Osteoarthritis and Cartilage (2008) 16, 579e583 2007 Osteoarthritis Research Society International. Published by Elsevier Ltd.

. All rights reserved. doi:10.1016/j.joca.2007.09.007

International Cartilage Repair Society

Effect of fatty acids on bone marrow lesions and knee cartilage in healthy, middle-aged subjects without clinical knee osteoarthritis
Y. Wang M.Med., Ph.D., Research Fellowy, A. E. Wluka M.B.B.S., F.R.A.C.P., Ph.D., Research Fellowyz, A. M. Hodge B.Agr.Sci., Grad.Dip.Diet., B.Sc., M.Env.Sc., Grad.Dip.Epi.Biostats., Research Fellowx, D. R. English Ph.D., Professork{, G. G. Giles Ph.D., Professork, R. OSullivan F.R.A.C.R.# and F. M. Cicuttini M.B.B.S., F.R.A.C.P., Ph.D., Professory*
y Department of Epidemiology and Preventive Medicine, Monash University, Central and Eastern Clinical School, Alfred Hospital, Melbourne, VIC 3004, Australia z Baker Heart Research Institute, Commercial Road, Melbourne, VIC 3004, Australia x Department of Medicine, University of Melbourne, St Vincents Hospital, Fitzroy, VIC 3065, Australia k Cancer Epidemiology Centre, The Cancer Council of Victoria, Carlton, VIC 3053, Australia { School of Population Health, University of Melbourne, Carlton, VIC 3053, Australia # MRI Unit, Mayne Health Diagnostic Imaging Group, Epworth Hospital, Richmond, VIC 3121, Australia Summary
Objective: There is evidence that omega-3 polyunsaturated fatty acids alleviate the progression of osteoarthritis (OA). However, little work has been done to investigate the effect of fatty acids on bone marrow lesions and knee cartilage in healthy subjects. We examined this in a cohort of healthy middle-aged subjects without clinical knee OA. Methods: Two hundred and ninety-three healthy adults without knee pain or injury were recruited from an existing community-based cohort. Intakes of fatty acids and food sources of these were estimated from a food frequency questionnaire at baseline. Tibial cartilage volume, tibial plateau bone area, tibiofemoral cartilage defects and bone marrow lesions were assessed approximately 10 years later using magnetic resonance imaging. Results: In multivariate analyses, higher intakes of monounsaturated fatty acids (OR 2.14, 95% CI 1.04e4.39, P 0.04), total (OR 1.77, 95% CI 1.13e2.77, P 0.01) and n-6 polyunsaturated fatty acids (OR 1.69, 95% CI 1.10e2.61, P 0.02) were associated with an increased risk of bone marrow lesions. Intake of fatty acids was not signicantly associated with cartilage volume or cartilage defects. Conclusion: These ndings support the dietary recommendation towards a shift to foods rich in n-3 polyunsaturated fatty acids in order to maintain an optimal balance between dietary n-3 and n-6 polyunsaturated fatty acids, which is also important in the prevention of atherosclerosis. Although our ndings will need to be conrmed in longitudinal studies, they suggest the potential of fatty acids to adversely effect the knee joint. 2007 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved. Key words: Fatty acids, Bone marrow lesions, Cartilage, Cartilage defects.

Introduction
Osteoarthritis (OA) is the most common form of joint disease and cause of musculoskeletal disability in the elderly1. The disease processes not only affect the articular cartilage, but also involve the entire joint, including the subchondral bone and soft tissues. There is growing recognition of the importance of nutritional factors in the maintenance of bone and joint health2. Nutrients and dietary supplements have been shown to be effective at relieving the symptoms

*Address correspondence and reprint requests to: Flavia Cicuttini, Department of Epidemiology and Preventive Medicine, Monash University, Central and Eastern Clinical School, Alfred Hospital, Melbourne, VIC 3004, Australia. Tel: 61-3-9903-0555; Fax: 61-3-9903-0556; E-mail: avia.cicuttini@med.monash.edu.au Received 20 May 2007; revision accepted 2 September 2007.

of OA, and some may have a role in inuencing the course of OA3. There is evidence that levels of fat and n-6 polyunsaturated fatty acids are elevated in OA bone, suggesting that lipids may play a signicant role in the pathogenesis of OA4. In contrast, n-3 polyunsaturated fatty acids have been shown to alleviate the progression of OA through an effect on the metabolism of articular cartilage5,6. Dietary supplementation with polyunsaturated fatty acids has been shown to decrease bone turnover and increase bone mineral density7,8. The relative amounts of polyunsaturated fatty acids may play a role in preserving skeletal integrity in older age, since a higher ratio of n-6 to n-3 polyunsaturated fatty acids has been shown to be associated with lower bone mineral density at the hip9. However, the mechanism by which polyunsaturated fatty acids affect the knee structure and consequently the risk of knee OA has not been fully elucidated. While saturated fatty acids have been shown to be associated with atherosclerosis 579

580 and cardiovascular diseases10, and monounsaturated fatty acids protect against atherosclerosis and reduce the risk of cardiovascular diseases11, there are no previous data on saturated fatty acids or monounsaturated fatty acids and joint health. Magnetic resonance imaging (MRI) has good tissue contrast and anatomical resolution12, thus allowing a non-invasive examination of the joint structure in pre-disease or early stages of OA. MRI can visualise joint structure directly13,14 and has been recognised as a valid, accurate and reproducible tool to measure articular cartilage volume14,15, cartilage defects16,17, and bone marrow lesions18,19, which have been shown to have an important role in OA15e17,19. Bone marrow lesions have been shown to be associated with knee pain18, increased cartilage defect score20, and radiographic progression of OA19. Cartilage defects predict cartilage loss in healthy subjects16 while loss of knee cartilage volume is a predictor of knee replacement21. A major strength of examining these structural features is that we can examine the state of knee from normal through to the pre-disease and early disease state. In this study, we utilise dietary data from the Melbourne Collaborative Cohort Study (MCCS)22 to examine the association of different types of fatty acids and food sources of these with bone marrow lesions and knee cartilage in healthy, community-based, middle-aged men and women with no clinical knee OA.
Patients and methods
SUBJECTS The study was conducted within the MCCS, which is a prospective cohort study of 41,528 residents of Melbourne, Australia aged between 27 and 75 years (99.3% were aged 40e69 years) at recruitment which occurred between 1990 and 1994, with the aim of examining the role of lifestyle and genetic factors in the risk of cancer and chronic diseases22. Participants were recruited via the electoral rolls (registration to vote is compulsory for adults in Australia), advertisements, and community announcements in local media (e.g., television, radio, and newspapers). Participants for this current study were recruited from the MCCS. Participants who presented the rst year of round three follow-up of the MCCS which commenced in 2003 were invited to attend the current study. As our intent was to investigate subjects with a healthy knee (i.e., no signicant current or past knee pain, injury, or disease), potential participants were excluded if they had had any of the following: a clinical diagnosis of knee OA as dened by American College of Rheumatology criteria23; knee pain lasting for >24 h in the last 5 years; a previous knee injury requiring non-weight bearing treatment for >24 h or surgery (including arthroscopy); or a history of any form of arthritis diagnosed by a medical practitioner. A further exclusion criterion was a contraindication to MRI including pacemaker, metal sutures, presence of shrapnel or iron lings in the eye, or claustrophobia. We used quota sampling whereby the recruitment ceased when our target sample of approximately 300 subjects was achieved. By the end of 2004, 297 eligible subjects were recruited into the current study. The study was approved by The Cancer Council Victorias Human Research Ethics Committee and the Standing Committee on Ethics in Research Involving Humans of Monash University. All participants gave written informed consent. ANTHROPOMETRIC AND DIETARY DATA Extensive information was collected at MCCS baseline (1990e1994) using questionnaires and physical measurements. Questionnaires covered demographic data and diet (via a 121-item food frequency questionnaire developed from a study of weighed food records24). Fatty acid intakes were calculated from the food frequency questionnaire using Australian food composition data25. For the analyses, foods were grouped as total meat, eggs, sh, dairy products (milk, yoghurt, cheese, custard, ice cream), or remained as single items (butter and margarine). Intakes of olive oil and vegetable or blended oil were calculated from the household consumption of each, accounting for the number of people in the household. All nutrient intakes reect those from food only without supplements. Weight was measured using electronic scales with bulky clothing removed. Height was measured using a stadiometer with shoes removed. Body mass index (BMI) (weight/height2, kg/m2) was calculated.

Y. Wang et al.: Fatty acids and knee structure

MRI AND THE MEASUREMENT OF BONE MARROW LESIONS, CARTILAGE VOLUME, CARTILAGE DEFECTS, AND BONE AREA During 2003e2004, each subject had an MRI performed on the dominant knee (dened as the lower limb the subject used to step off when walking). Knees were imaged on a 1.5-T whole body magnetic resonance unit (Philips 1.5 Tesla Intera, Philips Medical Systems, Eindhoven, the Netherlands) using a commercial transmitereceive extremity coil, with sagittal T1-weighted fat-suppressed 3D gradient recall acquisition and coronal T2-weighted fatsaturated acquisition as previously described15,18. Bone marrow lesions were dened as areas of increased signal intensity adjacent to subcortical bone in either the distal femur or the proximal tibia18,19. A lesion was identied as present if it appeared on two or more adjacent slices in either tibiofemoral compartment18,19. Two trained observers, who were blinded to the characteristics of subjects, together assessed the presence of lesions for each subject. The reproducibility for determination of bone marrow lesions was assessed by the same method used to measure bone marrow lesions, using 60 randomly selected knee MRIs (k value, 0.88; P < 0.001) from a different population measured on two occasions. Tibial cartilage volume was determined by image processing on an independent workstation using Osiris software (Geneva, Switzerland) as previously described14,15. The measurement was done by two independent trained observers. One observer measured all subjects, and the other observer did cross-checks, i.e., measured randomly selected subjects, measuring one out of ve subjects, in a blinded fashion. The coefcients of variation (CV) for cartilage volume measures were 2.1% for medial tibial and 2.2% for lateral tibial cartilage15. Interobserver reliability assessed in 99 subjects (expressed as intraclass correlation coefcient, ICC) was 0.88 and 0.92 for the medial and lateral tibial cartilage volumes, respectively. Cartilage defects of the medial and lateral tibial and femoral cartilages were graded on the MR images with a classication system previously described16,17. The measurement was done by a single trained observer, who measured all images in duplicate on separate occasions. A cartilage defect was identied as present if there was irregularity on cartilage surface or bottom with loss of cartilage thickness on at least two consecutive slices at any site of that compartment. Intraobserver reliability (expressed as ICC) was 0.90 for the medial tibiofemoral compartment and 0.89 for the lateral tibiofemoral compartment17. The reproducibility for determination of cartilage defects was assessed using the duplicate results, with k values of 0.92 for the medial tibiofemoral compartment and 0.93 for the lateral compartment (all P < 0.001). Tibial plateau cross-sectional area was used as a measure of tibial bone size from images reformatted in the axial plane using Osiris software, as previously described15,26. The measurement was done by two independent trained observers. One observer measured all subjects, and the other observer did cross-checks, i.e., measured randomly selected subjects, measuring one out of ve subjects, in a blinded fashion. CVs for the medial and lateral tibial plateau areas were 2.3% and 2.4%, respectively15. Interobserver reliability assessed in 33 subjects (expressed as ICC) was 0.95 and 0.86 for the medial and lateral tibial plateau bone areas, respectively.

STATISTICAL ANALYSES The descriptive statistics of the characteristics of study participants was tabulated. With 297 subjects, this study had 80% power to show a correlation as low as 0.16 between various risk factors and knee cartilage volume (alpha error 0.05, two-sided signicance), thus explaining up to 2.6% of the variance of cartilage volume. This study had 80% power to detect an odds ratio (OR) of 1.4 for cartilage defects or 1.7 for bone marrow lesions, associated with a one standard deviation (SD) increase in a continuous predictor (alpha error 0.05, two-sided signicance). Outcomes were tibial cartilage volume and the presence of tibiofemoral cartilage defects and bone marrow lesions. Tibial cartilage volume was initially assessed for normality before being regressed against intakes of foods and fatty acids. It showed normal distribution, thus linear regression was used. The presence/absence of tibiofemoral cartilage defects and bone marrow lesions were dichotomous outcomes, thus binary logistic regression was used. The assumptions of linear and logistic regression including collinearity, interaction effects, and the goodness of t were checked before using these analysis techniques. No violation of these assumptions was found. Participants with self-reported total energy intakes in the top or bottom 1% of the sex-specic distributions were excluded. Multivariate regression models were constructed to explore the relationship between foods or fatty acid intakes and bone marrow lesions and knee cartilage measures, adjusting for potential confounders of age, gender, BMI, and energy intake. There is evidence that age, gender, and BMI are associated with knee structure (cartilage volume, cartilage defects, and bone marrow lesions) and the risk of knee OA27. They may also be related to the dietary intakes of some nutrients. Furthermore, most of the dietary papers have adjusted for energy intake. So, we included these variables in the multivariate analyses since they were potential confounders. Intake of fatty acids were standardised so that the coefcients represent the effect of a one SD increment in intake. P-values of less than 0.05 were considered

Osteoarthritis and Cartilage Vol. 16, No. 5

to be statistically signicant. All analyses were performed using the SPSS statistical package (standard version 14.0, SPSS, Chicago, IL, USA).

581 acids were positively associated with tibial cartilage volume, except for n-6/n-3 ratio. However, there were no signicant associations between intake of fatty acids and tibial cartilage volume after adjusting for energy intake, age, gender, BMI, and tibial plateau bone area (Table III).
RELATIONSHIP BETWEEN FOOD GROUP INTAKE AND BONE MARROW LESIONS AND KNEE CARTILAGE

Results
Two hundred and ninety-seven participants entered the study. After excluding four subjects with energy intakes in the top or bottom 1% of the sex-specic distributions, there were 293 subjects, 63% females, aged 58.0 5.5 years with BMI 25.2 3.8 kg/m2 remaining in the analyses (Table I). There were no signicant differences between this population and the original MCCS population, which has the following prole: 61% females, aged 57.8 3.0 years, and BMI 25.7 3.8 kg/m2. There were no signicant differences in dietary intakes or other health-related behaviours such as smoking: 60% subjects never smoked in this population vs 57% in the MCCS population.
RELATIONSHIP BETWEEN FATTY ACID INTAKE AND THE PRESENCE OF BONE MARROW LESIONS

After adjusting for potential confounders, intakes of dairy products, eggs, meat, sh, margarine, butter, olive oil, and vegetable or blended oils were not signicantly associated with bone marrow lesions or knee cartilage volume or defects (results not shown). Adding lifestyle factors such as physical activity, education level, smoking, and alcohol consumption in the multivariate models did not alter the results (data not shown).

Discussion
In this population of healthy middle-aged people with no clinical knee OA, intakes of both monounsaturated fatty acids and polyunsaturated fatty acids were associated with an increased prevalence of bone marrow lesions. In the case of polyunsaturated fatty acids, this seemed to be due to an association of n-6 rather than n-3 polyunsaturated fatty acids. Bone marrow lesions are associated with static malalignment19 and related to increased bone mineral density locally28, emphasizing the likely relationship between joint loading and bone marrow lesions. Although there is increasing interest in bone marrow lesions apart from biomechanical factors, this study, to our knowledge, is the rst to examine the effect of nutritional factors on bone marrow lesions. We found associations of monounsaturated fatty acids and polyunsaturated fatty acids with increased risk of bone marrow lesions, which have been shown to be associated with knee pain and predictive of cartilage loss in knee OA18,19. In the case of polyunsaturated fatty acids, the association seemed to be due to n-6 rather than n-3 polyunsaturated fatty acids, since there was no association of n-3 polyunsaturated fatty acids with bone marrow lesions. Our ndings suggest an adverse effect of monounsaturated fatty acids and n-6 polyunsaturated fatty acids on knee bone in healthy people. The health of articular cartilage and bone is dependent upon the regular provision of nutrients, and diets decient in such nutrients have been shown to lead to arthropathy2,29. The effect of foods and nutrients on knee structure is likely to be complex. We observed that the intake of fatty acids affected knee bone rather than knee cartilage. Polyunsaturated fatty acids have been shown to play a role in bone metabolism and health7,8. Bone marrow lesions have an adverse effect on the risk of knee OA18,19. The pathogenesis of bone marrow lesions is unclear, but available histological evidence suggests that these lesions include bone marrow necrosis, necrotic or remodelled trabeculae, bone marrow brosis and oedema30. It may be that the relationships found between fatty acids and bone marrow lesions are mediated via a vascular effect. This may explain why the effect of polyunsaturated fatty acids was seen for n-6 polyunsaturated fatty acids, which are thought to be more atherogenic31, rather than n-3 polyunsaturated fatty acids. These subchondral bone vascular changes may alter the mechanical properties of the bone, and subsequently increase the susceptibility to cartilage

In univariate analyses, intakes of total and n-6 polyunsaturated fatty acids were positively associated with the presence of bone marrow lesions (P 0.02 and 0.03, respectively). After adjusting for energy intake, age, gender, and BMI, intake of total (OR 1.77, 95% CI 1.13e2.77, P 0.01) and n-6 polyunsaturated fatty acids (OR 1.69, 95% CI 1.10e2.61, P 0.02) remained positively associated with the presence of bone marrow lesions. Intake of monounsaturated fatty acids was positively associated with the presence of bone marrow lesions in the multivariate analyses (OR 2.14, 95% CI 1.04e4.39, P 0.04). Intake of saturated fatty acids and n-3 polyunsaturated fatty acids, and n-6/n-3 ratio were not signicantly associated with the presence of bone marrow lesions (Table II).
RELATIONSHIP BETWEEN FATTY ACID INTAKE AND KNEE CARTILAGE VOLUME AND DEFECTS

None of the fatty acids were signicantly associated with the presence of tibiofemoral cartilage defects in univariate analyses and after adjusting for energy intake, age, gender, BMI, and tibial cartilage volume. In contrast, all of the fatty
Table I Characteristics of study subjects Total (n 293) Age when MRI performed (years) Female, number (%) Variables in 1990e1994 Body mass index (kg/m2) Energy from dietary intake (kJ/d) Saturated fatty acids (g/d) Monounsaturated fatty acids (g/d) Polyunsaturated fatty acids (g/d) n-3 Polyunsaturated fatty acids (g/d) n-6 Polyunsaturated fatty acids (g/d) Variables in 2003e2004 Tibial cartilage volume (mm3) Presence of tibiofemoral cartilage defects, number (%) Tibial plateau bone area (mm2) Presence of tibiofemoral bone marrow lesions, number (%) 58.0 (5.5) 184 (63%) 25.2 9364.9 34.1 28.8 12.8 1.2 11.7 (3.8) (3067.5) (14.0) (10.8) (5.6) (0.4) (5.1)

3731 (1118) 181 (62%) 3302 (475) 39 (13%)

Values are reported as mean (SD), except for percentages.

582

Y. Wang et al.: Fatty acids and knee structure

Table II Relationship between fatty acids and the presence of bone marrow lesions Univariate analysis OR (95% CI) P-value 0.19 0.07 0.02 0.03 0.15 0.15 Multivariate analysis OR (95% CI)* 1.47 2.14 1.77 1.69 1.42 1.27 (0.69, (1.04, (1.13, (1.10, (0.88, (0.91, 3.13) 4.39) 2.77) 2.61) 2.28) 1.77) P-value 0.32 0.04 0.01 0.02 0.15 0.17

Saturated fatty acids Monounsaturated fatty acids Polyunsaturated fatty acids n-6 Polyunsaturated fatty acids n-3 Polyunsaturated fatty acids n-6/n-3 Ratio

1.25 1.35 1.44 1.41 1.27 1.26

(0.89, (0.97, (1.05, (1.04, (0.92, (0.92,

1.75) 1.86) 1.96) 1.91) 1.77) 1.74)

*OR of tibiofemoral bone marrow lesions being present per SD increase in the respective fatty acid intake after adjusting for energy intake, age, gender, and BMI.

breakdown32. Although atheromatous vascular disease has been suggested to be important in the progression of OA to severe joint damage33, no data have been available in early, pre-OA state. Our ndings in subjects with no clinical OA imply that vascular disease in subchondral bone may have a role in the pathogenesis of OA. However, these will need to be conrmed by longitudinal studies. This study has a number of limitations. We were able to measure dietary fatty acids in a valid fashion34. However, a single measure of nutrient intakes 10 years earlier was used as the exposure measure in our study. While not all studies have shown dietary stability in adults, there is some evidence that nutrient intake is relatively stable and tends to be more stable with increasing age35,36. Longitudinal studies have suggested that individuals may move towards a more healthy diet over time37,38. The participants in our study are likely to represent the healthier and more health conscious of those initially recruited into the MCCS, since they were selected from those who took part in the rst year of the followup. They may have already adopted a healthy diet and thus be less likely to change in this direction. Although selection bias towards the healthier subjects may have affected the estimates of nutrient intake and knee structure measures, it is unlikely that this would modify the relationships between

nutrient intakes and knee structure. While nutritional data collected 10 years earlier may have resulted in some misclassication of exposure, such misclassication is likely to have been non-differential in relation to knee structure, since only subjects with no history of knee symptoms or injury were included. Non-differential misclassication tends to underestimate the strength of any observed associations. The prospective design is also a potential strength of our study since a substantial period of time has elapsed between ascertainment of exposure to nutritional factors and development of outcome (cartilage and bone measures). In the current study we did not measure knee alignment, which has been shown to be associated with bone marrow lesions19. Another limitation of this study is that no information on dietary supplements was available, thus we were unable to adjust the effect of them in the statistical analyses. Our study suggests that fatty acids may have an adverse effect on the knee joint with an increase in bone marrow lesions. These ndings support the dietary recommendation towards a shift to foods rich in n-3 polyunsaturated fatty acids in order to maintain an optimal balance between dietary n-3 and n-6 polyunsaturated fatty acids, which is also important in the prevention of atherosclerosis. Although our ndings will need to be conrmed in longitudinal

Table III Relationship between fatty acids and knee cartilage volume and defects Univariate analysis regression coefcient/OR (95% CI) Saturated fatty acids Cartilage volume* Cartilage defectsy Monounsaturated fatty acids Cartilage volume* Cartilage defectsy Polyunsaturated fatty acids Cartilage volume* Cartilage defectsy n-6 Polyunsaturated fatty acids Cartilage volume* Cartilage defectsy n-3 Polyunsaturated fatty acids Cartilage volume* Cartilage defectsy n-6/n-3 Ratio Cartilage volume* Cartilage defectsy 194.2 (57.4, 331.0) 1.19 (0.91, 1.55) 265.7 (134.8, 396.7) 1.11 (0.86, 1.42) 291.0 (165.1, 416.8) 1.00 (0.78, 1.26) 275.5 (149.5, 401.5) 0.97 (0.77, 1.23) 283.8 (156.2, 411.4) 1.12 (0.87, 1.43) 58.9 (69.2, 186.9) 0.91 (0.72, 1.15) P-value Multivariate analysis regression coefcient/OR (95% CI) 2.1 (169.8, 165.6) 1.72 (0.96, 3.06) 8.0 (146.4, 162.4) 1.39 (0.83, 2.34) 46.3 (54.0, 146.6) 1.03 (0.74, 1.43) 48.9 (49.6, 147.4) 0.98 (0.71, 1.36) 65.6 (41.1, 172.3) 1.24 (0.86, 1.78) 5.8 (80.5, 68.9) 0.94 (0.73, 1.19) P-value

0.01 0.20 <0.001 0.42 <0.001 0.97 <0.001 0.80 <0.001 0.38 0.37 0.44

0.98 0.07 0.92 0.21 0.37 0.88 0.33 0.92 0.23 0.24 0.88 0.59

*Change in tibial cartilage volume (mm3) per SD increase in the respective fatty acid intake, before and after adjusting for energy intake, age, gender, BMI, and tibial plateau bone area. yOR of tibiofemoral cartilage defects being present per SD increase in the respective fatty acid intake, before and after adjusting for energy intake, age, gender, BMI, and tibial cartilage volume.

Osteoarthritis and Cartilage Vol. 16, No. 5 studies, they suggest the potential of fatty acids to adversely affect the knee joint.

583
16. Cicuttini F, Ding C, Wluka A, Davis S, Ebeling PR, Jones G. Association of cartilage defects with loss of knee cartilage in healthy, middle-age adults: a prospective study. Arthritis Rheum 2005;52(7): 2033e9. 17. Ding C, Garnero P, Cicuttini F, Scott F, Cooley H, Jones G. Knee cartilage defects: association with early radiographic osteoarthritis, decreased cartilage volume, increased joint surface area and type II collagen breakdown. Osteoarthritis Cartilage 2005;13(3): 198e205. 18. Felson DT, Chaisson CE, Hill CL, Totterman SM, Gale ME, Skinner KM, et al. The association of bone marrow lesions with pain in knee osteoarthritis. Ann Intern Med 2001;134(7):541e9. 19. Felson DT, McLaughlin S, Goggins J, LaValley MP, Gale ME, Totterman S, et al. Bone marrow edema and its relation to progression of knee osteoarthritis. Ann Intern Med 2003;139(5 Pt 1):330e6. 20. Hunter DJ, Zhang Y, Niu J, Goggins J, Amin S, LaValley MP, et al. Increase in bone marrow lesions associated with cartilage loss: a longitudinal magnetic resonance imaging study of knee osteoarthritis. Arthritis Rheum 2006;54(5):1529e35. 21. Cicuttini FM, Jones G, Forbes A, Wluka AE. Rate of cartilage loss at two years predicts subsequent total knee arthroplasty: a prospective study. Ann Rheum Dis 2004;63(9):1124e7. 22. Giles GG, English DR. The Melbourne Collaborative Cohort Study. IARC Sci Publ 2002;156:69e70. 23. Altman R, Asch E, Bloch D, Bole G, Borenstein D, Brandt K, et al. Development of criteria for the classication and reporting of osteoarthritis. Classication of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association. Arthritis Rheum 1986;29(8):1039e49. 24. Ireland P, Jolley D, Giles G. Development of the Melbourne FFQ: a food frequency questionnaire for use in an Australian prospective study involving and ethnically diverse cohort. Asia Pac J Clin Nutr 1994;3: 19e31. 25. RMIT Lipid Research Group. Fatty Acid Compositional Database. Brisbane: Xyris Software; 2001. 26. Wang Y, Wluka AE, Cicuttini FM. The determinants of change in tibial plateau bone area in osteoarthritic knees: a cohort study. Arthritis Res Ther 2005;7(3):R687e93. 27. Felson DT, Lawrence RC, Dieppe PA, Hirsch R, Helmick CG, Jordan JM, et al. Osteoarthritis: new insights. Part 1: the disease and its risk factors. Ann Intern Med 2000;133(8):635e46. 28. Lo GH, Hunter DJ, Zhang Y, McLennan CE, Lavalley MP, Kiel DP, et al. Bone marrow lesions in the knee are associated with increased local bone density. Arthritis Rheum 2005;52(9):2814e21. 29. Okma-Keulen P, Hopman-Rock M. The onset of generalized osteoarthritis in older women: a qualitative approach. Arthritis Rheum 2001; 45(2):183e90. 30. Zanetti M, Bruder E, Romero J, Hodler J. Bone marrow edema pattern in osteoarthritic knees: correlation between MR imaging and histologic ndings. Radiology 2000;215(3):835e40. 31. Yamashita T, Oda E, Sano T, Yamashita T, Ijiru Y, Giddings JC, et al. Varying the ratio of dietary n-6/n-3 polyunsaturated fatty acid alters the tendency to thrombosis and progress of atherosclerosis in apoE/ LDLR/ double knockout mouse. Thromb Res 2005; 116(5):393e401. 32. Simkin PA. Bone pain and pressure in osteoarthritis joints. In: Chadwick DJ, Goode J, Eds. Osteoarthritic Joint Pain. Chichester: Wiley 2004:179e86. 33. Conaghan PG, Vanharanta H, Dieppe PA. Is progressive osteoarthritis an atheromatous vascular disease? Ann Rheum Dis 2005;64(11): 1539e41. 34. Hodge AM, Simpson JA, Gibson RA, Sinclair AJ, Makrides M, ODea K, et al. Plasma phospholipid fatty acid composition as a biomarker of habitual dietary fat intake in an ethnically diverse cohort. Nutr Metab Cardiovasc Dis 2007;17(6):415e26. 35. Goldbohm RA, van t Veer P, van den Brandt PA, van t Hof MA, Brants HA, Sturmans F, et al. Reproducibility of a food frequency questionnaire and stability of dietary habits determined from ve annually repeated measurements. Eur J Clin Nutr 1995;49(6): 420e9. 36. Fernyhough LK, Horwath CC, Campbell AJ, Robertson MC, Busby WJ. Changes in dietary intake during a 6-year follow-up of an older population. Eur J Clin Nutr 1999;53(3):216e25. 37. Osler M, Heitmann BL, Schroll M. Ten year trends in the dietary habits of Danish men and women. Cohort and cross-sectional data. Eur J Clin Nutr 1997;51(8):535e41. 38. Zhang H, Hsu-Hage BH, Wahlqvist ML. Longitudinal changes in nutrient intakes in the Melbourne Chinese Cohort Study. Public Health Nutr 2002;5(3):433e9.

Conict of interest
None.

Acknowledgements
This study was funded by a programme grant from the National Health and Medical Research Council (NHMRC; 209057) and was further supported by infrastructure provided by The Cancer Council of Victoria. This study was also supported by NHMRC (project grant 334150), Colonial Foundation and Shepherd Foundation. Dr Wang is the recipient of an NHMRC Public Health (Australia) Fellowship (NHMRC 465142). Dr Wluka is the recipient of an NHMRC Public Health (Australia) Fellowship (NHMRC 317840) and co-recipient of the Cottrell Fellowship, Royal Australasian College of Physicians. We would especially like to thank the study participants who made this study possible.
References
1. Lawrence R, Helmick C, Arnett F, Deyo R, Felson D, Giannini E, et al. Estimates of the prevalence of arthritis and selected musculoskeletal disorders in the United States. Arthritis Rheum 1998; 41(5):778e99. 2. Goggs R, Vaughan-Thomas A, Clegg PD, Carter SD, Innes JF, Mobasheri A, et al. Nutraceutical therapies for degenerative joint diseases: a critical review. Crit Rev Food Sci Nutr 2005;45(3):145e64. 3. Wang Y, Prentice LF, Vitetta L, Wluka AE, Cicuttini FM. The effect of nutritional supplements on osteoarthritis. Altern Med Rev 2004;9(3): 275e96. 4. Plumb MS, Aspden RM. High levels of fat and (n-6) fatty acids in cancellous bone in osteoarthritis. Lipids Health Dis 2004;3:12. 5. Curtis CL, Hughes CE, Flannery CR, Little CB, Harwood JL, Caterson B. n-3 Fatty acids specically modulate catabolic factors involved in articular cartilage degradation. J Biol Chem 2000;275(2):721e4. 6. Curtis CL, Rees SG, Little CB, Flannery CR, Hughes CE, Wilson C, et al. Pathologic indicators of degradation and inammation in human osteoarthritic cartilage are abrogated by exposure to n-3 fatty acids. Arthritis Rheum 2002;46(6):1544e53. 7. Weaver CM, Peacock M, Johnston CC Jr. Adolescent nutrition in the prevention of postmenopausal osteoporosis. J Clin Endocrinol Metab 1999;84(6):1839e43. 8. Kruger MC, Coetzer H, de Winter R, Gericke G, van Papendorp DH. Calcium, gamma-linolenic acid and eicosapentaenoic acid supplementation in senile osteoporosis. Aging (Milano) 1998;10(5):385e94. 9. Weiss LA, Barrett-Connor E, von Muhlen D. Ratio of n-6 to n-3 fatty acids and bone mineral density in older adults: the Rancho Bernardo Study. Am J Clin Nutr 2005;81(4):934e8. 10. Bemelmans WJ, Lefrandt JD, Feskens EJ, Broer J, Tervaert JW, May JF, et al. Change in saturated fat intake is associated with progression of carotid and femoral intimaemedia thickness, and with levels of soluble intercellular adhesion molecule-1. Atherosclerosis 2002;163(1):113e20. 11. Mensink RP, Katan MB. Effect of a diet enriched with monounsaturated or polyunsaturated fatty acids on levels of low-density and highdensity lipoprotein cholesterol in healthy women and men. N Engl J Med 1989;321(7):436e41. 12. Peterfy CG. MR imaging. Baillieres Clin Rheumatol 1996;10(4):635e78. 13. Eckstein F, Schnier M, Haubner M, Priebsch J, Glaser C, Englmeier KH, et al. Accuracy of cartilage volume and thickness measurements with magnetic resonance imaging. Clin Orthop 1998;352:137e48. 14. Wluka AE, Stuckey S, Snaddon J, Cicuttini FM. The determinants of change in tibial cartilage volume in osteoarthritic knees. Arthritis Rheum 2002;46(8):2065e72. 15. Jones G, Glisson M, Hynes K, Cicuttini F. Sex and site differences in cartilage development: a possible explanation for variations in knee osteoarthritis in later life. Arthritis Rheum 2000;43(11):2543e9.