Article Text

Extended report
The association between objectively measured physical activity and knee structural change using MRI
  1. Dawn A Doré1,
  2. Tania Maree Winzenberg1,
  3. Changhai Ding1,2,
  4. Petr Otahal1,
  5. Jean-Pierre Pelletier3,
  6. Johanne Martel-Pelletier3,
  7. Flavia M Cicuttini2,
  8. Graeme Jones1
  1. 1Menzies Research Institute Tasmania, University of Tasmania, Hobart, Tasmania, Australia
  2. 2Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Victoria, Australia
  3. 3Osteoarthritis Research Unit, University of Montreal Hospital Research Centre (CRCHUM), Notre-Dame Hospital, Montreal, Quebec, Canada
  1. Correspondence to Dr Dawn A Doré, Menzies Research Institute Tasmania, University of Tasmania, Private Bag 23, Hobart, Tasmania 7000, Australia; Dawn.Dore{at}utas.edu.au

Abstract

Objectives This study describes the longitudinal association between objectively assessed physical activity (PA) and knee structural change measured using MRI.

Methods 405 community-dwelling adults aged 51–81 years were measured at baseline and approximately 2.7 years later. MRI of the right knee at baseline and follow-up was performed to evaluate bone marrow lesions (BMLs), meniscal pathology, cartilage defects, and cartilage volume. PA was assessed at baseline by pedometer (steps/day).

Results Doing ≥10 000 steps/day was associated with BML increases (RR 1.97, 95% CI 1.19 to 3.27, p=0.009). Participants doing ≥10 000 steps/day had a 1.52 times (95% CI 1.05 to 2.20, p=0.027) greater risk of increasing meniscal pathology score, which increased to 2.49 (95% CI 1.05 to 3.93, p=0.002) in those with adverse meniscal pathology at baseline. Doing ≥10 000 steps/day was associated with a greater risk of increasing cartilage defect score in those with prevalent BMLs at baseline (RR 1.36, 95% CI 1.03 to 1.69, p=0.013). Steps/day was protective against volume loss in those with more baseline cartilage volume but led to increased cartilage loss in those with less baseline cartilage volume. (p=0.046 for interaction).

Conclusions PA was deleteriously associated with knee structural change, especially in those with pre-existing knee structural abnormalities. This suggests individuals with knee abnormalities should avoid doing ≥10 000 steps/day. Alternatives to weight-bearing activity may be needed in order to maintain PA levels required for other aspects of health.

  • Osteoarthritis
  • Epidemiology
  • Knee Osteoarthritis
  • Magnetic Resonance Imaging
View Full Text

Statistics from Altmetric.com

Introduction

While there is concern that physical activity (PA) may increase the risk of osteoarthritis (OA) of the knee, epidemiologic studies examining the relationship between PA and knee OA are conflicting. Some studies suggest PA has a detrimental effect,1–3 while others show either no4–6 or a beneficial effect.7 ,8

A number of factors may contribute to this inconsistency. First, many studies used radiographs to assess OA, which provides only a limited view of the disease process.9 ,10 Second, many studies have used self-report surveys to assess PA11–17 which tend to over-report activity, demonstrate only moderate reproducibility, and have a modest correlation with actual PA.18–20 Third, many studies are cross-sectional. Lastly, injury is one of the strongest risk factors for OA development,21 and may confound any relationship between PA and OA.

MRI can directly visualise joint structures including bone, cartilage, menisci, synovium, and ligaments. Knee structural change can be measured both reliably and with good responsiveness on MRI.22 Therefore, using MRI may provide a better understanding of the effect PA has on knee structures. Although there have been a number of cross-sectional studies, there are very few longitudinal studies.23

The aim of this study was to examine the longitudinal association between objectively assessed PA (steps/day by pedometer) and knee structural change measured using MRI.

Methods

Subjects

This study was conducted as part of the Tasmanian Older Adult Cohort study. Subjects between the ages of 50 and 80 years were randomly selected from the electoral roll in Southern Tasmania (population 229 000), with an equal number of men and women (response rate 57%, 1099/1904). Exclusion criteria included contraindication for MRI and institutionalisation. Follow-up data was collected for 875 participants approximately 2.7 years later. The MRI machine was decommissioned halfway through the follow-up period; therefore, follow-up MRI scans were only available for approximately half of the participants (n=425/875). This research was conducted in compliance with the Declaration of Helsinki and was approved by the Southern Tasmanian Health and Medical Human Research Ethics Committee. All subjects gave informed written consent.

Anthropometrics

Weight was measured to the nearest 0.1 kg (Seca Delta Model 707). Height was measured to the nearest 0.1 cm using a stadiometer. Body mass index (BMI) was calculated (kg/m2).

Physical activity

PA was assessed at baseline as steps/day determined by pedometer (Omron HJ–003 & HJ-102, Omron Healthcare, Kyoto, Japan). Each participant was instructed to wear a pedometer for seven consecutive days. This was repeated 6 months later to account for seasonal variation. Mean steps/day was calculated as the average of the days worn at both time points. See supplementary text 1 for more information.

MRI

Images of the right knee at baseline and follow-up were acquired with a 1.5T whole-body magnetic resonance unit (Picker, Cleveland, Ohio, USA). Sagittal image sequences included: (1) a T1-weighted fat saturation three–dimensional (3D) gradient-recalled acquisition in the steady state, flip angle 30°, repetition time 31 ms, echo time 6.71 ms, field of view 16 cm, 60 partitions, 512×512–pixel matrix, slice thickness of 1.5 mm without a interslice gap; and (2) a T2-weighted fat saturation 2D fast spin echo, flip angle 90°, repetition time 3067 ms, echo time 112 ms, field of view 16 cm, 15 partitions, 228×256–pixel matrix, slice thickness of 4 mm with a interslice gap of 0.5–1.0 mm.

Bone marrow lesions

Bone marrow lesions (BMLs) were assessed on T2-weighted MR images and defined as areas of increased signal adjacent to the subcortical bone at the medial tibial, medial femoral, lateral tibial, and lateral femoral sites. One trained observer scored BMLs by measuring the maximum area of the lesion (mm2) at baseline and follow-up, as previously described.24 The intraclass correlation coefficient (ICC) was 0.97 for intra-observer repeatability. BML score at all four sites was summed to create a total BML score. Change in BML size was calculated as: follow-up total BML score—baseline total BML score.

Meniscal damage

Meniscal damage was assessed by a trained observer on T1-weighted MR images as previously described.25 The proportion of the menisci affected by a tear, partial or full extrusion was scored separately (yes/no) at the anterior, middle, and posterior horns (medially/laterally). These scores were summed to create a total meniscal pathology score which had a possible range from 0–18 (0–6 for tears, 0–6 for partial extrusions, and 0–6 for full extrusion). A meniscal pathology score increase was defined as an increase in tear, partial or full extrusion score.

Cartilage defects

Cartilage defects were assessed by a trained observer on T1-weighted MR images at the medial tibial, medial femoral, lateral tibial, and lateral femoral sites, as previously described26 as follows: grade 0=normal cartilage; grade 1=focal blistering and intracartilaginous low-signal intensity area with an intact surface and base; grade 2=irregularities on the surface or base and loss of thickness <50%; grade 3=deep ulceration with loss of thickness >50%; and grade 4=full-thickness chondral wear with exposure of subchondral bone. ICCs ranged from 0.80–0.95 for intra-observer repeatability. A cartilage defect score increase was defined as an increase of one or more on the 0–4 scale at any site.

Cartilage volume

Tibial cartilage volume was assessed by a trained observer on T1-weighted MR images using Osiris (University of Geneva, Geneva, Switzerland) software as previously described.27 ,28 The coefficient of variation (CV) ranged from 2.1–2.2% for intra-observer repeatability.27 Femoral cartilage volume was determined using Cartiscope (ArthroLab, Montreal, Quebec, Canada), as previously described.25 ,29 ,30 The CV was approximately 2% for intra-observer and inter-scan repeatability.29 Total cartilage volume was calculated as: tibial+femoral cartilage volume. Change in cartilage volume was calculated as: follow-up total cartilage volume—baseline total cartilage volume.

Tibial bone area

Tibial plateau bone area was assessed on T1-weighted MR images and defined as the cross-sectional surface area of the tibial plateau, as previously described.10 ,31 ,32 The CV was 2.2–2.6% for intra-observer repeatability.10 Change in tibial bone area was calculated as: follow-up bone area—baseline bone area.

Knee injury and surgery

At baseline, participants were asked, ‘Have you had knee surgery?’, and at follow-up, ‘Have you had any knee surgery since your first interview for this study?’. Knee injury was assessed at follow-up by asking ‘Have you had a previous knee injury requiring non-weight-bearing treatment for more than 24 h or surgery?’.

x-Ray

A baseline standing anteroposterior semiflexed view of the right knee with 15° of fixed knee flexion was performed and scored individually for osteophytes and joint space narrowing (JSN) on a scale of 0–3.33 The presence of radiographic OA (ROA) was defined as any score ≥1 for JSN or osteophytes.

Data analysis

T-tests and χ2 tests were used to compare differences in means and proportions.

The exposure for all regression analyses was steps/day. Steps/day was analysed in two ways; as a continuous variable, and also dichotomised according to the well-recognised recommendation calling for healthy adults to achieve 10 000 steps/day34 as an priori analysis. Five outcomes of knee structural change were analysed; change in BML area, increase in meniscal pathology score, increase in cartilage defect score, change in cartilage volume, and change in tibial bone area.

Log-binomial regression analyses were used to examine the associations between steps/day with increase in cartilage defect score and increase in meniscal pathology score.

Linear regression was used to examine changes in cartilage volume and tibial bone area with steps/day.

Absolute change in BML area was not normally distributed and no sensible transformation could validly normalise model residuals; therefore we analysed this variable two ways. Firstly, in the whole sample, change in BML area was dichotomised around 25 mm2 (based on previous work indicating that only an increase larger than this represents a genuine change after considering observer variability in scoring BMLs24 ,35) so that the outcome represented absence/presence of BML area increase of >25 mm2 (a deleterious change). Log-binomial regression was used to examine the association with steps/day.

Secondly we examined only persons showing a deleterious change in BML area (>25 mm2). A logarithmic transformation of change in BML area for these persons appropriately normalised model residuals, and thus linear regression was used to analyse these data.

Covariates common to all models included age, sex, BMI, ROA, and history of knee injury or surgery. Depending on the structural outcome, models were further adjusted for the baseline variable of the outcome of interest.

Interactions between steps/day and each of the five baseline knee structure variables, and steps/day and BMI were tested in all multivariable models, a total of six tests for each outcome (five outcomes). See supplementary text 2 for further information.

Standard diagnostic checks of model adequacy and unusual observations were performed on all models.

A p value less than 0.05 (two-tailed) was considered statistically significant. All statistical analyses were performed on Intercooled Stata V.12.0 for windows (StataCorp LP).

Results

Subjects

Four hundred and five participants with MRI measures at baseline and follow-up and pedometer measures at baseline were included. The average time to follow-up was 2.7 years (SD 0.4, range 2.0–4.7). There were small differences in baseline steps/day and meniscal pathology score between the subjects in the current study (mean±SD steps/day 8896±3345; mean±SD meniscal pathology score 5.6±1.3) and the rest of the cohort (mean±SD steps/day 8443±3352, p=0.03; mean±SD meniscal pathology score 5.3±1.5, p<0.01) but no other significant differences in study factors.

Characteristics of participants are presented in table 1 split by 10 000 steps/day. Participants who did ≥10 000 steps/day were younger, had lower BMI and a higher proportion had a BML and meniscal pathology score increase at follow-up.

Table 1

Study characteristics split by 10000 steps/day*

PA and knee structural changes

BML area

Whole sample

In multivariable analysis, the risk of having a deleterious BML change (increase >25 mm2) was 1.97 times (95% CI 1.19 to 3.27) greater for participants doing ≥10 000 steps/day, compared with participants doing <10 000 steps/day (p=0.009). For every 1000 step/day increase in PA, there was a 1.10 times greater risk of having a deleterious change in BML area (p< 0.001).

Deleterious change in BML

In participants with a deleterious BML change (n=64), those doing ≥10 000 steps/day had a 48.96 mm2 (95% CI 8.64 to 89.29 mm2) greater increase in BML area compared with those doing <10 000 steps/day (figure 1).

Figure 1

Bone marrow lesion (BML) change for participants with a deleterious increase (>25 mm2) in BML area against steps/day, after adjustment for age, sex, body mass index, radiographic osteoarthritis, history of knee injury or surgery, and baseline BML.

Meniscal pathology

In multivariable analysis, participants doing ≥10 000 steps/day had a 1.52 times (95% CI 1.05 to 2.20) greater risk of increasing meniscal pathology score compared to those doing <10 000 (p=0.027). The effect was modified by baseline meniscal pathology score (figure 2A). Participants with higher baseline meniscal pathology scores (5–10) had a 2.49 times (95% CI 1.05 to 3.93) higher risk (p=0.002) of increase in meniscal pathology score if they did ≥10 000 steps/day compared to those doing <10 000. Due to a likely ceiling effect in those with higher baseline meniscal pathology scores, a larger proportion of participants with a lower baseline meniscal pathology score (0–4) had a meniscal pathology score increase (albeit small in magnitude); however, this was not related to steps/day (figure 2A). When analysing steps/day as a continuous variable the interaction was also significant (p=0.012).

Figure 2

Interactions between steps/day and baseline knee structures. There was a significant interaction between (A) steps/day and baseline meniscal pathology score on meniscal pathology score increases; (B) steps/day and baseline bone marrow lesions (BMLs) on cartilage defect score increases; and (C) steps/day and baseline cartilage volume on cartilage volume loss. Covariates common to all models included age, sex, body mass index, radiographic osteoarthritis, and history of knee injury or surgery. Depending on the structural outcome, models were further adjusted for baseline meniscal pathology score (A), baseline cartilage defect score and baseline BML (B), or baseline cartilage volume (C).

Cartilage defects

PA dichotomised at 10 000 steps/day was not associated with cartilage defect score increases as a main effect. However, there was an interaction between steps/day and baseline BML presence (figure 2B). Participants doing ≥10 000 steps/day who had a prevalent BML at baseline were 1.36 times (95% CI 1.03 to 1.69) more likely to increase their defect score compared to those with a prevalent BML doing <10 000 steps/day (p=0.013). There was no association seen in those without a prevalent BML at baseline.

When steps/day was analysed as a continuous variable the risk of a defect increase was 1.03 (95% CI 1.00 to 1.06) for each 1000 steps/day increase in PA (p=0.037). The interaction with baseline BML presence was borderline significant (p=0.070).

Cartilage volume

Steps/day was not associated with cartilage volume loss as a main effect, however, there was an interaction between steps/day and baseline cartilage volume (figure 2C). For participants in the lowest and middle third of baseline cartilage volume, doing ≥10 000 steps/day resulted in greater cartilage volume loss over time, while those in the highest third of cartilage volume had a smaller loss when compared with participants doing <10 000 steps/day. This interaction was also present when steps/day and baseline cartilage volume were analysed as continuous variables (p=0.031).

Tibial bone area

There was no association between steps/day and tibial bone area change (β=1.27, 95% CI −21.56 to 24.10).

Results were similar after adjustment for length of follow-up; exclusion of ROA from multivariable models, and; exclusion of those with any knee injury or surgery.

There were no significant interactions between BMI and steps/day on any knee structural change outcomes (data not shown).

Discussion

This 3-year longitudinal study has uniquely used objectively assessed PA and sensitive MRI techniques to assess knee structural change. Its findings suggest that PA is deleteriously associated with knee structural change, and this is mostly seen in those with pre-existing knee abnormalities.

Using the widely accepted cut-off of 10 000 steps/day34 we found that doing ≥10 000 steps/day was deleteriously associated with BML change, independent of important confounders including history of knee injury or surgery. Two recent cross-sectional studies have shown that PA was positively associated with BMLs.12 ,13 In addition, in vivo data demonstrated that exercise in horses was positively related to BMLs.36 Therefore the evidence for BMLs and PA is becoming consistent and the current study suggests ambulatory activity is deleteriously associated with BML development and progression.

Doing ≥10 000 step/day was associated with a 1.52 times greater risk of meniscal pathology score increase. In those with worse meniscal pathology at baseline, the risk increased to 2.49 times. This suggests that ambulatory activity is deleterious for meniscal health, especially in those with existing meniscal abnormalities. Two cross-sectional studies have shown that physically active individuals have a higher prevalence of meniscal abnormalities12 ,13; however, to the best of our knowledge this is the first longitudinal study.

PA was associated with cartilage defect increases and this relationship was also modified by underlying knee structure. Doing ≥10 000 step/day was associated with a 1.36 times greater risk of defect score increase in those with a prevalent BML at baseline. This interaction may help to explain the conflicting data seen in the literature. Cross-sectionally, Hanna et al17 found no association between exercise and cartilage defect presence; while Stehling et al12 ,13 showed that physically active individuals had more cartilage lesions. In contrast, Racunica et al16 demonstrated that vigorous PA was inversely associated with cartilage defects and data from a longitudinal study37 suggested that strenuous exercise was protective against cartilage defect progression.

Our finding that PA is significantly associated with adverse knee structural change is also clinically significant as all of these structural abnormalities have been linked with cartilage volume loss and knee joint replacement surgery.24 ,25 ,3841 Importantly, our data demonstrates the detrimental effect PA has on knee structure is especially seen in those with pre-existing knee abnormalities (ie, prevalent BMLs and/or meniscal pathology). This novel finding may help to explain the controversy surrounding PA and knee OA. PA exerts mechanical stimuli on the knee joint which may result in structural adaptations which could include chondrocyte death, disruption of the extracellular matrix, and microfractures within the subchondral bone.42 ,43 The presence of BMLs and meniscal pathology indicate early degenerative changes which may then make the knee joint more susceptible to the mechanical load from PA. We also found an important interaction between PA and baseline cartilage volume. PA was protective against volume loss in those with more cartilage volume at baseline but led to increased cartilage loss in those with less baseline cartilage volume. If these results are confirmed in other longitudinal studies they have implications for PA recommendations in people with knee abnormalities. PA may need to be tailored and future studies are needed to explore what type of activity is best for those with knee abnormalities.

We found no association between PA and tibial bone area change. This contrasts with our previous study showing that higher endurance fitness at baseline resulted in enlargement of the subchondral bone over 2 years37 in females. In the current study we did not find any sex differences when examining PA and tibial bone area change.

Obesity is a well-known major risk factor for the development of knee OA,44 but it is unclear whether overweight individuals who exercise may further increase their risk for joint damage.2 After adjusting for BMI and testing for BMI interaction terms we found the association between PA and knee structural change was not significantly different for those participants who were overweight or obese.

The strengths of our study include its longitudinal design, large sample size, objective measure of PA, and the use of MRI to assess joint structure. This study also has potential limitations. First, for the current study, 694 individuals were not included due to decommissioning of the MRI scanner at follow-up. There were no significant differences between the individuals studied and the rest of the cohort for many study factors and only a slightly higher baseline steps/day and meniscal pathology score in studied individuals. Second, our study assesses current PA rather than lifelong PA; therefore, we cannot make inferences about temporal changes in walking behaviour. Third, as PA was measured by pedometer, we were unable to examine the effect of impact or intensity. Fourth, we did not have data on knee alignment. Lastly, although we were able to adjust for history of knee injury/surgery, questions did not specify whether the injury/surgery was on the right or left knee. Also knee injury was only assessed at follow-up which may increase the risk of recall bias.

In conclusion, PA was deleteriously associated with knee structural change, especially in those with pre-existing knee structural abnormalities. This suggests individuals with knee abnormalities should avoid doing greater than 10 000 steps/day. Alternatives to weight-bearing activity may be needed in order to maintain PA levels required for other aspects of health.

Acknowledgments

We thank the subjects who made this study possible, and Catrina Boon and Pip Boon for their role in collecting the data, and André Pelletier for his expertise in MRI reading.

References

View Abstract

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

Footnotes

  • This work was presented at The European Congress on Osteoporosis and Osteoarthritis in Bordeaux, France, 21–24 March 2012, where it was accepted for an oral presentation.

  • Contributors DD was responsible for data collection, data management and cleaning, carried out analysis and interpretation of data, prepared the initial manuscript draft, and completed manuscript revisions. TW and PO participated in analysis and interpretation of the data, and critically revised the manuscript. CD was responsible for data collection, data management and cleaning, participated in analysis and interpretation of the data, and critically revised the manuscript. JPP and JMP participated in the study planning, carried out data collection, data management and cleaning, and critically revised the manuscript. FC designed and carried out the study planning, participated in analysis and interpretation of data, and critically revised the manuscript. GJ designed and carried out the study planning, participated in analysis and interpretation of the data, assisted with the initial manuscript draft, and critically revised the manuscript. All authors have approved the final manuscript and had full access to all of the data in the study. DD and GJ are the guarantors.

  • Funding This work was supported by the National Health and Medical Research Council of Australia; Tasmanian Community Fund; Masonic Centenary Medical Research Foundation; Royal Hobart Hospital Research Foundation; and Arthritis Foundation of Australia. The study sponsor had no role in the design of the study; the collection, analysis, and interpretation of the data; or the writing of the article and the decision to submit it for publication. The researchers work independently of their funders.

  • Competing interests JPP and JMP is a consultant for and shareholder in ArthroLab; the other authors declare no completing interests.

  • Patient consent Obtained.

  • Ethics approval Southern Tasmanian Health and Medical Human Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.