- Email: [email protected]
Metabolic syndrome and trajectory of knee pain in older adults
Osteoarthritis and Cartilage xxx (xxxx) xxx
Metabolic syndrome and trajectory of knee pain in older adults F. Pan y *, J. Tian y, F. Cicuttini z, G. Jones y y Menzies Institute for Medical Research, University of Tasmania, Private Bag 23, Hobart, Tasmania 7000, Australia z Department of Epidemiology and Preventive Medicine, Monash University Medical School, Commercial Road, Melbourne 3181, Australia
a r t i c l e i n f o
s u m m a r y
Article history: Received 15 December 2018 Accepted 15 May 2019
Objectives: To examine the association of metabolic syndrome (MetS) and its components with knee pain severity trajectories. Methods: Data from a population-based cohort study were utilised. Baseline blood pressure, glucose, triglycerides and high-density lipoprotein (HDL) cholesterol were measured. MetS was defined according to the National Cholesterol Education Program-Adult Treatment Panel III criteria. Radiographic knee osteoarthritis (ROA) was assessed by X-ray. Pain severity was measured by the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain questionnaire at each time-point. Groupbased trajectory modelling was used to identify pain trajectories and multi-nominal logistic regression was used for analysis. Mediation analysis was performed to assess whether body mass index (BMI)/ central obesity mediated the association between MetS, its components and pain trajectories. Results: Among 985 participants (Mean ± SD age: 62.9 ± 7.4, 50% female), 32% had MetS and 60% had ROA. Three pain trajectories were identified: ‘Minimal pain’ (52%), ‘Mild pain’ (33%) and ‘Moderate pain’ (15%). After adjustment for potential confounders, central obesity increased risk of belonging to both ‘Mild pain’ and ‘Moderate pain’ trajectories as compared to the ‘Minimal pain’ trajectory group, but MetS [relative risk ratio (RRR): 2.26, 95%CI 1.50e3.39], hypertriglyceridemia (RRR: 1.75, 95%CI 1.16e2.62) and low HDL (RRR: 1.67, 95%CI 1.10e2.52) were only associated with ‘Moderate pain’ trajectory. BMI/central obesity explained 37e70% of these associations. Results were similar in those with ROA. Conclusion: MetS and its components are predominantly associated with worse pain trajectories through central obesity, suggesting that the development and maintenance of worse pain trajectories may be caused by MetS. © 2019 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Keywords: Metabolic syndrome Pain Trajectory Knee osteoarthritis
Introduction Musculoskeletal conditions are prevalent and are the most common causes of years lived with disability (YLD) worldwide, with musculoskeletal pain and osteoarthritis (OA) being the most frequent conditions1. It is estimated that the disability due to musculoskeletal conditions will continue to rise with an increasingly obese and ageing population2. The causes of musculoskeletal pain are complex and heterogeneous, but OA is the most common cause of pain especially in elderly3. Obesity is an established risk factor for OA and severity of symptoms4. It has been long proposed that the pathophysiological
* Address correspondence and reprint requests to: F. Pan, Menzies Institute for Medical Research, University of Tasmania, Private Bag 23, Hobart, Tasmania 7000, Australia. Tel: 61-3 6226-7700; Fax: 61-3 6226-7704. E-mail addresses: [email protected] (F. Pan), [email protected] (J. Tian), [email protected] (F. Cicuttini), [email protected] (G. Jones).
mechanisms underlying obesity-related OA are due to a mechanical overload, but metabolic and systemic inflammation related to adipose tissue have been considered additional mechanisms, leading to a new subtype of OA e metabolic OA5,6. Metabolic syndrome (MetS), generally consisting of increased waist circumference, dyslipidaemia (elevated triglycerides and reduced high-density lipoprotein (HDL)), hypertension and elevated serum glucose, has shown a relatively consistent association with the development and progression of OA7e12 and joint replacement13,14 before adjustment for body mass index (BMI) in previous studies, although the positive associations between MetS and its components and OA disappeared or weakened after controlling for BMI. Furthermore, most prior studies have been cross-sectional, limiting the ability to determine causality and raising the possibility that OA and MetS simply co-exist through shared risk factors such as age and obesity. Pain, the most prominent symptom of OA, is a leading cause of restriction in physical function, impaired mobility, and reduced
https://doi.org/10.1016/j.joca.2019.05.030 1063-4584/© 2019 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
2
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
quality of life in older adults15. It is also a major reason for patients to seek healthcare. The severity of OA-related pain and disability are associated with an increased risk of all-cause death16. This highlights that understanding of pain mechanisms is more clinically relevant especially when there is no proven curative treatment to slow or halt the progression of OA. In the musculoskeletal field, there are few reports investigating the relationship between MetS and OA-related pain10,12,17e20 and musculoskeletal pain21e26, most of which have been cross-sectional and the results have been inconsistent. Pain is often considered to get worse at the group level with time using traditional statistical approaches for longitudinal data analysis; however, pain at the individual level may be stable, improved or worsening. Prior studies including our own research have demonstrated that pain consists of homogeneous subgroups following distinct pain trajectories and pain level in each trajectory appears to be stable27e31. This suggests that pain is a nonprogressive symptom and maintains its level unless important factors are targeted. Group-based trajectory modelling (GBTM) can estimate and identify pain patterns and unobserved distinct subgroups of individuals with similar trajectories32. Given this, traditional analyses assuming that there is only one homogenous group may lead to misleading results. In this context, examining the association between MetS and pain trajectories in a longitudinal study may aid the understanding of underlying mechanisms of development and maintenance of pain, and thus allows for early interventions to shift pain trajectory to favour course33. The aim of study, therefore, was to examine the association between MetS and its components and pain trajectories in a general older population. Material and methods Participants Data utilized in this study were from the Tasmanian Older Adult Cohort Study (TASOAC), which is a longitudinal, population-based cohort study. Participants aged 50e80 years were randomly selected from the electoral roll in Southern Tasmania (population n ¼ 229,000) Australia in 2002. Of 2,530 subjects selected, a total of 1,099 participants were recruited and had their baseline examinations (including questionnaire, general interview, clinical assessment and blood tests). 875, 768 and 563 participants attended the second, third and fourth clinic during a mean follow-up period of 2.6, 5.1 and 10.7 years, respectively, where questionnaires, general interview and clinical assessment were completed, and blood samples were collected. The study was approved by the Southern Tasmanian Health and Medical Human Research Ethics Committee and written informed consent of all participants was obtained.
ilium in the mid-auxiliary plane. Blood pressure was measured using sphygmomanometer (Omron HEMe907) on the right arm with the correct sized cuff according to the arm circumference and at least 5 min has elapsed since participant was seated before recording the first blood pressure measurement. Measurement of systolic and diastolic blood pressure was then made two times with at least 30 s interval. The mean of two times measurements was used for the analysis. Venous fasting blood samples were taken from the antecubital vein and the concentrations of HDL, glucose and triglycerides were analysed on an Olympus AU5400 automated analyser (Olympus Optical, Tokyo, Japan). According to the National Cholesterol Education Program-Adult Treatment Panel III criteria, MetS is defined as the presence of 3 components out of the following: high waist circumference (102 cm in men and 88 cm in women), low HDL (<1.03 mmol/L in men and <1.29 mmol/L in women), hypertension (systolic blood pressure 130 mm Hg or diastolic blood pressure 85 mm Hg), hyperglycemia (fasting plasma glucose >5.6 mmol/L) and hypertriglyceridemia (triglycerides 1.7 mmol/L). Self-reported prescribed medication by a doctor was collected. Antihypertensive, antidiabetic and lipid-lowering medication were extracted. We also defined MetS by considering medication use. Participants receiving antihypertensive, antidiabetic and/or lipidlowering medication were classified as having hypertension, hyperglycemia, and/or low HDL and hypertriglyceridemia. Measurement of covariates Weight and height were measured, and BMI (kg/m2) was then calculated. Pedometer (Omron HJe003 & HJe102, Omron Healthcare, Kyoto, Japan) was used to measure physical activity (steps/day) for consecutive 7-day, as previously described35. Self-reported employment status was collected and grouped into two categories: ‘employed’ was defined as those with full/part-time job; home duties, student, sole parent/disability pension, retired or unemployed were considered ‘unemployment’. Highest education level the participants had completed was categorised as: ‘low level’ was those who completed school only, ‘medium level’ was those with a trade/vocational certificate, and ‘high level’ was those who had a university degree or above. Smoking history was assessed by the question, ‘Have you ever been a ‘regular smoker’? A ‘regular smoker’ is defined as someone who has smoked at least seven cigarettes, cigars or pipes each week for at least 3-month?’. Participants had a standing anteroposterior semiflexed radiograph of the right knee. The Altman atlas was used to score osteophytes and joint space narrowing (JSN) on a 0e3 point scale, as previously described36. Any score of JSN and/or osteophytes 1 was defined as radiographic knee OA (ROA). Statistical analysis
Measurement of outcome Knee pain The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain questionnaire was used to assess knee pain. WOMAC pain questionnaire consists of five items including walking on flat surface, going up or down stairs, at night in the bed, sitting or lying and standing upright. Each item scores on a 10-point scale from 0 to 934. A total pain score is a sum of five items (0e45) with higher scores indicating greater pain. Measurement of factors comprising the MetS Waist circumference measurement (cm) was taken at the halfway between the bottom of the lowest rib and the top of the
Identifying pain trajectories Unobserved subgroups of individuals with similar pain trajectories over a-10.7-year course were identified using GBTM. GBTM utilises finite mixture modelling and multinomial modelling strategy to identify homogenous subgroups across a study population32, where trajectory parameters (i.e., shape and number of trajectory groups) are tested using maximum likelihood. We used censored normal models to estimate pain trajectories with considering possible trajectory shapes (linear, quadratic, or cubic) and number of trajectories (two to six). The Bayesian information criterion (BIC) and Aikaike's information criterion (AIC) was used to determine the best optimal model, with lower BIC and AIC values (absolute value) indicating a better model fit. Other criteria were also considered in the model selection process including: (1) model
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
3
interpretability:each trajectory group had >10% of the total sample; (2) pragmatic evaluation: posterior probabilities were >0.732. Posterior probabilities refer to the average probability of each participant that was assigned to in a given trajectory. Descriptive analysis Continuous and categorical variables were reported as Mean ± SD and percentages as appropriate.The characteristics of participants were compared across pain trajectory groups using ANOVA or c2 test where appropriate. Association between MetS and pain trajectory groups Age, sex, physical activity, smoking history, unemployment, education level and ROA were considered potential confounders based on existing literature within the field. Univariable and multivariable multi-nominal logistic regression were used to examine the associations between MetS and each of components and pain trajectory groups, where relative rate ratios (RRRs) and 95% confidence intervals (CIs) were produced. Mediation analysis was performed by using the KarlsonHolm-Breen decomposition method37,38 for multinomial outcome to determine whether obesity (BMI/central obesity) is the main mediating factor of the association between MetS and its components and pain trajectories. This method is not affected by the rescaling or attenuation issues that arise from crossmodel comparison of nonlinear models38e40. Mediation analysis was only performed for statistically significant associations between MetS and its components and pain trajectories in multivariable analyses. Secondary analyses were also conducted by restricting analyses to those with and without ROA. Sensitivity analyses were also performed on those with data at all four time-points. We also reanalysed the data using MetS definition based on use of medication. Stata V.15 was used for all analyses. P values 0.05 were regarded as statistically significant. Results There were 1,099 participants who attended the baseline examination including questionnaires, clinical assessment and blood tests. Trajectory model analysis was conducted for 985 participants who had complete baseline data. Of these, 802, 706 and 526 participants had complete pain data at the following three follow-ups, respectively. 60% of participants had ROA with a mean WOMAC pain score of 3.6 (SD ¼ 6.1) at baseline; 32% of participants had MetS. Participants who werelost to follow-up were older, had a greater BMI, fewer steps per day and a greater WOMAC pain score, and had a larger proportion of participants who were ever smoked, being unemployed, had low education level and had high waist circumference, hyperglycemia and MetS compared to those who completed the study (Supplementary Table I). Fig. 1 shows three distinct pain trajectory groups identified from the GBTM analysis. Pain severity within each of trajectories was not different but the pattern appeared similar with no trajectories showing a substantial pain improvement or worsening over time. There was almost half of participants (‘Minimal pain’ group) with a consistently low level of pain, approximately one third of participant (‘Mild pain’ group) with a mild level of pain consistent throughout the follow-up and 14% of participants (‘Moderate pain’) with a relatively high level of pain consistent throughout the follow-up. All groups included at least of 10% of participants with a clear difference in pain score suggesting that three-group model is a well-separated and optimal model. In addition, the average posterior probabilities
Fig. 1. Three distinct trajectories of WOMAC pain over 10.7 years.
were 0.97 (‘Minimal pain’), 0.95 (‘Mild pain’) and 0.97 (‘Moderate pain’), indicating a good model fit. Table I describes the characteristics of participants at baseline across three pain trajectory groups. ‘Moderate pain’ trajectory group appeared to have the highest BMI, had the least physical activity, had the greatest WOMAC pain score and the largest proportion of participants who were unemployed, had a low education level and had a ROA. Further, the ‘Moderate pain’ trajectory group was more likely to have the largest proportion of participants with central obesity, low HDL and MetS defined by the clinical criteria. BMI, waist circumference, WOMAC pain score, the proportion of participants with unemployed status, low education level, ROA, low HDL and MetS increased across ‘Minimal’, ‘Mild’ and ‘Moderate’ pain trajectory groups; alternatively, physical activity decreased across pain trajectory groups. The associations of MetS and its components with pain trajectories in the whole population regardless of ROA status are displayed in Table II. In univariable analysis, MetS, central obesity and low HDL were associated with both the ‘Mild pain’ and ‘Moderate pain’ trajectory groups; hypertriglyceridemia increased risk of being in the ‘Moderate pain’ trajectory relative to the ‘Minimal pain’ trajectory group. After adjustment for potential confounders including age, sex, physical activity, smoking history, unemployment, education level and ROA, the association of central obesity with both the ‘Mild pain’ and ‘Moderate pain’ trajectory groups remained statistically significant, whereas ‘Moderate pain’ trajectory group was only associated with MetS (RRR: 2.26, 95%CI 1.50e3.39], hypertriglyceridemia (RRR: 1.75, 95%CI 1.16e2.62) and low HDL (RRR: 1.67, 95%CI 1.10e2.52) as compared to the ‘Minimal pain’ trajectory group. The effect size of MetS and its components apart from hyperglycaemia tended to be higher with the ‘Moderate pain’ than that with ‘Mild pain’ trajectory group. Mediation analysis was performed to evaluate whether the association of MetS and its components with pain trajectories is mediated through BMI/central obesity. BMI was found to mediate the associations between central obesity, hypertriglyceridemia, low HDL and MetS and pain trajectories; the estimated mediation effect explained 37, 38, 38, 52% of each association, respectively (Table III). Central obesity also mediated 39e70% of the association between hypertriglyceridemia, low HDL and MetS and pain trajectories. Secondary analyses in those with ROA did not reveal additional pain trajectory groups. As compared to the ‘Minimal pain’ trajectory group, similar associations between MetS and its components and
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
4
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
Table I Characteristics of the three pain trajectory groups Characteristics
Minimal pain (N ¼ 512)
Mild pain (N ¼ 328)
Moderate pain (N ¼ 145)
Age, years Female (%) BMI (kg/m2) Physical activity (steps/day) Smoking history (%) Unemployment (%) Education level (%) School only Vocation training University or higher Radiographic knee OA (%) WOMAC pain score Waist circumference, cm Systolic BP, mm Hg Diastolic BP, mm Hg Fasting glucose, mmol/L Triglycerides, mmol/L HDL cholesterol, mmol/L High waist circumference* (%) Hypertensiony (%) Hyperglycemiaz (%) Hypertriglyceridemiax (%) Low HDLk (%) MetS¶ (%)
62.9 ± 7.4 48 26.8 ± 4.1 8910.1 ± 3363.4 52 57
63.0 ± 7.6 51 28.3 ± 4.6 8550.7 ± 3261.8 47 61
62.8 ± 7.2 57 30.1 ± 5.8 7839.7 ± 3523.7 52 72
56 30 14 57 0.3 ± 0.8 91.7 ± 12.5 140.0 ± 19.5 85.3 ± 10.1 5.3 ± 1.1 1.7 ± 4.0 1.4 ± 0.4 33 73 22 28 25 28
53 36 11 65 3.9 ± 3.2 95.0 ± 13.1 140.6 ± 21.4 84.9 ± 11.0 5.3 ± 1.1 1.5 ± 1.0 1.4 ± 0.4 47 70 24 30 31 35
64 31 4 64 14.7 ± 8.0 99.3 ± 14.3 139.6 ± 18.8 84.8 ± 10.0 5.3 ± 1.2 1.7 ± 0.9 1.3 ± 0.4 63 74 22 38 35 45
Bold denotes statistically significant result determined by ANOVA or Chi-square test as appropriate. BMI, body mass index; OA, osteoarthritis; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index; BP blood pressure; HDL, high density lipoprotein; MetS, Metabolic syndrome. * Defined as waist circumference 102/88 cm for men/women. y Defined as systolic or diastolic BP 130/85 mm Hg. z Defined as fasting plasma glucose 5.6 mmol/L. x Defined as triglycerides 1.7 mmol/L. k Defined as HDL<1.03/1.29 mmol/L for men/women. ¶ Defined as presence of three or more components above.
risk of being worse pain trajectory groups were observed in univariable and multivariable analysis with adjustment for potential confounders (Table IV). Further adjustment for BMI or central obesity showed a protective effect of hypertension with ‘Mild pain’ trajectory group, all other associations was attenuated and became statistically non-significant. Similarly, we found greater effect sizes of MetS and its components except for hyperglycaemia with the ‘Moderate pain’ than with ‘Mild pain’ trajectory. Sensitivity analyses performed on those with data at all four time-points found that MetS and central obesity (but not hypertriglyceridemia and low HDL) were associated with the ‘Moderate pain’ trajectory relative to the ‘Minimal pain’ trajectory group after adjustment for potential confounders.
We observed similar results based on a medication use definition (Supplementary Tables II and III); however, the protective effect of hypertension with the ‘Mild pain’ trajectory disappeared. Discussion Three knee pain trajectory groups were identified among community-dwelling elderly individuals and those with ROA over a course of 10.7-year in the current study. In multivariable analysis, central obesity was associated with an increased risk of being both the ‘Mild pain’ and ‘Moderate pain’ trajectory group relative to ‘Minimal pain’ trajectory group; whereas MetS and its components (hypertriglyceridemia and low HDL) only increased risk of being in
Table II Metabolic syndrome, its components, and knee pain trajectories among participants regardless of radiographic knee osteoarthritis status over 10.7 years (N ¼ 985) RRR (95%CI)* Mild vs Minimal pain Univariable Abdominal obesity Hypertension Fasting hyperglycaemia Hypertriglyceridemia Low HDL MetS
Model 1
Moderate vs Minimal pain Model 2
Model 3
Univariable
Model 1
Model 2
Model 3
1.69 (1.29, 2.22) 1.68 (1.25, 2.25) 1.30 (0.87, 1.94) 3.28 (2.28, 4.70) 3.19 (2.12, 4.80) 2.07 (1.18, 3.64) 0.89 (0.66, 1.19) 0.78 (0.56, 1.08) 0.71 (0.51, 0.99) 0.73 (0.52, 1.02) 0.99 (0.67, 1.47) 1.05 (0.67, 1.66) 0.90 (0.56, 1.43) 0.89 (0.56, 1.43) 1.12 (0.82, 1.54) 1.12 (0.79, 1.58) 0.95 (0.67, 1.35) 0.98 (0.69, 1.40) 0.97 (0.63, 1.49) 1.06 (0.66, 1.70) 0.74 (0.45, 1.22) 0.75 (0.46, 1.23) 1.18 (0.88, 1.57) 1.04 (0.76, 1.43) 1.43 (1.06, 1.92) 1.32 (0.96, 1.82) 1.47 (1.10, 1.96) 1.32 (0.96, 1.80)
0.93 (0.68, 1.29) 1.20 (0.87, 1.67) 1.02 (0.73, 1.44)
0.97 (0.70, 1.34) 1.79 (1.25, 2.57) 1.75 (1.16, 2.62) 1.44 (0.95, 2.18) 1.24 (0.89, 1.72) 1.86 (1.29, 2.70) 1.67 (1.10, 2.52) 1.41 (0.93, 2.16) 0.95 (0.66, 1.38) 2.22 (1.54, 3.20) 2.26 (1.50, 3.39) 1.49 (0.95, 2.33)
1.42 (0.93, 2.15) 1.37 (0.89, 2.10) 1.29 (0.81, 2.07)
Bold denotes statistically significant result; HDL, high density lipoprotein; MetS, Metabolic syndrome. *RRR (95%CI): relative risk ratio (95% confidence interval) for belonging in each trajectory relative to reference trajectory (Minimal pain). Model 1: adjusted for age, sex, physical activity, smoking history, unemployment, education level and radiographic knee osteoarthritis. Model 2: Model 1 þ body mass index. Model 3: Model 1 þ abdominal obesity.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
5
Table III Estimated mediation of the association between metabolic syndrome, its components, and knee pain trajectories by body mass index/central obesity in whole population (N ¼ 985) Moderate vs Minimal pain
BMI Total effect Direct effect Indirect effect Mediation percentage Central obesity Total effect Direct effect Indirect effect Mediation percentage
Abdominal obesity
Hypertriglyceridemia
Low HDL
MetS
3.20 (2.12, 4.83) 2.07 (1.18, 3.64) 1.54 (1.04, 2.29) 37
1.79 (1.18, 2.71) 1.44 (0.95, 2.18) 1.25 (1.12, 1.40) 38
1.74 (1.14, 2.66) 1.41 (0.93, 2.16) 1.23 (1.10, 1.38) 38
2.30 (1.52, 3.47) 1.49 (0.95, 2.33) 1.55 (1.26, 1.90) 52
1.77 (1.17, 2.67) 1.42 (0.93, 2.15) 1.25 (1.12, 1.39) 39
1.78 (1.17, 2.72) 1.37 (0.89, 2.10) 1.30 (1.15, 1.47) 46
2.32 (1.53, 3.49) 1.29 (0.81, 2.07) 1.79 (1.38, 2.34) 70
Table shows relative risk ratio and 95% confidence intervals; HDL, high density lipoprotein; MetS, Metabolic syndrome; BMI, body mass index. Indirect effect represents mediation of BMI/central obesity in the association between MetS, its components, and knee pain trajectories. Mediation percentage represents percentage of the total effect explained by the mediation of BMI/central obesity.
the ‘Moderate pain’ trajectory group. These associations were markedly attenuated and became statistically non-significant with the ‘Moderate pain’ trajectory group after further adjusting for BMI or central obesity. Further, mediation analysis shows that BMI/ central obesity is an important mediator of the association between MetS and hypertriglyceridemia and low HDL and pain trajectory. These findings suggest that the potential mechanistic links between MetS and its components and pain severity trajectory are mediated through BMI (largely by central obesity). To the best of our knowledge, this is the first study examining MetS and its components with pain trajectories in a general older population and among those with ROA. The findings that three distinct pain trajectories were identified in this study were consistent with prior studies which identified three to five stable pain severity trajectories, showing that the level of pain does not markedly change over time at a group level27,28,30,41. Combining prior studies with this study, our results suggest that pain appears to be a non-progressive symptom and on average, pain levels at baseline are maintained overtime unless important risk factors are modified. There was a slight difference in the number of pain trajectories between the current study and prior studies. This could be explained by differences in participants included, length of follow-up and pain measurement. However, clear differences in characteristics of participants and high posterior probabilities among three trajectories in the current study provide support for the construct and validity of identified trajectories. There is a growing body of literature existing on the role of MetS in relation to pathogenesis of OA. Few studies have reported its
relationship with OA-related pain10,12,17e20, all of which have been cross-sectional, and results are contradictory. This study goes beyond prior studies to investigate the role of MetS in the evolution of pain over time. Our finding that MetS was associated with an increased risk of belonging to worse pain trajectories before adjustment for BMI is in line with prior studies showing that pain intensity is higher in those with than without MetS. A more recent study with 60 knee OA patients by Afifi et al. reported that MetS was related to a higher level of WOMAC pain12. Li et al.19 reported that pain score was higher in those with greater number of metabolic disorders. In another study, Sowers et al.17 found that obese women with cardiometabolic clustering were more likely to report persistent knee pain compared with non-obese and obese without cardiometabolic clustering. The statistically significant association between MetS and worse pain trajectories was weakened and disappeared after controlling for BMI or central obesity. This is consistent with one recent study which found that MetS was associated with knee pain but this association disappeared after adjustment for BMI18. In contrast, Shin reported that accumulation of MetS components was associated with a higher intensity of OA pain even after considering BMI10. Although the discrepancy may be explained by differences in study design, characteristics of included participants, potential confounders considered and the use of questionnaires to assess pain, it likely reflects that MetS plays a different role in the group level of pain and individual course of pain. Obesity is a well-established risk factor for pain and OA, but there is a limited understanding of how obesity contributes to pain especially for OA-related pain. Several potential mechanisms of the
Table IV Metabolic syndrome, its components, and knee pain trajectories among participants with radiographic knee osteoarthritis status over 10.7 years (N ¼ 551) RRR (95%CI)* Mild vs Minimal pain Univariable Abdominal obesity Hypertension Fasting hyperglycaemia Hypertriglyceridemia Low HDL MetS
Model 1
Moderate vs Minimal pain Model 2
Model 3
Univariable
Model 1
Model 2
Model 3
1.72 (1.23, 2.41) 1.66 (1.15, 2.41) 1.16 (0.69, 1.93) 3.12 (1.97, 4.92) 2.87 (1.70, 4.84) 1.58 (0.76, 3.28) 0.83 (0.57, 1.21) 0.67 (0.44, 1.02) 0.60 (0.39, 0.93) 0.61 (0.40, 0.94) 0.73 (0.45, 1.18) 0.69 (0.40, 1.22) 0.57 (0.32, 1.03) 0.59 (0.33, 1.05) 1.25 (0.84, 1.85) 1.22 (0.79, 1.88) 1.02 (0.65, 1.59) 1.06 (0.68, 1.65) 0.88 (0.50, 1.54) 0.92 (0.49, 1.73) 0.64 (0.33, 1.23) 0.67 (0.25, 1.28) 1.17 (0.82, 1.68) 1.07 (0.72, 1.59) 1.44 (0.99, 2.11) 1.33 (0.88, 2.02) 1.51 (1.05, 2.17) 1.27 (0.86, 1.90)
0.94 (0.62, 1.41) 1.15 (0.75, 1.76) 0.94 (0.61, 1.45)
0.98 (0.65, 1.47) 1.23 (0.80, 1.90) 0.90 (0.56, 1.43)
1.68 (1.06, 2.65) 1.94 (1.16, 3.25) 1.57 (0.92, 2.66) 1.60 (0.94, 2.72) 2.18 (1.36, 3.50) 2.14 (1.27, 3.63) 1.67 (0.97, 2.88) 1.76 (1.02, 3.04) 2.27 (1.43, 3.59) 2.44 (1.45, 4.09) 1.59 (0.90, 2.82) 1.51 (0.82, 2.77)
Bold denotes statistically significant result; HDL, high density lipoprotein; MetS, Metabolic syndrome. RRR (95% CI): relative risk ratio (95% confidence interval) for belonging in each trajectory relative to reference trajectory (Minimal pain). Model 1: adjusted for age, sex, physical activity, smoking history, unemployment and education level. Model 2: Model 1 þ body mass index. Model 3: Model 1 þ abdominal obesity.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
6
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
relationship between obesity and pain have been proposed42. Mechanical-structural factors are the most common mechanism including increased joint loading and subsequent structural damage. Metabolic-inflammatory factors are being considered another important mechanism as adipose tissue is recognised as an endocrine organ where proinflammatory cytokines and adipokines are produced and released43. BMI is commonly used to classify obesity in prior studies examining obesity-pain relationship; central obesity measured by waist circumference, an important component of MetS, has been found to be associated with elevated levels of inflammation-mediating cytokines44. Furthermore, central fat has been suggested as a better marker of obesity risks in the elderly than BMI45. In the current study, among five components of MetS, central obesity was found to be associated with both the ‘Mild’ and ‘Moderate’ pain trajectories suggesting a critical role of central obesity in the development and maintenance of pain. However, this study cannot differentiate the relative contribution of mechanicalstructural and metabolic-inflammatory factors to pain trajectories. In a subsample of this study, we previously reported that inflammatory markers (C-reactive protein (CRP), tumour necrosis factor alpha (TNF-a) and interleukin-6 (IL-6)) were associated with worsening knee pain over 5 years independent of confounders including BMI46. We found that 7% of the association between abdominal obesity and the ‘Moderate pain’ trajectory may be explained by the inflammatory markers (Supplementary Table IV). Taken together, the findings from this study suggest that metabolicinflammatory mechanism may have a role in the relationship between MetS and worse pain trajectories. Decreased level of HDL and elevated level of triglycerides, although associated with either the ‘Mild pain’ or ‘Moderate pain’ trajectories in univariable analysis, remain statistically significant with the ‘Moderate pain’ trajectory after adjustment for potential confounders. The mechanism for this association is unknown, it is possible that chronic inflammation lowers the level of HDL and worse pain trajectory may be associated with a higher level of inflammation24. Another possible explanation is that elevated level of triglycerides results in structural damage including cartilage damage6 and bone marrow lesions47. However, we found that these associations disappeared after adjustment for BMI or central obesity indicating that these two components contributing to worse pain trajectory may be largely mediated by obesity. However, the magnitude of effect sizes and confidence intervals may suggest that this may be an issue of power, further studies with a larger sample size may reveal the contribution of dyslipidemia to the development of pain trajectories. Other than weight loss through exercise and/or diet recommended by current guidelines, this may open up an opportunity for knee pain and/or OA management if confirmed. For instance, it is worthwhile testing lipid-lowering medication (e.g., statins) in clinical trials and therapies aimed specifically at abdominal obesity. Surprisingly, there was an independent and inverse association between high blood pressure and the ‘Mild pain’ trajectory that differed from two recent studies where high blood pressure was reported to be positively associated with high intensity of pain18,19. It was hypothesised that the positive association involved subchondral bone ischemia due to a decreased blood flow in the subchondral mirco-vessels and reduced nutrients and oxygen supply between subchondral bone and overlying cartilage. However, there is emerging evidence that elevated resting blood pressure is associated with diminished pain sensitivity involving the functional interaction between the cardiovascular and pain regulatory systems48,49. This is a homeostatic feedback loop to help restore arousal levels in the presence of painful stimuli. In chronic pain conditions, the homeostatic feedback loop is significantly
altered, and thus affecting responsiveness to pain stimuli. The other possible explanation is ascribed to use of antihypertensive medication in particular b-adrenergic blockers which may have antinociceptive properties50. Supporting this, the protective effect disappeared after considering the use of antihypertensive medication. It also likely reflects possible misclassification of hypertension as people taking the medication may have been classified as having normal blood pressure on measurement. Given the inconsistent result, the relationship between hypertension and pain warrants further investigations. The community-dwelling individuals, relatively large sample size, long term follow-up and consideration of multiple potential confounders are strengths of this study. Our results can be, therefore, extrapolated to other elderly populations and perhaps those with ROA. However, there are several limitations with this study. First, the time interval between each pain assessment was large, therefore whether more frequent pain assessment may change pain trajectories cannot be determined from this study. However, pain patterns were similar with prior studies with more frequent assessments. Second, there is a strong correlation between BMI and waist circumference with an estimated R2 of 0.48 in this study. The corresponding variance inflation factor (VIF) was 1.94, suggesting the variance of waist circumference is 94% bigger than it would be if there was no correlation between BMI and waist circumference. However, the VIF is less than four which is widely considered an accepted level51. Third, pain medications and other techniques may have been using among participants with pain due to the nature of observational study. However, after adjusting for pain medication use, the results were similar (data not shown), suggesting that the results were robust. Fourth, blood tests (i.e., HDL, glucose and triglycerides) were not performed for all time-points, therefore, we are unable to define MetS and account for changes in MetS status after baseline. This may lead to misclassification; however, body weight and blood pressure did track strongly over time suggesting that our results may not have been affected by the exposure status change. Fifth, psychological factors (e.g., anxiety/depression) are linked to both MetS and pain with a bi-directional relationship between psychological factors and pain52 and between psychological factors and MetS53. It is hard to understand causal directions, so psychological factors were not considered confounders in this study. However, our results did not change after further adjustment for emotional problems (Supplementary Table V), suggesting that psychological factors are not biasing these results. In summary, MetS and its components are predominantly associated with worse pain trajectories through central obesity, suggesting that the development and maintenance of worse pain trajectories may be caused by MetS. Author contributions FP participated in the design of the study, analysis and interpretation of the data and manuscript preparation. JT participated in analysis and interpretation of the data and manuscript preparation. FC participated in the design of the study and interpretation of the data. GJ participated in the design of the study, interpretation of the data and manuscript preparation. All authors read and approved the final manuscript. Conflict of interest statement The authors declare no conflict of interest. Funding source This study was supported by grants from the National Health and Medical Research Council (NHMRC) of Australia (302204), the Tasmanian Community Fund (D0015018), the Arthritis Foundation of Australia (MRI06161), and the University of Tasmania Grant-
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
Institutional Research Scheme (D0015019). FP is funded by the NHMRC Early Career Fellowship; GJ is funded by the NHMRC Practitioner Fellowship. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.joca.2019.05.030. References 1. Disease GBD, Injury I, Prevalence C. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet 2017;390:1211e59. 2. Storheim K, Zwart JA. Musculoskeletal disorders and the global burden of disease study. Ann Rheum Dis 2014;73:949e50. 3. Breivik H, Collett B, Ventafridda V, Cohen R, Gallacher D. Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment. Eur J Pain 2006;10:287e333. 4. Urban H, Little CB. The role of fat and inflammation in the pathogenesis and management of osteoarthritis. Rheumatology (Oxford) 2018;57:iv10e21. 5. Courties A, Gualillo O, Berenbaum F, Sellam J. Metabolic stressinduced joint inflammation and osteoarthritis. Osteoarthr Cartil 2015;23:1955e65. 6. Sellam J, Berenbaum F. Is osteoarthritis a metabolic disease? Jt Bone Spine 2013;80:568e73. 7. Yoshimura N, Muraki S, Oka H, Kawaguchi H, Nakamura K, Akune T. Association of knee osteoarthritis with the accumulation of metabolic risk factors such as overweight, hypertension, dyslipidemia, and impaired glucose tolerance in Japanese men and women: the ROAD study. J Rheumatol 2011;38:921e30. 8. Yoshimura N, Muraki S, Oka H, Tanaka S, Kawaguchi H, Nakamura K, et al. Accumulation of metabolic risk factors such as overweight, hypertension, dyslipidaemia, and impaired glucose tolerance raises the risk of occurrence and progression of knee osteoarthritis: a 3-year follow-up of the ROAD study. Osteoarthr Cartil 2012;20:1217e26. 9. Han CD, Yang IH, Lee WS, Park YJ, Park KK. Correlation between metabolic syndrome and knee osteoarthritis: data from the Korean national health and nutrition examination survey (KNHANES). BMC Public Health 2013;13:603. 10. Shin D. Association between metabolic syndrome, radiographic knee osteoarthritis, and intensity of knee pain: results of a national survey. J Clin Endocrinol Metab 2014;99:3177e83. 11. Niu J, Clancy M, Aliabadi P, Vasan R, Felson DT. Metabolic syndrome, its components, and knee osteoarthritis: the Framingham osteoarthritis study. Arthritis Rheum 2017;69: 1194e203. 12. Afifi AEA, Shaat RM, Gharbia OM, Boghdadi YE, Eshmawy MME, El-Emam OA. Osteoarthritis of knee joint in metabolic syndrome. Clin Rheumatol 2018;37:2855e61. 13. Monira Hussain S, Wang Y, Cicuttini FM, Simpson JA, Giles GG, Graves S, et al. Incidence of total knee and hip replacement for osteoarthritis in relation to the metabolic syndrome and its components: a prospective cohort study. Semin Arthritis Rheum 2014;43:429e36. 14. Hellevik AI, Johnsen MB, Langhammer A, Baste V, Furnes O, Storheim K, et al. Metabolic syndrome as a risk factor for total hip or knee replacement due to primary osteoarthritis: a prospective cohort study (the HUNT study and the Norwegian Arthroplasty Register). Clin Epidemiol 2018;10:83e96.
7
15. Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull World Health Organ 2003;81:646e56. 16. Hawker GA, Croxford R, Bierman AS, Harvey PJ, Ravi B, Stanaitis I, et al. All-cause mortality and serious cardiovascular events in people with hip and knee osteoarthritis: a population based cohort study. PLoS One 2014;9, e91286. 17. Sowers M, Karvonen-Gutierrez CA, Palmieri-Smith R, Jacobson JA, Jiang Y, Ashton-Miller JA. Knee osteoarthritis in obese women with cardiometabolic clustering. Arthritis Rheum 2009;61:1328e36. 18. Abourazzak F, Talbi S, Lazrak F, Azzouzi H, Aradoini N, Keita S, et al. Does metabolic syndrome or its individual components affect pain and function in knee osteoarthritis women? Curr Rheumatol Rev 2015;11:8e14. 19. Li H, George DM, Jaarsma RL, Mao X. Metabolic syndrome and components exacerbate osteoarthritis symptoms of pain, depression and reduced knee function. Ann Transl Med 2016;4:133. 20. Sanchez-Santos MT, Judge A, Gulati M, Spector TD, Hart DJ, Newton JL, et al. Association of metabolic syndrome with knee and hand osteoarthritis: a community-based study of women. Semin Arthritis Rheum 2019;48:791e8. 21. Mantyselka P, Kautiainen H, Vanhala M. Prevalence of neck pain in subjects with metabolic syndrome–a cross-sectional population-based study. BMC Muscoskelet Disord 2010;11: 171. 22. Ray L, Lipton RB, Zimmerman ME, Katz MJ, Derby CA. Mechanisms of association between obesity and chronic pain in the elderly. Pain 2011;152:53e9. 23. Ono R, Yamazaki S, Takegami M, Otani K, Sekiguchi M, Onishi Y, et al. Gender difference in association between low back pain and metabolic syndrome: locomotive syndrome and health outcome in Aizu cohort study (LOHAS). Spine (Phila Pa 1976) 2012;37:1130e7. 24. Goodson NJ, Smith BH, Hocking LJ, McGilchrist MM, Dominiczak AF, Morris A, et al. Cardiovascular risk factors associated with the metabolic syndrome are more prevalent in people reporting chronic pain: results from a cross-sectional general population study. Pain 2013;154:1595e602. 25. Park IY, Cho NH, Lim SH, Kim HA. Gender-specific associations between fat mass, metabolic syndrome and musculoskeletal pain in community residents: a three-year longitudinal study. PLoS One 2018;13, e0200138. 26. Rechardt M, Shiri R, Lindholm H, Karppinen J, ViikariJuntura E. Associations of metabolic factors and adipokines with pain in incipient upper extremity soft tissue disorders: a cross-sectional study. BMJ Open 2013;3, e003036. 27. Collins JE, Katz JN, Dervan EE, Losina E. Trajectories and risk profiles of pain in persons with radiographic, symptomatic knee osteoarthritis: data from the osteoarthritis initiative. Osteoarthr Cartil 2014;22:622e30. 28. Nicholls E, Thomas E, van der Windt DA, Croft PR, Peat G. Pain trajectory groups in persons with, or at high risk of, knee osteoarthritis: findings from the Knee Clinical Assessment Study and the Osteoarthritis Initiative. Osteoarthr Cartil 2014;22:2041e50. 29. Wesseling J, Bastick AN, ten Wolde S, Kloppenburg M, Lafeber FP, Bierma-Zeinstra SM, et al. Identifying trajectories of pain severity in early symptomatic knee osteoarthritis: a 5year followup of the cohort hip and cohort knee (CHECK) study. J Rheumatol 2015;42:1470e7. 30. Dai Z, Lu N, Niu J, Felson DT, Zhang Y. Dietary fiber intake in relation to knee pain trajectory. Arthritis Care Res (Hoboken) 2017;69:1331e9.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
8
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
31. Pan F, Tian J, Aitken D, Cicuttini F, Jones G. Predictors of pain severity trajectory in older adults: a 10.7-year follow-up study. Osteoarthr Cartil 2018;26:1619e26. 32. Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol 2010;6:109e38. 33. Pan F, Jones G. Clinical perspective on pain and pain phenotypes in osteoarthritis. Curr Rheumatol Rep 2018;20:79. 34. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833e40. 35. Dore DA, Winzenberg TM, Ding C, Otahal P, Pelletier JP, MartelPelletier J, et al. The association between objectively measured physical activity and knee structural change using MRI. Ann Rheum Dis 2013;72:1170e5. 36. Altman RD, Hochberg M, Murphy Jr WA, Wolfe F, Lequesne M. Atlas of individual radiographic features in osteoarthritis. Osteoarthr Cartil 1995;3(Suppl A):3e70. 37. Karlson KB, Holm A. Decomposing primary and secondary effects: a new decomposition method. Res Soc Stratif Mobil 2011;29:221e37. 38. Kohler U, Karlson KB, Holm A. Comparing coefficients of nested nonlinear probability models. STATA J 2011;11:420e38. 39. Breen R, Karlson KB, Holm A. Total, direct, and indirect effects in logit and probit models. Sociol Methods Res 2013;42:164e91. 40. Breen R, Karlson KB, Holm A. Interpreting and understanding logits, probits, and other nonlinear probability models. Annu Rev Sociol 2018;44:39e54. 41. Bastick AN, Wesseling J, Damen J, Verkleij SP, Emans PJ, Bindels PJ, et al. Defining knee pain trajectories in early symptomatic knee osteoarthritis in primary care: 5-year results from a nationwide prospective cohort study (CHECK). Br J Gen Pract 2016;66:e32e9. 42. Narouze S, Souzdalnitski D. Obesity and chronic pain: systematic review of prevalence and implications for pain practice. Reg Anesth Pain Med 2015;40:91e111.
43. Ronti T, Lupattelli G, Mannarino E. The endocrine function of adipose tissue: an update. Clin Endocrinol 2006;64:355e65. 44. Schmidt FM, Weschenfelder J, Sander C, Minkwitz J, Thormann J, Chittka T, et al. Inflammatory cytokines in general and central obesity and modulating effects of physical activity. PLoS One 2015;10. e0121971. 45. Zamboni M, Mazzali G, Zoico E, Harris TB, Meigs JB, Di Francesco V, et al. Health consequences of obesity in the elderly: a review of four unresolved questions. Int J Obes (Lond) 2005;29:1011e29. 46. Stannus OP, Jones G, Blizzard L, Cicuttini FM, Ding C. Associations between serum levels of inflammatory markers and change in knee pain over 5 years in older adults: a prospective cohort study. Ann Rheum Dis 2013;72:535e40. 47. Davies-Tuck ML, Hanna F, Davis SR, Bell RJ, Davison SL, Wluka AE, et al. Total cholesterol and triglycerides are associated with the development of new bone marrow lesions in asymptomatic middle-aged women - a prospective cohort study. Arthritis Res Ther 2009;11:R181. 48. Bruehl S, Chung OY. Interactions between the cardiovascular and pain regulatory systems: an updated review of mechanisms and possible alterations in chronic pain. Neurosci Biobehav Rev 2004;28:395e414. 49. Bruehl S, Chung OY, Jirjis JN, Biridepalli S. Prevalence of clinical hypertension in patients with chronic pain compared to nonpain general medical patients. Clin J Pain 2005;21:147e53. 50. Valdes AM, Abhishek A, Muir K, Zhang W, Maciewicz RA, Doherty M. Association of beta-blocker use with less prevalent joint pain and lower opioid requirement in people with osteoarthritis. Arthritis Care Res (Hoboken) 2017;69:1076e81. 51. O’brien RM. A caution regarding rules of thumb for variance inflation factors. Qual Quantity 2007;41:673e90. 52. Edwards RR, Cahalan C, Calahan C, Mensing G, Smith M, Haythornthwaite JA. Pain, catastrophizing, and depression in the rheumatic diseases. Nat Rev Rheumatol 2011;7:216e24. 53. Repousi N, Masana MF, Sanchez-Niubo A, Haro JM, Tyrovolas S. Depression and metabolic syndrome in the older population: a review of evidence. J Affect Disord 2018;237:56e64.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
Metabolic syndrome and trajectory of knee pain in older adults F. Pan y *, J. Tian y, F. Cicuttini z, G. Jones y y Menzies Institute for Medical Research, University of Tasmania, Private Bag 23, Hobart, Tasmania 7000, Australia z Department of Epidemiology and Preventive Medicine, Monash University Medical School, Commercial Road, Melbourne 3181, Australia
a r t i c l e i n f o
s u m m a r y
Article history: Received 15 December 2018 Accepted 15 May 2019
Objectives: To examine the association of metabolic syndrome (MetS) and its components with knee pain severity trajectories. Methods: Data from a population-based cohort study were utilised. Baseline blood pressure, glucose, triglycerides and high-density lipoprotein (HDL) cholesterol were measured. MetS was defined according to the National Cholesterol Education Program-Adult Treatment Panel III criteria. Radiographic knee osteoarthritis (ROA) was assessed by X-ray. Pain severity was measured by the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain questionnaire at each time-point. Groupbased trajectory modelling was used to identify pain trajectories and multi-nominal logistic regression was used for analysis. Mediation analysis was performed to assess whether body mass index (BMI)/ central obesity mediated the association between MetS, its components and pain trajectories. Results: Among 985 participants (Mean ± SD age: 62.9 ± 7.4, 50% female), 32% had MetS and 60% had ROA. Three pain trajectories were identified: ‘Minimal pain’ (52%), ‘Mild pain’ (33%) and ‘Moderate pain’ (15%). After adjustment for potential confounders, central obesity increased risk of belonging to both ‘Mild pain’ and ‘Moderate pain’ trajectories as compared to the ‘Minimal pain’ trajectory group, but MetS [relative risk ratio (RRR): 2.26, 95%CI 1.50e3.39], hypertriglyceridemia (RRR: 1.75, 95%CI 1.16e2.62) and low HDL (RRR: 1.67, 95%CI 1.10e2.52) were only associated with ‘Moderate pain’ trajectory. BMI/central obesity explained 37e70% of these associations. Results were similar in those with ROA. Conclusion: MetS and its components are predominantly associated with worse pain trajectories through central obesity, suggesting that the development and maintenance of worse pain trajectories may be caused by MetS. © 2019 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Keywords: Metabolic syndrome Pain Trajectory Knee osteoarthritis
Introduction Musculoskeletal conditions are prevalent and are the most common causes of years lived with disability (YLD) worldwide, with musculoskeletal pain and osteoarthritis (OA) being the most frequent conditions1. It is estimated that the disability due to musculoskeletal conditions will continue to rise with an increasingly obese and ageing population2. The causes of musculoskeletal pain are complex and heterogeneous, but OA is the most common cause of pain especially in elderly3. Obesity is an established risk factor for OA and severity of symptoms4. It has been long proposed that the pathophysiological
* Address correspondence and reprint requests to: F. Pan, Menzies Institute for Medical Research, University of Tasmania, Private Bag 23, Hobart, Tasmania 7000, Australia. Tel: 61-3 6226-7700; Fax: 61-3 6226-7704. E-mail addresses: [email protected] (F. Pan), [email protected] (J. Tian), [email protected] (F. Cicuttini), [email protected] (G. Jones).
mechanisms underlying obesity-related OA are due to a mechanical overload, but metabolic and systemic inflammation related to adipose tissue have been considered additional mechanisms, leading to a new subtype of OA e metabolic OA5,6. Metabolic syndrome (MetS), generally consisting of increased waist circumference, dyslipidaemia (elevated triglycerides and reduced high-density lipoprotein (HDL)), hypertension and elevated serum glucose, has shown a relatively consistent association with the development and progression of OA7e12 and joint replacement13,14 before adjustment for body mass index (BMI) in previous studies, although the positive associations between MetS and its components and OA disappeared or weakened after controlling for BMI. Furthermore, most prior studies have been cross-sectional, limiting the ability to determine causality and raising the possibility that OA and MetS simply co-exist through shared risk factors such as age and obesity. Pain, the most prominent symptom of OA, is a leading cause of restriction in physical function, impaired mobility, and reduced
https://doi.org/10.1016/j.joca.2019.05.030 1063-4584/© 2019 Osteoarthritis Research Society International. Published by Elsevier Ltd. All rights reserved.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
2
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
quality of life in older adults15. It is also a major reason for patients to seek healthcare. The severity of OA-related pain and disability are associated with an increased risk of all-cause death16. This highlights that understanding of pain mechanisms is more clinically relevant especially when there is no proven curative treatment to slow or halt the progression of OA. In the musculoskeletal field, there are few reports investigating the relationship between MetS and OA-related pain10,12,17e20 and musculoskeletal pain21e26, most of which have been cross-sectional and the results have been inconsistent. Pain is often considered to get worse at the group level with time using traditional statistical approaches for longitudinal data analysis; however, pain at the individual level may be stable, improved or worsening. Prior studies including our own research have demonstrated that pain consists of homogeneous subgroups following distinct pain trajectories and pain level in each trajectory appears to be stable27e31. This suggests that pain is a nonprogressive symptom and maintains its level unless important factors are targeted. Group-based trajectory modelling (GBTM) can estimate and identify pain patterns and unobserved distinct subgroups of individuals with similar trajectories32. Given this, traditional analyses assuming that there is only one homogenous group may lead to misleading results. In this context, examining the association between MetS and pain trajectories in a longitudinal study may aid the understanding of underlying mechanisms of development and maintenance of pain, and thus allows for early interventions to shift pain trajectory to favour course33. The aim of study, therefore, was to examine the association between MetS and its components and pain trajectories in a general older population. Material and methods Participants Data utilized in this study were from the Tasmanian Older Adult Cohort Study (TASOAC), which is a longitudinal, population-based cohort study. Participants aged 50e80 years were randomly selected from the electoral roll in Southern Tasmania (population n ¼ 229,000) Australia in 2002. Of 2,530 subjects selected, a total of 1,099 participants were recruited and had their baseline examinations (including questionnaire, general interview, clinical assessment and blood tests). 875, 768 and 563 participants attended the second, third and fourth clinic during a mean follow-up period of 2.6, 5.1 and 10.7 years, respectively, where questionnaires, general interview and clinical assessment were completed, and blood samples were collected. The study was approved by the Southern Tasmanian Health and Medical Human Research Ethics Committee and written informed consent of all participants was obtained.
ilium in the mid-auxiliary plane. Blood pressure was measured using sphygmomanometer (Omron HEMe907) on the right arm with the correct sized cuff according to the arm circumference and at least 5 min has elapsed since participant was seated before recording the first blood pressure measurement. Measurement of systolic and diastolic blood pressure was then made two times with at least 30 s interval. The mean of two times measurements was used for the analysis. Venous fasting blood samples were taken from the antecubital vein and the concentrations of HDL, glucose and triglycerides were analysed on an Olympus AU5400 automated analyser (Olympus Optical, Tokyo, Japan). According to the National Cholesterol Education Program-Adult Treatment Panel III criteria, MetS is defined as the presence of 3 components out of the following: high waist circumference (102 cm in men and 88 cm in women), low HDL (<1.03 mmol/L in men and <1.29 mmol/L in women), hypertension (systolic blood pressure 130 mm Hg or diastolic blood pressure 85 mm Hg), hyperglycemia (fasting plasma glucose >5.6 mmol/L) and hypertriglyceridemia (triglycerides 1.7 mmol/L). Self-reported prescribed medication by a doctor was collected. Antihypertensive, antidiabetic and lipid-lowering medication were extracted. We also defined MetS by considering medication use. Participants receiving antihypertensive, antidiabetic and/or lipidlowering medication were classified as having hypertension, hyperglycemia, and/or low HDL and hypertriglyceridemia. Measurement of covariates Weight and height were measured, and BMI (kg/m2) was then calculated. Pedometer (Omron HJe003 & HJe102, Omron Healthcare, Kyoto, Japan) was used to measure physical activity (steps/day) for consecutive 7-day, as previously described35. Self-reported employment status was collected and grouped into two categories: ‘employed’ was defined as those with full/part-time job; home duties, student, sole parent/disability pension, retired or unemployed were considered ‘unemployment’. Highest education level the participants had completed was categorised as: ‘low level’ was those who completed school only, ‘medium level’ was those with a trade/vocational certificate, and ‘high level’ was those who had a university degree or above. Smoking history was assessed by the question, ‘Have you ever been a ‘regular smoker’? A ‘regular smoker’ is defined as someone who has smoked at least seven cigarettes, cigars or pipes each week for at least 3-month?’. Participants had a standing anteroposterior semiflexed radiograph of the right knee. The Altman atlas was used to score osteophytes and joint space narrowing (JSN) on a 0e3 point scale, as previously described36. Any score of JSN and/or osteophytes 1 was defined as radiographic knee OA (ROA). Statistical analysis
Measurement of outcome Knee pain The Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain questionnaire was used to assess knee pain. WOMAC pain questionnaire consists of five items including walking on flat surface, going up or down stairs, at night in the bed, sitting or lying and standing upright. Each item scores on a 10-point scale from 0 to 934. A total pain score is a sum of five items (0e45) with higher scores indicating greater pain. Measurement of factors comprising the MetS Waist circumference measurement (cm) was taken at the halfway between the bottom of the lowest rib and the top of the
Identifying pain trajectories Unobserved subgroups of individuals with similar pain trajectories over a-10.7-year course were identified using GBTM. GBTM utilises finite mixture modelling and multinomial modelling strategy to identify homogenous subgroups across a study population32, where trajectory parameters (i.e., shape and number of trajectory groups) are tested using maximum likelihood. We used censored normal models to estimate pain trajectories with considering possible trajectory shapes (linear, quadratic, or cubic) and number of trajectories (two to six). The Bayesian information criterion (BIC) and Aikaike's information criterion (AIC) was used to determine the best optimal model, with lower BIC and AIC values (absolute value) indicating a better model fit. Other criteria were also considered in the model selection process including: (1) model
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
3
interpretability:each trajectory group had >10% of the total sample; (2) pragmatic evaluation: posterior probabilities were >0.732. Posterior probabilities refer to the average probability of each participant that was assigned to in a given trajectory. Descriptive analysis Continuous and categorical variables were reported as Mean ± SD and percentages as appropriate.The characteristics of participants were compared across pain trajectory groups using ANOVA or c2 test where appropriate. Association between MetS and pain trajectory groups Age, sex, physical activity, smoking history, unemployment, education level and ROA were considered potential confounders based on existing literature within the field. Univariable and multivariable multi-nominal logistic regression were used to examine the associations between MetS and each of components and pain trajectory groups, where relative rate ratios (RRRs) and 95% confidence intervals (CIs) were produced. Mediation analysis was performed by using the KarlsonHolm-Breen decomposition method37,38 for multinomial outcome to determine whether obesity (BMI/central obesity) is the main mediating factor of the association between MetS and its components and pain trajectories. This method is not affected by the rescaling or attenuation issues that arise from crossmodel comparison of nonlinear models38e40. Mediation analysis was only performed for statistically significant associations between MetS and its components and pain trajectories in multivariable analyses. Secondary analyses were also conducted by restricting analyses to those with and without ROA. Sensitivity analyses were also performed on those with data at all four time-points. We also reanalysed the data using MetS definition based on use of medication. Stata V.15 was used for all analyses. P values 0.05 were regarded as statistically significant. Results There were 1,099 participants who attended the baseline examination including questionnaires, clinical assessment and blood tests. Trajectory model analysis was conducted for 985 participants who had complete baseline data. Of these, 802, 706 and 526 participants had complete pain data at the following three follow-ups, respectively. 60% of participants had ROA with a mean WOMAC pain score of 3.6 (SD ¼ 6.1) at baseline; 32% of participants had MetS. Participants who werelost to follow-up were older, had a greater BMI, fewer steps per day and a greater WOMAC pain score, and had a larger proportion of participants who were ever smoked, being unemployed, had low education level and had high waist circumference, hyperglycemia and MetS compared to those who completed the study (Supplementary Table I). Fig. 1 shows three distinct pain trajectory groups identified from the GBTM analysis. Pain severity within each of trajectories was not different but the pattern appeared similar with no trajectories showing a substantial pain improvement or worsening over time. There was almost half of participants (‘Minimal pain’ group) with a consistently low level of pain, approximately one third of participant (‘Mild pain’ group) with a mild level of pain consistent throughout the follow-up and 14% of participants (‘Moderate pain’) with a relatively high level of pain consistent throughout the follow-up. All groups included at least of 10% of participants with a clear difference in pain score suggesting that three-group model is a well-separated and optimal model. In addition, the average posterior probabilities
Fig. 1. Three distinct trajectories of WOMAC pain over 10.7 years.
were 0.97 (‘Minimal pain’), 0.95 (‘Mild pain’) and 0.97 (‘Moderate pain’), indicating a good model fit. Table I describes the characteristics of participants at baseline across three pain trajectory groups. ‘Moderate pain’ trajectory group appeared to have the highest BMI, had the least physical activity, had the greatest WOMAC pain score and the largest proportion of participants who were unemployed, had a low education level and had a ROA. Further, the ‘Moderate pain’ trajectory group was more likely to have the largest proportion of participants with central obesity, low HDL and MetS defined by the clinical criteria. BMI, waist circumference, WOMAC pain score, the proportion of participants with unemployed status, low education level, ROA, low HDL and MetS increased across ‘Minimal’, ‘Mild’ and ‘Moderate’ pain trajectory groups; alternatively, physical activity decreased across pain trajectory groups. The associations of MetS and its components with pain trajectories in the whole population regardless of ROA status are displayed in Table II. In univariable analysis, MetS, central obesity and low HDL were associated with both the ‘Mild pain’ and ‘Moderate pain’ trajectory groups; hypertriglyceridemia increased risk of being in the ‘Moderate pain’ trajectory relative to the ‘Minimal pain’ trajectory group. After adjustment for potential confounders including age, sex, physical activity, smoking history, unemployment, education level and ROA, the association of central obesity with both the ‘Mild pain’ and ‘Moderate pain’ trajectory groups remained statistically significant, whereas ‘Moderate pain’ trajectory group was only associated with MetS (RRR: 2.26, 95%CI 1.50e3.39], hypertriglyceridemia (RRR: 1.75, 95%CI 1.16e2.62) and low HDL (RRR: 1.67, 95%CI 1.10e2.52) as compared to the ‘Minimal pain’ trajectory group. The effect size of MetS and its components apart from hyperglycaemia tended to be higher with the ‘Moderate pain’ than that with ‘Mild pain’ trajectory group. Mediation analysis was performed to evaluate whether the association of MetS and its components with pain trajectories is mediated through BMI/central obesity. BMI was found to mediate the associations between central obesity, hypertriglyceridemia, low HDL and MetS and pain trajectories; the estimated mediation effect explained 37, 38, 38, 52% of each association, respectively (Table III). Central obesity also mediated 39e70% of the association between hypertriglyceridemia, low HDL and MetS and pain trajectories. Secondary analyses in those with ROA did not reveal additional pain trajectory groups. As compared to the ‘Minimal pain’ trajectory group, similar associations between MetS and its components and
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
4
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
Table I Characteristics of the three pain trajectory groups Characteristics
Minimal pain (N ¼ 512)
Mild pain (N ¼ 328)
Moderate pain (N ¼ 145)
Age, years Female (%) BMI (kg/m2) Physical activity (steps/day) Smoking history (%) Unemployment (%) Education level (%) School only Vocation training University or higher Radiographic knee OA (%) WOMAC pain score Waist circumference, cm Systolic BP, mm Hg Diastolic BP, mm Hg Fasting glucose, mmol/L Triglycerides, mmol/L HDL cholesterol, mmol/L High waist circumference* (%) Hypertensiony (%) Hyperglycemiaz (%) Hypertriglyceridemiax (%) Low HDLk (%) MetS¶ (%)
62.9 ± 7.4 48 26.8 ± 4.1 8910.1 ± 3363.4 52 57
63.0 ± 7.6 51 28.3 ± 4.6 8550.7 ± 3261.8 47 61
62.8 ± 7.2 57 30.1 ± 5.8 7839.7 ± 3523.7 52 72
56 30 14 57 0.3 ± 0.8 91.7 ± 12.5 140.0 ± 19.5 85.3 ± 10.1 5.3 ± 1.1 1.7 ± 4.0 1.4 ± 0.4 33 73 22 28 25 28
53 36 11 65 3.9 ± 3.2 95.0 ± 13.1 140.6 ± 21.4 84.9 ± 11.0 5.3 ± 1.1 1.5 ± 1.0 1.4 ± 0.4 47 70 24 30 31 35
64 31 4 64 14.7 ± 8.0 99.3 ± 14.3 139.6 ± 18.8 84.8 ± 10.0 5.3 ± 1.2 1.7 ± 0.9 1.3 ± 0.4 63 74 22 38 35 45
Bold denotes statistically significant result determined by ANOVA or Chi-square test as appropriate. BMI, body mass index; OA, osteoarthritis; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index; BP blood pressure; HDL, high density lipoprotein; MetS, Metabolic syndrome. * Defined as waist circumference 102/88 cm for men/women. y Defined as systolic or diastolic BP 130/85 mm Hg. z Defined as fasting plasma glucose 5.6 mmol/L. x Defined as triglycerides 1.7 mmol/L. k Defined as HDL<1.03/1.29 mmol/L for men/women. ¶ Defined as presence of three or more components above.
risk of being worse pain trajectory groups were observed in univariable and multivariable analysis with adjustment for potential confounders (Table IV). Further adjustment for BMI or central obesity showed a protective effect of hypertension with ‘Mild pain’ trajectory group, all other associations was attenuated and became statistically non-significant. Similarly, we found greater effect sizes of MetS and its components except for hyperglycaemia with the ‘Moderate pain’ than with ‘Mild pain’ trajectory. Sensitivity analyses performed on those with data at all four time-points found that MetS and central obesity (but not hypertriglyceridemia and low HDL) were associated with the ‘Moderate pain’ trajectory relative to the ‘Minimal pain’ trajectory group after adjustment for potential confounders.
We observed similar results based on a medication use definition (Supplementary Tables II and III); however, the protective effect of hypertension with the ‘Mild pain’ trajectory disappeared. Discussion Three knee pain trajectory groups were identified among community-dwelling elderly individuals and those with ROA over a course of 10.7-year in the current study. In multivariable analysis, central obesity was associated with an increased risk of being both the ‘Mild pain’ and ‘Moderate pain’ trajectory group relative to ‘Minimal pain’ trajectory group; whereas MetS and its components (hypertriglyceridemia and low HDL) only increased risk of being in
Table II Metabolic syndrome, its components, and knee pain trajectories among participants regardless of radiographic knee osteoarthritis status over 10.7 years (N ¼ 985) RRR (95%CI)* Mild vs Minimal pain Univariable Abdominal obesity Hypertension Fasting hyperglycaemia Hypertriglyceridemia Low HDL MetS
Model 1
Moderate vs Minimal pain Model 2
Model 3
Univariable
Model 1
Model 2
Model 3
1.69 (1.29, 2.22) 1.68 (1.25, 2.25) 1.30 (0.87, 1.94) 3.28 (2.28, 4.70) 3.19 (2.12, 4.80) 2.07 (1.18, 3.64) 0.89 (0.66, 1.19) 0.78 (0.56, 1.08) 0.71 (0.51, 0.99) 0.73 (0.52, 1.02) 0.99 (0.67, 1.47) 1.05 (0.67, 1.66) 0.90 (0.56, 1.43) 0.89 (0.56, 1.43) 1.12 (0.82, 1.54) 1.12 (0.79, 1.58) 0.95 (0.67, 1.35) 0.98 (0.69, 1.40) 0.97 (0.63, 1.49) 1.06 (0.66, 1.70) 0.74 (0.45, 1.22) 0.75 (0.46, 1.23) 1.18 (0.88, 1.57) 1.04 (0.76, 1.43) 1.43 (1.06, 1.92) 1.32 (0.96, 1.82) 1.47 (1.10, 1.96) 1.32 (0.96, 1.80)
0.93 (0.68, 1.29) 1.20 (0.87, 1.67) 1.02 (0.73, 1.44)
0.97 (0.70, 1.34) 1.79 (1.25, 2.57) 1.75 (1.16, 2.62) 1.44 (0.95, 2.18) 1.24 (0.89, 1.72) 1.86 (1.29, 2.70) 1.67 (1.10, 2.52) 1.41 (0.93, 2.16) 0.95 (0.66, 1.38) 2.22 (1.54, 3.20) 2.26 (1.50, 3.39) 1.49 (0.95, 2.33)
1.42 (0.93, 2.15) 1.37 (0.89, 2.10) 1.29 (0.81, 2.07)
Bold denotes statistically significant result; HDL, high density lipoprotein; MetS, Metabolic syndrome. *RRR (95%CI): relative risk ratio (95% confidence interval) for belonging in each trajectory relative to reference trajectory (Minimal pain). Model 1: adjusted for age, sex, physical activity, smoking history, unemployment, education level and radiographic knee osteoarthritis. Model 2: Model 1 þ body mass index. Model 3: Model 1 þ abdominal obesity.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
5
Table III Estimated mediation of the association between metabolic syndrome, its components, and knee pain trajectories by body mass index/central obesity in whole population (N ¼ 985) Moderate vs Minimal pain
BMI Total effect Direct effect Indirect effect Mediation percentage Central obesity Total effect Direct effect Indirect effect Mediation percentage
Abdominal obesity
Hypertriglyceridemia
Low HDL
MetS
3.20 (2.12, 4.83) 2.07 (1.18, 3.64) 1.54 (1.04, 2.29) 37
1.79 (1.18, 2.71) 1.44 (0.95, 2.18) 1.25 (1.12, 1.40) 38
1.74 (1.14, 2.66) 1.41 (0.93, 2.16) 1.23 (1.10, 1.38) 38
2.30 (1.52, 3.47) 1.49 (0.95, 2.33) 1.55 (1.26, 1.90) 52
1.77 (1.17, 2.67) 1.42 (0.93, 2.15) 1.25 (1.12, 1.39) 39
1.78 (1.17, 2.72) 1.37 (0.89, 2.10) 1.30 (1.15, 1.47) 46
2.32 (1.53, 3.49) 1.29 (0.81, 2.07) 1.79 (1.38, 2.34) 70
Table shows relative risk ratio and 95% confidence intervals; HDL, high density lipoprotein; MetS, Metabolic syndrome; BMI, body mass index. Indirect effect represents mediation of BMI/central obesity in the association between MetS, its components, and knee pain trajectories. Mediation percentage represents percentage of the total effect explained by the mediation of BMI/central obesity.
the ‘Moderate pain’ trajectory group. These associations were markedly attenuated and became statistically non-significant with the ‘Moderate pain’ trajectory group after further adjusting for BMI or central obesity. Further, mediation analysis shows that BMI/ central obesity is an important mediator of the association between MetS and hypertriglyceridemia and low HDL and pain trajectory. These findings suggest that the potential mechanistic links between MetS and its components and pain severity trajectory are mediated through BMI (largely by central obesity). To the best of our knowledge, this is the first study examining MetS and its components with pain trajectories in a general older population and among those with ROA. The findings that three distinct pain trajectories were identified in this study were consistent with prior studies which identified three to five stable pain severity trajectories, showing that the level of pain does not markedly change over time at a group level27,28,30,41. Combining prior studies with this study, our results suggest that pain appears to be a non-progressive symptom and on average, pain levels at baseline are maintained overtime unless important risk factors are modified. There was a slight difference in the number of pain trajectories between the current study and prior studies. This could be explained by differences in participants included, length of follow-up and pain measurement. However, clear differences in characteristics of participants and high posterior probabilities among three trajectories in the current study provide support for the construct and validity of identified trajectories. There is a growing body of literature existing on the role of MetS in relation to pathogenesis of OA. Few studies have reported its
relationship with OA-related pain10,12,17e20, all of which have been cross-sectional, and results are contradictory. This study goes beyond prior studies to investigate the role of MetS in the evolution of pain over time. Our finding that MetS was associated with an increased risk of belonging to worse pain trajectories before adjustment for BMI is in line with prior studies showing that pain intensity is higher in those with than without MetS. A more recent study with 60 knee OA patients by Afifi et al. reported that MetS was related to a higher level of WOMAC pain12. Li et al.19 reported that pain score was higher in those with greater number of metabolic disorders. In another study, Sowers et al.17 found that obese women with cardiometabolic clustering were more likely to report persistent knee pain compared with non-obese and obese without cardiometabolic clustering. The statistically significant association between MetS and worse pain trajectories was weakened and disappeared after controlling for BMI or central obesity. This is consistent with one recent study which found that MetS was associated with knee pain but this association disappeared after adjustment for BMI18. In contrast, Shin reported that accumulation of MetS components was associated with a higher intensity of OA pain even after considering BMI10. Although the discrepancy may be explained by differences in study design, characteristics of included participants, potential confounders considered and the use of questionnaires to assess pain, it likely reflects that MetS plays a different role in the group level of pain and individual course of pain. Obesity is a well-established risk factor for pain and OA, but there is a limited understanding of how obesity contributes to pain especially for OA-related pain. Several potential mechanisms of the
Table IV Metabolic syndrome, its components, and knee pain trajectories among participants with radiographic knee osteoarthritis status over 10.7 years (N ¼ 551) RRR (95%CI)* Mild vs Minimal pain Univariable Abdominal obesity Hypertension Fasting hyperglycaemia Hypertriglyceridemia Low HDL MetS
Model 1
Moderate vs Minimal pain Model 2
Model 3
Univariable
Model 1
Model 2
Model 3
1.72 (1.23, 2.41) 1.66 (1.15, 2.41) 1.16 (0.69, 1.93) 3.12 (1.97, 4.92) 2.87 (1.70, 4.84) 1.58 (0.76, 3.28) 0.83 (0.57, 1.21) 0.67 (0.44, 1.02) 0.60 (0.39, 0.93) 0.61 (0.40, 0.94) 0.73 (0.45, 1.18) 0.69 (0.40, 1.22) 0.57 (0.32, 1.03) 0.59 (0.33, 1.05) 1.25 (0.84, 1.85) 1.22 (0.79, 1.88) 1.02 (0.65, 1.59) 1.06 (0.68, 1.65) 0.88 (0.50, 1.54) 0.92 (0.49, 1.73) 0.64 (0.33, 1.23) 0.67 (0.25, 1.28) 1.17 (0.82, 1.68) 1.07 (0.72, 1.59) 1.44 (0.99, 2.11) 1.33 (0.88, 2.02) 1.51 (1.05, 2.17) 1.27 (0.86, 1.90)
0.94 (0.62, 1.41) 1.15 (0.75, 1.76) 0.94 (0.61, 1.45)
0.98 (0.65, 1.47) 1.23 (0.80, 1.90) 0.90 (0.56, 1.43)
1.68 (1.06, 2.65) 1.94 (1.16, 3.25) 1.57 (0.92, 2.66) 1.60 (0.94, 2.72) 2.18 (1.36, 3.50) 2.14 (1.27, 3.63) 1.67 (0.97, 2.88) 1.76 (1.02, 3.04) 2.27 (1.43, 3.59) 2.44 (1.45, 4.09) 1.59 (0.90, 2.82) 1.51 (0.82, 2.77)
Bold denotes statistically significant result; HDL, high density lipoprotein; MetS, Metabolic syndrome. RRR (95% CI): relative risk ratio (95% confidence interval) for belonging in each trajectory relative to reference trajectory (Minimal pain). Model 1: adjusted for age, sex, physical activity, smoking history, unemployment and education level. Model 2: Model 1 þ body mass index. Model 3: Model 1 þ abdominal obesity.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
6
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
relationship between obesity and pain have been proposed42. Mechanical-structural factors are the most common mechanism including increased joint loading and subsequent structural damage. Metabolic-inflammatory factors are being considered another important mechanism as adipose tissue is recognised as an endocrine organ where proinflammatory cytokines and adipokines are produced and released43. BMI is commonly used to classify obesity in prior studies examining obesity-pain relationship; central obesity measured by waist circumference, an important component of MetS, has been found to be associated with elevated levels of inflammation-mediating cytokines44. Furthermore, central fat has been suggested as a better marker of obesity risks in the elderly than BMI45. In the current study, among five components of MetS, central obesity was found to be associated with both the ‘Mild’ and ‘Moderate’ pain trajectories suggesting a critical role of central obesity in the development and maintenance of pain. However, this study cannot differentiate the relative contribution of mechanicalstructural and metabolic-inflammatory factors to pain trajectories. In a subsample of this study, we previously reported that inflammatory markers (C-reactive protein (CRP), tumour necrosis factor alpha (TNF-a) and interleukin-6 (IL-6)) were associated with worsening knee pain over 5 years independent of confounders including BMI46. We found that 7% of the association between abdominal obesity and the ‘Moderate pain’ trajectory may be explained by the inflammatory markers (Supplementary Table IV). Taken together, the findings from this study suggest that metabolicinflammatory mechanism may have a role in the relationship between MetS and worse pain trajectories. Decreased level of HDL and elevated level of triglycerides, although associated with either the ‘Mild pain’ or ‘Moderate pain’ trajectories in univariable analysis, remain statistically significant with the ‘Moderate pain’ trajectory after adjustment for potential confounders. The mechanism for this association is unknown, it is possible that chronic inflammation lowers the level of HDL and worse pain trajectory may be associated with a higher level of inflammation24. Another possible explanation is that elevated level of triglycerides results in structural damage including cartilage damage6 and bone marrow lesions47. However, we found that these associations disappeared after adjustment for BMI or central obesity indicating that these two components contributing to worse pain trajectory may be largely mediated by obesity. However, the magnitude of effect sizes and confidence intervals may suggest that this may be an issue of power, further studies with a larger sample size may reveal the contribution of dyslipidemia to the development of pain trajectories. Other than weight loss through exercise and/or diet recommended by current guidelines, this may open up an opportunity for knee pain and/or OA management if confirmed. For instance, it is worthwhile testing lipid-lowering medication (e.g., statins) in clinical trials and therapies aimed specifically at abdominal obesity. Surprisingly, there was an independent and inverse association between high blood pressure and the ‘Mild pain’ trajectory that differed from two recent studies where high blood pressure was reported to be positively associated with high intensity of pain18,19. It was hypothesised that the positive association involved subchondral bone ischemia due to a decreased blood flow in the subchondral mirco-vessels and reduced nutrients and oxygen supply between subchondral bone and overlying cartilage. However, there is emerging evidence that elevated resting blood pressure is associated with diminished pain sensitivity involving the functional interaction between the cardiovascular and pain regulatory systems48,49. This is a homeostatic feedback loop to help restore arousal levels in the presence of painful stimuli. In chronic pain conditions, the homeostatic feedback loop is significantly
altered, and thus affecting responsiveness to pain stimuli. The other possible explanation is ascribed to use of antihypertensive medication in particular b-adrenergic blockers which may have antinociceptive properties50. Supporting this, the protective effect disappeared after considering the use of antihypertensive medication. It also likely reflects possible misclassification of hypertension as people taking the medication may have been classified as having normal blood pressure on measurement. Given the inconsistent result, the relationship between hypertension and pain warrants further investigations. The community-dwelling individuals, relatively large sample size, long term follow-up and consideration of multiple potential confounders are strengths of this study. Our results can be, therefore, extrapolated to other elderly populations and perhaps those with ROA. However, there are several limitations with this study. First, the time interval between each pain assessment was large, therefore whether more frequent pain assessment may change pain trajectories cannot be determined from this study. However, pain patterns were similar with prior studies with more frequent assessments. Second, there is a strong correlation between BMI and waist circumference with an estimated R2 of 0.48 in this study. The corresponding variance inflation factor (VIF) was 1.94, suggesting the variance of waist circumference is 94% bigger than it would be if there was no correlation between BMI and waist circumference. However, the VIF is less than four which is widely considered an accepted level51. Third, pain medications and other techniques may have been using among participants with pain due to the nature of observational study. However, after adjusting for pain medication use, the results were similar (data not shown), suggesting that the results were robust. Fourth, blood tests (i.e., HDL, glucose and triglycerides) were not performed for all time-points, therefore, we are unable to define MetS and account for changes in MetS status after baseline. This may lead to misclassification; however, body weight and blood pressure did track strongly over time suggesting that our results may not have been affected by the exposure status change. Fifth, psychological factors (e.g., anxiety/depression) are linked to both MetS and pain with a bi-directional relationship between psychological factors and pain52 and between psychological factors and MetS53. It is hard to understand causal directions, so psychological factors were not considered confounders in this study. However, our results did not change after further adjustment for emotional problems (Supplementary Table V), suggesting that psychological factors are not biasing these results. In summary, MetS and its components are predominantly associated with worse pain trajectories through central obesity, suggesting that the development and maintenance of worse pain trajectories may be caused by MetS. Author contributions FP participated in the design of the study, analysis and interpretation of the data and manuscript preparation. JT participated in analysis and interpretation of the data and manuscript preparation. FC participated in the design of the study and interpretation of the data. GJ participated in the design of the study, interpretation of the data and manuscript preparation. All authors read and approved the final manuscript. Conflict of interest statement The authors declare no conflict of interest. Funding source This study was supported by grants from the National Health and Medical Research Council (NHMRC) of Australia (302204), the Tasmanian Community Fund (D0015018), the Arthritis Foundation of Australia (MRI06161), and the University of Tasmania Grant-
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
Institutional Research Scheme (D0015019). FP is funded by the NHMRC Early Career Fellowship; GJ is funded by the NHMRC Practitioner Fellowship. Supplementary data Supplementary data to this article can be found online at https://doi.org/10.1016/j.joca.2019.05.030. References 1. Disease GBD, Injury I, Prevalence C. Global, regional, and national incidence, prevalence, and years lived with disability for 328 diseases and injuries for 195 countries, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet 2017;390:1211e59. 2. Storheim K, Zwart JA. Musculoskeletal disorders and the global burden of disease study. Ann Rheum Dis 2014;73:949e50. 3. Breivik H, Collett B, Ventafridda V, Cohen R, Gallacher D. Survey of chronic pain in Europe: prevalence, impact on daily life, and treatment. Eur J Pain 2006;10:287e333. 4. Urban H, Little CB. The role of fat and inflammation in the pathogenesis and management of osteoarthritis. Rheumatology (Oxford) 2018;57:iv10e21. 5. Courties A, Gualillo O, Berenbaum F, Sellam J. Metabolic stressinduced joint inflammation and osteoarthritis. Osteoarthr Cartil 2015;23:1955e65. 6. Sellam J, Berenbaum F. Is osteoarthritis a metabolic disease? Jt Bone Spine 2013;80:568e73. 7. Yoshimura N, Muraki S, Oka H, Kawaguchi H, Nakamura K, Akune T. Association of knee osteoarthritis with the accumulation of metabolic risk factors such as overweight, hypertension, dyslipidemia, and impaired glucose tolerance in Japanese men and women: the ROAD study. J Rheumatol 2011;38:921e30. 8. Yoshimura N, Muraki S, Oka H, Tanaka S, Kawaguchi H, Nakamura K, et al. Accumulation of metabolic risk factors such as overweight, hypertension, dyslipidaemia, and impaired glucose tolerance raises the risk of occurrence and progression of knee osteoarthritis: a 3-year follow-up of the ROAD study. Osteoarthr Cartil 2012;20:1217e26. 9. Han CD, Yang IH, Lee WS, Park YJ, Park KK. Correlation between metabolic syndrome and knee osteoarthritis: data from the Korean national health and nutrition examination survey (KNHANES). BMC Public Health 2013;13:603. 10. Shin D. Association between metabolic syndrome, radiographic knee osteoarthritis, and intensity of knee pain: results of a national survey. J Clin Endocrinol Metab 2014;99:3177e83. 11. Niu J, Clancy M, Aliabadi P, Vasan R, Felson DT. Metabolic syndrome, its components, and knee osteoarthritis: the Framingham osteoarthritis study. Arthritis Rheum 2017;69: 1194e203. 12. Afifi AEA, Shaat RM, Gharbia OM, Boghdadi YE, Eshmawy MME, El-Emam OA. Osteoarthritis of knee joint in metabolic syndrome. Clin Rheumatol 2018;37:2855e61. 13. Monira Hussain S, Wang Y, Cicuttini FM, Simpson JA, Giles GG, Graves S, et al. Incidence of total knee and hip replacement for osteoarthritis in relation to the metabolic syndrome and its components: a prospective cohort study. Semin Arthritis Rheum 2014;43:429e36. 14. Hellevik AI, Johnsen MB, Langhammer A, Baste V, Furnes O, Storheim K, et al. Metabolic syndrome as a risk factor for total hip or knee replacement due to primary osteoarthritis: a prospective cohort study (the HUNT study and the Norwegian Arthroplasty Register). Clin Epidemiol 2018;10:83e96.
7
15. Woolf AD, Pfleger B. Burden of major musculoskeletal conditions. Bull World Health Organ 2003;81:646e56. 16. Hawker GA, Croxford R, Bierman AS, Harvey PJ, Ravi B, Stanaitis I, et al. All-cause mortality and serious cardiovascular events in people with hip and knee osteoarthritis: a population based cohort study. PLoS One 2014;9, e91286. 17. Sowers M, Karvonen-Gutierrez CA, Palmieri-Smith R, Jacobson JA, Jiang Y, Ashton-Miller JA. Knee osteoarthritis in obese women with cardiometabolic clustering. Arthritis Rheum 2009;61:1328e36. 18. Abourazzak F, Talbi S, Lazrak F, Azzouzi H, Aradoini N, Keita S, et al. Does metabolic syndrome or its individual components affect pain and function in knee osteoarthritis women? Curr Rheumatol Rev 2015;11:8e14. 19. Li H, George DM, Jaarsma RL, Mao X. Metabolic syndrome and components exacerbate osteoarthritis symptoms of pain, depression and reduced knee function. Ann Transl Med 2016;4:133. 20. Sanchez-Santos MT, Judge A, Gulati M, Spector TD, Hart DJ, Newton JL, et al. Association of metabolic syndrome with knee and hand osteoarthritis: a community-based study of women. Semin Arthritis Rheum 2019;48:791e8. 21. Mantyselka P, Kautiainen H, Vanhala M. Prevalence of neck pain in subjects with metabolic syndrome–a cross-sectional population-based study. BMC Muscoskelet Disord 2010;11: 171. 22. Ray L, Lipton RB, Zimmerman ME, Katz MJ, Derby CA. Mechanisms of association between obesity and chronic pain in the elderly. Pain 2011;152:53e9. 23. Ono R, Yamazaki S, Takegami M, Otani K, Sekiguchi M, Onishi Y, et al. Gender difference in association between low back pain and metabolic syndrome: locomotive syndrome and health outcome in Aizu cohort study (LOHAS). Spine (Phila Pa 1976) 2012;37:1130e7. 24. Goodson NJ, Smith BH, Hocking LJ, McGilchrist MM, Dominiczak AF, Morris A, et al. Cardiovascular risk factors associated with the metabolic syndrome are more prevalent in people reporting chronic pain: results from a cross-sectional general population study. Pain 2013;154:1595e602. 25. Park IY, Cho NH, Lim SH, Kim HA. Gender-specific associations between fat mass, metabolic syndrome and musculoskeletal pain in community residents: a three-year longitudinal study. PLoS One 2018;13, e0200138. 26. Rechardt M, Shiri R, Lindholm H, Karppinen J, ViikariJuntura E. Associations of metabolic factors and adipokines with pain in incipient upper extremity soft tissue disorders: a cross-sectional study. BMJ Open 2013;3, e003036. 27. Collins JE, Katz JN, Dervan EE, Losina E. Trajectories and risk profiles of pain in persons with radiographic, symptomatic knee osteoarthritis: data from the osteoarthritis initiative. Osteoarthr Cartil 2014;22:622e30. 28. Nicholls E, Thomas E, van der Windt DA, Croft PR, Peat G. Pain trajectory groups in persons with, or at high risk of, knee osteoarthritis: findings from the Knee Clinical Assessment Study and the Osteoarthritis Initiative. Osteoarthr Cartil 2014;22:2041e50. 29. Wesseling J, Bastick AN, ten Wolde S, Kloppenburg M, Lafeber FP, Bierma-Zeinstra SM, et al. Identifying trajectories of pain severity in early symptomatic knee osteoarthritis: a 5year followup of the cohort hip and cohort knee (CHECK) study. J Rheumatol 2015;42:1470e7. 30. Dai Z, Lu N, Niu J, Felson DT, Zhang Y. Dietary fiber intake in relation to knee pain trajectory. Arthritis Care Res (Hoboken) 2017;69:1331e9.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030
8
F. Pan et al. / Osteoarthritis and Cartilage xxx (xxxx) xxx
31. Pan F, Tian J, Aitken D, Cicuttini F, Jones G. Predictors of pain severity trajectory in older adults: a 10.7-year follow-up study. Osteoarthr Cartil 2018;26:1619e26. 32. Nagin DS, Odgers CL. Group-based trajectory modeling in clinical research. Annu Rev Clin Psychol 2010;6:109e38. 33. Pan F, Jones G. Clinical perspective on pain and pain phenotypes in osteoarthritis. Curr Rheumatol Rep 2018;20:79. 34. Bellamy N, Buchanan WW, Goldsmith CH, Campbell J, Stitt LW. Validation study of WOMAC: a health status instrument for measuring clinically important patient relevant outcomes to antirheumatic drug therapy in patients with osteoarthritis of the hip or knee. J Rheumatol 1988;15:1833e40. 35. Dore DA, Winzenberg TM, Ding C, Otahal P, Pelletier JP, MartelPelletier J, et al. The association between objectively measured physical activity and knee structural change using MRI. Ann Rheum Dis 2013;72:1170e5. 36. Altman RD, Hochberg M, Murphy Jr WA, Wolfe F, Lequesne M. Atlas of individual radiographic features in osteoarthritis. Osteoarthr Cartil 1995;3(Suppl A):3e70. 37. Karlson KB, Holm A. Decomposing primary and secondary effects: a new decomposition method. Res Soc Stratif Mobil 2011;29:221e37. 38. Kohler U, Karlson KB, Holm A. Comparing coefficients of nested nonlinear probability models. STATA J 2011;11:420e38. 39. Breen R, Karlson KB, Holm A. Total, direct, and indirect effects in logit and probit models. Sociol Methods Res 2013;42:164e91. 40. Breen R, Karlson KB, Holm A. Interpreting and understanding logits, probits, and other nonlinear probability models. Annu Rev Sociol 2018;44:39e54. 41. Bastick AN, Wesseling J, Damen J, Verkleij SP, Emans PJ, Bindels PJ, et al. Defining knee pain trajectories in early symptomatic knee osteoarthritis in primary care: 5-year results from a nationwide prospective cohort study (CHECK). Br J Gen Pract 2016;66:e32e9. 42. Narouze S, Souzdalnitski D. Obesity and chronic pain: systematic review of prevalence and implications for pain practice. Reg Anesth Pain Med 2015;40:91e111.
43. Ronti T, Lupattelli G, Mannarino E. The endocrine function of adipose tissue: an update. Clin Endocrinol 2006;64:355e65. 44. Schmidt FM, Weschenfelder J, Sander C, Minkwitz J, Thormann J, Chittka T, et al. Inflammatory cytokines in general and central obesity and modulating effects of physical activity. PLoS One 2015;10. e0121971. 45. Zamboni M, Mazzali G, Zoico E, Harris TB, Meigs JB, Di Francesco V, et al. Health consequences of obesity in the elderly: a review of four unresolved questions. Int J Obes (Lond) 2005;29:1011e29. 46. Stannus OP, Jones G, Blizzard L, Cicuttini FM, Ding C. Associations between serum levels of inflammatory markers and change in knee pain over 5 years in older adults: a prospective cohort study. Ann Rheum Dis 2013;72:535e40. 47. Davies-Tuck ML, Hanna F, Davis SR, Bell RJ, Davison SL, Wluka AE, et al. Total cholesterol and triglycerides are associated with the development of new bone marrow lesions in asymptomatic middle-aged women - a prospective cohort study. Arthritis Res Ther 2009;11:R181. 48. Bruehl S, Chung OY. Interactions between the cardiovascular and pain regulatory systems: an updated review of mechanisms and possible alterations in chronic pain. Neurosci Biobehav Rev 2004;28:395e414. 49. Bruehl S, Chung OY, Jirjis JN, Biridepalli S. Prevalence of clinical hypertension in patients with chronic pain compared to nonpain general medical patients. Clin J Pain 2005;21:147e53. 50. Valdes AM, Abhishek A, Muir K, Zhang W, Maciewicz RA, Doherty M. Association of beta-blocker use with less prevalent joint pain and lower opioid requirement in people with osteoarthritis. Arthritis Care Res (Hoboken) 2017;69:1076e81. 51. O’brien RM. A caution regarding rules of thumb for variance inflation factors. Qual Quantity 2007;41:673e90. 52. Edwards RR, Cahalan C, Calahan C, Mensing G, Smith M, Haythornthwaite JA. Pain, catastrophizing, and depression in the rheumatic diseases. Nat Rev Rheumatol 2011;7:216e24. 53. Repousi N, Masana MF, Sanchez-Niubo A, Haro JM, Tyrovolas S. Depression and metabolic syndrome in the older population: a review of evidence. J Affect Disord 2018;237:56e64.
Please cite this article as: Pan F et al., Metabolic syndrome and trajectory of knee pain in older adults, Osteoarthritis and Cartilage, https:// doi.org/10.1016/j.joca.2019.05.030