion, data from these two cohorts suggests that a diet high in saturated fats increases the risk of OA/pain as do the serum levels of lipids. Increased consumption of a Mediterranean diet is suggested to be protective of hand OA. Although BMI is strongly associated with hip and knee OA and incidence and change in hip/knee pain over time, both BMI and triglycerides were found not to be causal of OA when explained by genetic variants responsible for changes in these two exposures. Although wider replication is needed to confirm these observations, the results from this research suggest potential dietary modifications for prevention and treatment of OA and pain associated with OA that could be tested further in randomised trials.