Radiographic OA progression over one-to-two years occurred in 58 (12%) participants. Results of this longitudinal study show that people who used NSAIDs and those who did not meet physical activity guidelines independently had double the odds of radiographic knee OA progression compared to people who did not. The study also confirms that people with more severe structural disease at baseline have greater odds of radiographic OA progression. These findings have implications for understanding the pathogenesis of medial knee OA and provide guidance regarding potential interventions for secondary prevention.
Clinical guidelines recommend use of NSAIDs for people with chronic knee pain unrelieved by simple analgesics30,31. International guidelines also state that NSAIDs should not be recommended for long-term use due to gastrointestinal and cardiovascular adverse effects32,33. While causality cannot be confirmed with this longitudinal cohort study, there may be several reasons why we identified a two-fold increased odds of radiographic OA progression among people who reported NSAID use at baseline. Potential mechanisms of elevated progression risk may be due to a physiological effect of the drugs, the representation of people with more severe symptoms, or a representation of people more likely to be physically active due to pharmacological treatment (if activity was proven detrimental to OA). Post-hoc analysis provides some support that NSAID users may represent people with greater symptoms, and thus more severe disease. However, this theory is unlikely to fully explain the findings because high baseline pain was not associated with radiographic OA progression in the multivariable analysis. There is also evidence of a potential physiological effect of NSAIDs. Other longitudinal observational and RCTs studies have identified an increased risk of OA progression in people who use NSAIDs, and this may be dependent on the type of NSAID. A placebo-controlled RCT reported that indomethacin use increased progression over a one-year period, compared with placebo, while tiaprofenic acid did not34. Similarly, a longitudinal cohort study of 2,330 people with hip or knee OA examined risk of radiographic OA progression with four common NSAIDs over a mean 7-year follow-up35. People who used diclofenac had 2.5-fold increased odds of knee OA progression if diclofenac was taken for 31–180 days, and 3.2-fold increased odds if diclofenac was taken > 180 days; whereas ibuprofen, naproxen and piroxicam were not associated with increased progression. Conversely, cohort data from the Osteoarthritis Initiative (OAI) did not support an increased risk with NSAIDs among their sample of 2,890 participants36. There may also be support for protective effects of some NSAID classes in other longitudinal studies, with low dose aspirin use associated with less medial tibial cartilage loss over two years compared to non-aspirin users37. The high proportion of aspirin use in our study was likely for cardiovascular prophylaxis given the sample was middle aged, overweight, and had high blood pressure. A systematic evaluation of the clinical measures of disease progression with different NSAIDs is required for a clearer understanding of potential effects.
People who did not meet guidelines for adequate physical activity had double the odds of radiographic OA progression compared to those who met the guidelines. Due to the observational design, we can only speculate about the causes of the risk. These may include the physiological effect of low physical activity, an association with poor general health or comorbidity, or presence of more pain or activity limitations in these individuals. Post-hoc analysis does not provide support for the abovementioned potential hypotheses. Although a slightly greater proportion of participants not meeting physical activity guidelines were allocated to treatment groups other than the “double active” combined supplements group, thus potentially representing less people in the active therapy group which slowed OA progression in the main trial13. This theory is unlikely to explain the findings because treatment allocation was not associated with radiographic OA progression in the multivariable analysis. Despite anecdotal suggestions that high physical activity may be detrimental to OA progression via increased cumulative loading38, our results suggest that maintaining adequate physical activity is not associated with increased risk and may potentially slow OA progression. Despite the lack of statistical significance in previous studies of physical activity, and a 95% CI which crosses 1.0 in our study, the previously reported mean ORs (range: 0.4 to 0.7) suggest agreement with our findings and a potential protective effect of physical activity. Previous studies6,12 did not specifically evaluate physical activity duration or intensity at baseline, had smaller cohorts and did not consider radiographic factors in their analyses. Our findings for the protective effects of physical activity are also consistent with a systematic review which identified that runners had a 50% reduced odds of requiring surgery due to their knee OA compared to non-runners39. A mechanism of joint protection is unknown, however the link between inactivity and chronic low-grade inflammation may explain the identified risk40. Muscle strength may present another possible explanation, with high levels of physical activity associated with greater quadriceps strength41, which is speculated to support the joint with improved shock-absorption and load distribution42. However, associations between muscle strength and disease progression are conflicting and require further investigation5,11. Studies implementing a physical activity intervention should evaluate effects on joint biology, mediating factors and OA progression.
The strongest risk factors for radiographic OA progression were baseline structural disease severity measures, indicating that once OA is clearly established on x-ray, OA progression is likely. Our findings are consistent with previous studies which identified similar strengths of association (mean OR range: 1.7–4.5)11. Varus alignment is also a known factor associated with radiographic knee OA progression, with the current study identifying a two-fold increase in odds. This is in agreement with published studies (reported ORs range: 2.3–11.0)11. The variability in the magnitude of ORs is likely due to different criteria used to measure progression, the angle of varus alignment and the covariates in the statistical model.
We did not find an association between daily combined glucosamine and chondroitin supplements and the OARSI-OMERACT recommended target for defining disease progression (JSN ≥ 0.5 mm over two years). While no effect was identified for rapid progression, the original study confirmed that participants taking the combination supplement experienced a mean of 0.1 mm less JSN in comparison to the placebo group over two years13. This is in agreement with a longitudinal analysis involving the OAI cohort43, which identified a disease modifying effect of glucosamine and chondroitin sulfate among people with early OA but not among people with more severe OA. While this benefit has not been shown among people with moderate-to-severe disease44, the small disease-modifying effect of combined glucosamine and chondroitin during early stages of OA may potentially delay the requirement for joint replacement surgery.
Risk of radiographic OA progression will be higher for people with multiple risk factors. As demonstrated in Fig. 1, 5% of people with only one risk factor demonstrated radiographic OA progression, whereas 36% of people with four or more factors demonstrated progression. Clinicians should routinely assess known risk factors to identify people at greatest risk of progression and potential need for joint replacement surgery. Lastly, we did not find an association with age, sex, or obesity, consistent with findings from a systematic review evaluating risk factors of radiographic OA progression11.
Strengths of this study include the prospective design, consideration of baseline structural disease severity, evaluation of clinically-relevant modifiable factors, and participants with predominantly mild radiographic OA at baseline. We also employed a standardised radiographic protocol for image acquisition and evaluation by a single researcher. There are some limitations to be considered when interpreting these findings. As recruitment was for an intervention involving supplements, our results apply to people with symptomatic knee OA seeking treatment. Our analysis was based on self-reported regular NSAID use over the seven days prior to baseline assessment, which may not reflect continual use of NSAIDs throughout the study. As the study was not powered to separately evaluate risk of different NSAID classes, future studies should determine if there is a differential effect specific to the class and/or dosage of NSAIDs. However, the significant finding maintained in our post-hoc analysis for NSAIDs with the exclusion of aspirin confirms the presence of increased risk. Our analysis adjusted for nine identified and/or known confounding variables. It is possible that other factors not evaluated may be important to further understand the relationship between NSAID use, physical activity, and radiographic OA progression. With observational longitudinal study designs of radiographic disease progression, there is a possibility of collider bias45. As only participants with pre-existing radiographic knee OA were included, conditioning on radiographic OA may bias the effect of risk factors, such as baseline structural disease severity, towards the null effect. Good quality two-year follow-up radiographs were available for 61% of participants, limiting our ability to detect all people with JSN ≥ 0.5 mm. Finally, as this was an observational longitudinal study, risk factors imply an increase in the odds of disease progression, and causality needs to be confirmed using RCT designs.
Meeting physical activity guidelines should be encouraged among people with symptomatic knee OA with mild structural disease severity, as this may be protective in the risk of radiographic OA progression over one-to-two years. Clinicians, and people with knee OA, need to be aware of the increased risk of progression among people who use NSAIDs. The strongest associations with radiographic progression were features of disease severity. Interventions targeting prevention of radiographic OA progression are likely to be most effective during the early stage of knee OA, when only mild features are evident on radiographs.