Article Text
Abstract
Objective Obesity is a risk factor for knee osteoarthritis (KOA) development and progression. Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are indicated for type 2 diabetes mellitus (T2DM) and obesity. However, whether KOA patients can benefit from GLP-1RA therapies has not been sufficiently investigated, especially in the long term.
Methods The Shanghai Osteoarthritis Cohort study is a prospective, observational, multicentre study of >40 000 adults with clinically diagnosed osteoarthritis aged >45 years in Shanghai. We identified all KOA participants with comorbid T2DM enrolled from 1 January 2011 to 1 January 2017. Primary outcome was incidence of knee surgery after enrolment. Secondary outcomes included pain-relieving medication use, number of intra-articular therapies, Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) and medial femorotibial joint cartilage thickness. To evaluate the effects of GLP-1RA, we performed before-and-after comparison and comparison with participants who had no GLP-1RA exposure.
Results For an intergroup comparison (non-GLP-1RA vs GLP-1RA), more weight loss (adjusted mean difference in weight change from baseline −7.29 kg (95% CI −8.07 to −6.50 kg), p<0.001) and lower incidence of knee surgery (93/1574 (5.9%) vs 4/233 (1.7%), adjusted p=0.014) were observed in the GLP-1RA group. Statistically significant differences in mean change from baseline for the WOMAC total and pain subscale scores were observed (adjusted mean difference in WOMAC total score −1.46 (95% CI −2.84 to −0.08), p=0.038; adjusted mean difference in WOMAC pain subscore −3.37 (95% CI −5.79 to −0.94), p=0.007). Cartilage-loss velocity of the medial femorotibial joint was significantly lower in the GLP-1RA group postadjustment for baseline characteristics (adjusted mean difference −0.02 mm (95% CI −0.03 to −0.002 mm), p=0.004). For the before-and-after comparison within the GLP-1RA group, we observed a significant decrease of symptom-relieving medication consumption and cartilage loss velocity of medial femorotibial joint (after-treatment vs before-treatment: −0.03±0.05 vs −0.05±0.07 mm/year, p<0.001). The association between GLP-1RA exposure and decreased incidence of knee surgery was mediated by weight reduction (mediation proportion: 32.1%), instead of glycaemic control (too small to calculate).
Conclusion With sufficient treatment duration, GLP-1RA therapies might be disease-modifying for KOA patients with comorbid T2DM, possibly mediated by weight loss. Further investigation is needed to elucidate effects of GLP-1RA on disease process, joint structure and patient-reported outcomes of osteoarthritis.
- Knee Osteoarthritis
- Osteoarthritis, Knee
- Magnetic Resonance Imaging
- Therapeutics
- Patient Reported Outcome Measures
Data availability statement
All relevant anonymised patient-level data are available on reasonable request to the corresponding author.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
- Knee Osteoarthritis
- Osteoarthritis, Knee
- Magnetic Resonance Imaging
- Therapeutics
- Patient Reported Outcome Measures
WHAT IS ALREADY KNOWN ON THIS TOPIC
Weight management is considered as a first-line intervention for knee osteoarthritis (KOA). Recent evidence has indicated that glucagon-like peptide-1 receptor agonists (GLP-1RAs), which have effects on reducing weight, provide a small improvement in short-term patient-reported outcomes.
WHAT THIS STUDY ADDS
Results of this observational study of KOA patients with comorbid type 2 diabetes mellitus indicated long-term effects of GLP-1RAs on KOA progression, with a lower incidence rate of knee surgery in patients receiving GLP-1RAs than in the control group (non-GLP-1RA exposure).
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
GLP-1RAs might be potentially disease-modifying OA drugs for KOA, although the benefits might require a long treatment duration.
Introduction
Osteoarthritis (OA) is a highly prevalent, disabling disease with a tremendous individual and socioeconomic burden.1 According to the Global Burden of Disease Study 2019,2 the disease burden of OA has been growing rapidly worldwide over the past decades.3 Controlling modifiable risk factors, most importantly, maintaining an appropriate weight/body mass index (BMI), is vital for preventing disease development and progression.4 Multiple guidelines recommend weight control as a basic measure for long-term OA management.5–7 However, the disease burden of OA associated with a high weight/BMI has remained a continuous upward trend worldwide in the past decades.8 9 Notably, despite the long-term benefits of weight control on knee OA (KOA), weight loss via diet modification, physical activities and/or medications only lead to a small improvement in patient-reported outcomes (PROs) of uncertain clinical importance in short term (<2 years).10 11
Sustaining weight control in the long term remains a major challenge for the general population including KOA patients.12–14 Many clinical guidelines recommend adjunctive medications for obese persons, especially those with comorbidities (such as T2DM, cardiovascular diseases and non-alcoholic fatty liver disease).15–17 Glucagon-like peptide-1 receptor agonists (GLP-1RAs) are a class of medications that are effective treatment for patients with type 2 diabetes mellitus (T2DM) and weight control by stimulating insulin secretion, suppressing glucagon secretion, delaying gastric emptying and decreasing appetite.18 19 Many GLP-1RAs, including semaglutide, liraglutide and dulaglutide, have been approved for T2DM and weight management.20 Patients treated with liraglutide therapy lost more body weight than those treated with placebo for at 1 year (a difference of −5.6 kg; 95% CI −6.0 to −5.1 kg).21 In the STEP 1 trial, semaglutide treatment achieved sustained and clinically important reduction in body weight (estimated treatment difference compared with placebo −12.7 kg; 95% CI −13.7 to −11.7 kg).22
Thus, GLP-1RAs are believed to be potential disease-modifying OA drugs for treating OA, with the rationale that GLP-RAs could prevent cartilage loss by reducing the mechanical stress from body weight23 and PROs by reducing pain sensitivity.24 To the best of our knowledge, only one clinical study has investigated the efficacy of liraglutide on knee OA (KOA), in which liraglutide did not reduce knee pain compared with placebo at 1 year in KOA patients with obesity or those that are overweight.11 However, this trial started with a required weight loss of >5% before randomisation and only a small weight loss (<5%) was achieved after liraglutide treatment. This might explain why the results were not significant. Thus, a longer study duration with significant weight loss/maintenance and structural data might be needed to answer this question. In this study, to explore potential disease-modifying effects of GLP-1RAs, we analysed prospectively collected, multicentre, observational data from the Shanghai Osteoarthritis Cohort (SOC) study.
Methods
Patients
We identified all KOA participants who were enrolled in the SOC study from 1 January 2011 to 1 January 2017. The SOC study prospectively recruited more than 40 000 adults with clinically diagnosed OA aged >45 years from four hospitals in Shanghai. For this study, we included patients with comorbid T2DM at baseline who had at least completed the 5-year follow-up. Patients were eligible for this study if they had baseline bilateral plain radiographs demonstrating Kellgren and Lawrence (K-L) grades 1‒3 KOA (K-L grade was recorded according to the more severe side).25 Participants were excluded from the analyses if they had secondary OA, K‒L grade 0 or 4, diabetic vascular diseases, diabetic foot and knee surgery history at baseline. The clinical diagnosis of KOA was performed by clinical specialists in orthopaedic and/or sports medicine. It was determined based on the patient history, physical examination, and laboratory and radiographic findings.26 These patients were grouped according to whether they received GLP-1RA therapies for the treatment of T2DM. Notably, patients who received GLP-1RA for less than 2 years were also excluded from the analysis (online supplemental figure S1).
Supplemental material
Baseline data collection and PROs
Baseline demographic characteristics, including age, sex and weight, were self-reported by the participants. A weight change greater than 5% was considered clinically relevant for KOA.27 Standard weight-bearing anteroposterior and lateral plain radiographs of the knee were obtained at enrolement, and the K-L scoring system was used to grade the radiographic stages of KOA. The K-L grades were reviewed and rated by an independent radiographic evaluation committee consisting of three radiologists specialising in musculoskeletal radiology. A consensus on the grading was achieved after discussion. When the two knees had different K-L grades, the final readout used in the current study was recorded according to the more severe side (index side). For PROs, the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) questionnaire, a well-validated hip and KOA pain instrument consisting of 24 questions, was used to assess three separate dimensions (pain, physical function and stiffness) in KOA.28 In this study, for convenience and comparison with previously published literature, all subscales were normalised to scores within a range of 0–100. The minimal clinically important difference (MCID) for the WOMAC total score was 7 (95% CI 4 to 10); pain subscale, 9 (95% CI 6 to 12); function subscale, 6 (95% CI 3 to 9); and stiffness subscale, 7 (95% CI 6 to 9).29
Incident knee surgery
Incident knee surgery was defined as all surgical procedures performed to treat KOA after patient enrolment in SOC. In this study, incident knee surgery included total knee arthroplasty, unicompartmental knee arthroplasty, arthroscopic procedures and high tibial osteotomy. Notably, although arthroscopic procedures, including lavage, debridement and arthroscopic partial meniscectomy, are ineffective and even harmful for KOA patients,30 31 evidence published to date has not led to a major decline in arthroscopic procedures for managing KOA.32 33 Thus, we also considered incident arthroscopic procedures as meaningful events of poor symptom control in this observational study. We adopted the latest PROs prior to surgery and analysed the structural outcomes using the earliest and latest MRI scans during the study period.
Definition of recommended daily dose and morphine milligram equivalents
To allow direct comparisons of analgesics of different potencies and formulations, we converted the quantities of non-opioid medication use, including acetaminophen and topical and oral non-steroidal anti-inflammatory drugs (NSAIDs) into recommended daily doses (RDDs). RDD was defined as the RDD for treating OA, if applicable. When multiple recommended doses were present, the RDD was calculated by averaging the highest and lowest recommended doses. When no specific recommended dose regarding OA was present, the RDD was calculated by averaging the highest and lowest recommended doses for all indications. The consumption of opioid analgesics was converted into morphine milligram equivalents (MMEs) for each opioid-containing product. The MME conversion factors and RDD of analgesics were collected from various sources, and the details are available in online supplemental table S1 and S2.
Measurement of cartilage loss velocity
We measured the cartilage thickness in the medial femorotibial cartilage plates (tibial and weight-bearing femur): mean cartilage thickness over the total area of the subchondral bone (mm).34 The weight-bearing region of the femoral condyles was defined as the area between the intercondylar notch and 60% of the distance to the posterior end of the femoral condyles.35 The cartilage loss velocity was calculated as follows: cartilage loss velocity (mm/year)=(cartilage thickness at time point A)–(cartilage thickness at time point B)/(duration between time points A and B). To compare the GLP-1RA and non-GLP-1RA groups, we adopted the earliest and latest MRI scans for the index side knee during the study period. We only included patients who underwent at least two MRI scans with a minimal 6-month interval for comparison of cartilage loss. Full imaging methods appear in online supplemental material.
Before-and-after comparison within the GLP-1RA group
For before-and-after comparison, we included patients who had a minimal 2-year follow-up before the start of GLP-1RA therapy. We included 126 of 233 patients from the GLP-1RA group before and after the comparison. For structural comparison, we needed at least two MRI scans for both periods (before and after GLP-1RA therapy).
Statistical analysis
Continuous and categorical variables are presented as means±SD and counts (percentages), unless otherwise indicated (median (quartile)). Univariate analyses were conducted using the t-test (or Mann-Whitney U test) and Pearson’s χ2 test (or Fisher’s exact test). Multivariable regression models (linear or logistic regression) were used to compare the mean changes in related variables and knee surgery incidence between the GLP-1RA therapy and non-GLP-1RA therapy groups. All multivariable analyses were adjusted for baseline covariates (age, sex, BMI, K-L grade and WOMAC total score) or other covariates with a p<0.2 in univariate analysis. The effects were reported with an adjusted mean difference and its 95% CI.
Considering that the effect of GLP-1RA on outcomes of KOA may primarily rely on the change in weight after receiving GLP-1RA,36–38 and the poor control of blood glucose may be a risk factor for the progression of OA,39–42 we established an ‘exposure–mediator–outcome’ model and viewed weight and hemoglobin A1c (HbA1c) change as a mediator. In this model, the total effect of exposure (GLP-1RA usage) on the outcome (knee surgery) was divided into the ‘direct effect’ of exposure outcome and the ‘indirect effect’ on a pathway through the mediator (weight change after using GLP-1RA) (figure 1). We performed a model-based causal mediation analysis to calculate the proportion of ‘indirect effect’ and its 95% CI (simulated by the quasi-Bayesian Monte Carlo method based on normal approximation43) to estimate the proportion of GLP-1RA use to outcomes (knee surgery, cartilage loss velocity and WOMAC pain subscore) effect attributable to the pathway of weight and HbA1c change. The set of pre-exposure covariates (age, sex, WOMAC total score, BMI and K-L grade) satisfied the assumption of confounding adjustment for the exposure–mediator–outcome relationships.
Three sensitivity analyses were conducted. First, to minimise the potential bias due to weight change experienced by participants before GLP-1RA exposure, we conducted an analysis of weight and PROs using the most recent weight and PROs before GLP-1RA usage. Second, to prevent the occurrence of a possible floor effect where differences in score reductions might be difficult to discern, we excluded patients with WOMAC total score lower than 7 (which is the MCID for the WOMAC total score) at baseline. Third, it could find some participants had GLP-1RA exposure before enrolment and we were unable to collect data before enrolment. To partially overcome this problem, we excluded those patients who had GLP-1RA exposure within the initial half year after enrolment in the sensitivity analysis. In addition, we performed subgroup analyses that were stratified based on baseline KL grade and change in weight.
All statistical assessments were performed in a two-sided fashion, and a p<0.05 was considered statistically significant. Statistical analysis was conducted using IBM SPSS V.26.0, and the ‘mediation’ package in R V.4.1.2 was applied for mediation analysis in this study.44
Results
We included 1807 clinically diagnosed KOA patients with comorbid T2DM for analysis from the established cohort. GLP-1RA and non-GLP-1RA had 233 and 1574 participants, respectively (table 1).
The enrolment year for participants and the year of initial incident exposure to GLP-1RA of participants are presented in online supplemental table S3 and S4, respectively. The mean treatment duration of GLP-1RAs was 4.9±1.9 years. The GLP-1RA and non-GLP-1RA groups had similar mean weight at baseline (66.0±12.2 vs 65.1±12.3, p=0.31), whereas we observed a substantial reduction in weight in the GLP-1RA group at the last follow-up (change from baseline, GLP-1RA vs non-GLP-1RA, −4.60±8.07 vs 2.69±5.23, p<0.001) (table 2). Clinically relevant gains in weight were observed in 18.5% (43/233) and 45.9% (722/1574) of patients in the GLP-1RA and non-GLP-1RA groups, respectively; clinically relevant reductions in weight were observed in 57.9% (135/233) and 13.4% (211/1574) of the GLP-1RA and non-GLP-1RA groups, respectively (online supplemental table S5). Online supplemental tables S6 and S7 show the results for comparison of outcomes between participants who achieved a clinically significant reduction in weight and those who did not. The baseline demographic and clinical characteristics of the GLP-1RA (n=233) and non-GLP-1RA (n=1574) groups are summarised in table 1; the history of GLP-1RA use is shown in online supplemental table S8. Based on the knee radiographs, 42 out of 233 (18.0%) individuals in the GLP-1RA group and 298 out of 1574 (18.9%) individuals in the control group demonstrated predominantly lateral OA.
Comparison of PROs and incident knee surgery between the GLP-1RA and non-GLP-1RA groups
Statistically significant differences were observed in the mean absolute change from baseline for the WOMAC total and pain subscale scores (adjusted mean difference in WOMAC total score −1.46 (95% CI −2.84 to −0.08), p=0.038; adjusted mean difference in WOMAC pain subscore −3.37 (95% CI −5.79 to −0.94), p=0.007). Compared with the non-GLP-1RA (93/1574) group, we observed a substantially lower incidence of knee surgery in the GLP-1RA (4/233) group (5.9% vs 1.7%, adjusted p=0.014) (table 2). Sensitivity analysis of weight and PROs using the most recent weight and PROs before GLP-1RA usage also showed the statistically significant results for WOMAC total and pain subscale scores (adjusted mean difference in WOMAC total score −2.03 (95% CI −3.41 to −0.65), p=0.004; adjusted mean difference in WOMAC pain subscore −3.93 (95% CI −6.29 to −1.56), p=0.001) (online supplemental tables S9). Additionally, sensitivity analyses of PROs and incident knee surgery after excluding patients with low WOMAC total scores at baseline also showed stable results (adjusted mean difference in WOMAC total score −1.49 (95% CI −2.93 to −0.05), p=0.043; adjusted mean difference in WOMAC pain subscore −3.31 (95% CI −5.85 to −0.77), p=0.011; incidence of knee surgery, GLP-1RA vs non-GLP-1RA, 1.9% vs 5.7%, adjusted p=0.026) (online supplemental tables S10-11). The proportion of patients who achieved MCID improvement in the two groups were compared and are presented in online supplemental table S12 and S13. The association between GLP-1RA exposure and WOMAC pain subscore was not mediated by weight reduction and HbA1c change (online supplemental table S14). For the GLP-1RA group, 2 underwent TKA and 2 underwent arthroscopic procedures; for the non-GLP-1RA group, 30 underwent TKA, 1 underwent UKA, 6 underwent HTO, 53 underwent arthroscopic procedures and 3 underwent TKA following arthroscopic procedures (online supplemental table S15).The association between GLP-1RA exposure and decreased incidence of knee surgery was substantially mediated by weight reduction, but much less so by HbA1c change (the mediation proportion for weight reduction was 32.1% for GLP-1RA exposure) (table 3).
In subgroup analysis with patients who reached clinically relevant reduction on weight (defined >5%), we observed a trend towards significance for the comparison between GLP-1RA and non-GLP-1RA groups (0/135 (0.0%) vs 6/211 (2.8%), p=0.085) (online supplemental table S16). For sensitivity analysis with only incident users of GLP-1RA (defined as no exposure of GLP-1RA in the initial 6 months after enrolment), compared with the non-GLP-1RA group, we also observed significant differences in mean change from baseline for the WOMAC total and pain subscale scores (adjusted mean difference in WOMAC total score −1.57 (95% CI −3.03 to −0.12), p=0.034; adjusted mean difference in WOMAC pain subscore −3.15 (95% CI −5.70 to −0.60), p=0.016) and a lower incidence of knee surgery in the GLP-1RA group (5.9% vs 1.9%, adjusted p=0.027) (online supplemental table S17).
Comparison of symptom-relieving medication use between the GLP-1RA and non-GLP-1RA groups
We observed numerical but statically nonsignificant decrease in the GLP-1RA group compared with the non-GLP-1RA group in terms of annual consumption of oral NSAIDs and acetaminophen (15.2±13.8 vs 16.9±14.5 RDD/year, p=0.10), topical NSAIDs (21.7±22.7 vs 23.6±22.8 RDD/year, p=0.22), opioids (109.0±190.0 vs 126.2±205.6 MME/year, p=0.20), number of intra-articular therapies (1.07±1.99 vs 1.30±2.12, p=0.10). The GLP-1RA group required fewer number of intra-articular injection of steroids compared with the non-GLP-1RA group (0.13±0.28 vs 0.22±0.39, p<0.001) (online supplemental figure S2, table 2).
Comparison of structural outcomes between the GLP-1RA and non-GLP-1RA groups
We identified 188 and 1267 patients who received at least two MRIs in the GLP-1RA and non-GLP-1RA groups, respectively. The mean duration between the earliest and latest MRI scanning was 4.5±2.1 and 4.3±2.3 years for participants in the GLP-1RA and non-GLP-1RA groups, respectively. Thirty-two out of 188 (17.0%) individuals in the GLP-1RA group and 231 out of 1267 (18.2%) individuals in the control group showed predominantly lateral OA. In this subcohort, cartilage-loss velocity of the medial femorotibial joint was significantly lower in the GLP-1RA group than in the non-GLP-1RA group after adjustment for baseline characteristics, including age, sex, BMI, WOMAC total score and K-L grades (−0.05±0.08 vs −0.07±0.10 mm/year; adjusted mean difference, 0.02 mm (95% CI 0.002 to 0.033), adjusted p=0.026) (table 4). For subgroup and sensitivity analyses for (1) patients who reached clinically relevant reduction on weight; (2) incident users of GLP-1RA and (3) with different baseline K-L grades, we have summarised data in online supplemental table S18–20. The association between GLP-1RA exposure and cartilage loss velocity was not mediated by weight reduction and HbA1c change (online supplemental table S21).
Before-and-after comparison within the GLP-1RA group
We observed a significant decrease in pain-relieving medication use after GLP-1RA treatment compared with pretreatment (oral NSAIDs and acetaminophen (post-treatment vs pretreatment, 14.0±12.2 vs 16.8±14.7 RDD/year, p<0.001), topical NSAIDs (16.6±19.6 vs 24.5±24.3 RDD/year, p<0.001), opioids (93.5±182.3 vs 97.7±195.1 MME/year, p=0.70)). Patients required fewer intra-articular therapies (0.76±1.40 vs 1.35±2.67, p<0.001) and fewer intra-articular injections of steroids (0.10±0.21 vs 0.18±0.41, p<0.001) after GLP-1RA therapies. The cartilage loss velocity of the medial femorotibial joint was significantly lower after GLP-1RA treatment compared with the pretreatment level (−0.03±0.05 vs −0.05±0.07 mm/year, n=61, p<0.001) (online supplemental figure S3, table 5).
Discussion
This is the first clinical investigation to examine the long-term effects of GLP-1RA on KOA in patients with comorbid T2DM. We noticed significant differences in PROs pertaining to the WOMAC score, both total and pain subscale, between the GLP-1RA and non-GLP-1RA groups. Furthermore, the cartilage loss velocity and knee surgery incidence were both statistically lower in the GLP-1RA group. Importantly, the GLP-1RA group also required fewer intra-articular injections of steroids than the non-GLP-1RA group. The before-and-after comparison within the GLP-1RA group and additional sensitivity analyses further supported our findings.
Weight loss has been reported to be a highly effective approach for patients with OA, particularly those with obesity.5–7 According to conventional wisdom, a weight change greater than 5% is considered clinically relevant for KOA.27 In this study, the weight loss after GLP-1RA therapy (such as the use of semaglutide, liraglutide and dulaglutide) was substantial and consistent with previous reports.45 Our analysis revealed that the effects of GLP-1RA on arthritic knees were largely mediated by weight loss instead of glycaemic control. This observation is as expected because the long-term benefits of weight control in KOA have been well established.5–7 In contrast, several preclinical studies have revealed that GLP-1RA has anti-inflammatory and antidegradative effects.46 47 It is reasonable to speculate that GLP-1RA might have direct effects on KOA progression. Nonetheless, the ‘direct effects’ of GLP-1RA on knee surgery, apart from the weight loss-mediated pathway, did not reach statistical significance (table 3). We also fail to identify statistically significant effects within the weight loss subgroup (online supplemental table S18). Consequently, we hypothesise that benefit of GLP-1RA is mainly associated with weight loss while the direct effects remain unclear.
Because GLP-1RAs have a well-established profile in weight management, it is rational to consider that OA patients might benefit from GLP-1RA therapies via reduced mechanical stress and pain sensitivity elicited by overweight.5–7 20 24 However, a previous trial reported that liraglutide did not reduce knee pain compared with placebo at 1 year in KOA patients that were overweight or obese.11 Notably, as stated before, the design of this trial is problematic and could not provide a confirmative conclusion. However, a recent trial found a statistically significant correlation between weight loss and pain reduction in the diet and exercise intervention group (p<0.001), and the difference in knee pain was statistically significant but small (p=0.02) compared with an attention control group at the 18-month follow-up.10 Therefore, patients with OA might benefit from GLP-1RA therapy for over a longer duration. Thus, in this study, we compared data from patients who received GLP-1RA therapies for at least 2 years with those from the control group. We observed significant effects on the WOMAC total and pain subscale scores during the extended follow-up period. In addition, a larger mean intergroup difference in weight change was noted in this study compared with the previous research,10 11 which was consistent with earlier finding that substantial weight loss could exert its effect on knee pain improvement.24
Notably, different strategies of weight loss, including diet, exercise, and medications, differentially affect body weight, composition, and muscle strength.48 Although GLP-1RA therapies led to a greater reduction in fat mass, lean body mass was also reduced significantly after treatment due to the treatment-emergent hypocaloric diet.49 50 However, a longitudinal study demonstrated an increased risk of KOA with obesity and sarcopenic obesity, but not sarcopenia.51 This might avert part of the worry and concern regarding GLP-1RA therapies. For patients with sarcopenic obesity, physical exercise might be essential to preserve lean body mass, as exercise with weight loss could not only decrease ectopic fat but also improve insulin sensitivity.48 Future trials should consider an add-on design to evaluate the efficacy of GLP-1RA therapies for KOA.
We observed potentially preventive effects of GLP-1RA therapies on articular cartilage in the intergroup and before-and-after comparisons. Articular cartilage is a connective tissue composed of chondrocytes and chondrocyte-producing extracellular matrix.52 Obesity may impair cartilage homoeostasis and cause systemic and local inflammation, whereas weight loss can improve the quality and quantity of articular cartilage.53 54 Furthermore, as antianabolic effects of steroids on healthy cartilage are known, the use of intra-articular steroids could result in greater cartilage volume loss compared with placebo.55 56 Moreover, we observed less progression of knee cartilage loss velocity with fewer number of steroids required in the GLP-1RA group. In addition to indirect effects, GLP-1RA is expressed in normal and OA articular chondrocytes, suggesting that GLP-1RAs might have a direct impact on articular chondrocytes.57 Preclinical studies have shown that GLP-1RA signalling is associated with apoptosis prevention, anti-inflammatory activity and matrix protection.23 Future preclinical and clinical investigations may elucidate the underlying mechanisms.
A statistically insignificant decrease was observed for symptom-relieving medication use in the GLP-1RA group compared with the non-GLP-1RA group. This might be because most patients in the GLP-1RA group did not receive GLP-1RA therapies immediately after enrolling in SOC. In contrast, before-and-after comparisons within the GLP-1RA group showed a significant decrease in symptom-relieving medication use except for opioids after GLP-1RA treatment compared with the pretreatment level. Chronic use of opioids is associated with an increased risk of fractures, cardiovascular events, opioid dependence and mortality; opioids were generally considered as second-line choices before intra-articular therapies for KOA.58 59 Likewise, in this study, opioids were less frequently consumed by participants at about 100 MME per year on average. Thus, the consumption of opioids might not be of clinical importance.
The current study had several limitations. First, because of the large sample size of SOC study, we only had access to routine follow-up on PROs and were unable to perform on-site and routine radiographic follow-up (eg, measured body weight, MRI and X-ray) for our participants owing to limited funding. Future randomised trials are required to validate our findings as this study did not draw any confirmatory conclusions regarding the efficacy of GLP-1RA therapy for KOA. Second, the current study enrolled only KOA patients with comorbid T2DM. Extrapolation of our findings to a general KOA population should be done cautiously, especially given the fact that KOA patients with comorbid T2DM have a higher prevalence of obesity/overweight. Third, the rationale behind the decision to use GLP-1RA was not exactly recorded. Clearly, preferences of the treating physician and patients played important roles regarding this decision. Indication bias (eg, easier access to newer drugs, paying more attention to their health) is inevitable. For structural data, patients who underwent more frequent MRI scans may have been more vigilant about their health. Fourth, because of the observational nature of this study, the switch between different GLP-1RAs occurred frequently during our study period (history of GLP-1RAs use is shown in online supplemental table S8). For example, many patients in this study switched from liraglutide to semaglutide after the latter was made commercially available. Finally, approximately 20% of patients did not have available MRI in this study as MRI examination was not compulsory according to our protocol. Patients underwent MRI examination based on the physicians’ recommendations and their preference, resulting in irregular intervals between MRI scans. Therefore, it is not clear whether the cartilage loss in this study was linear or not.
In conclusion, with sufficient treatment duration, GLP-1RA therapies might be disease-modifying for KOA patients with comorbid T2DM. Further investigations are needed to elucidate the effects of GLP-1RA on the disease process, joint structure and PROs of OA.
Data availability statement
All relevant anonymised patient-level data are available on reasonable request to the corresponding author.
Ethics statements
Patient consent for publication
Ethics approval
This study was approved by the Ethics Committee of Shanghai Sixth People’s Hospital (approval no. 2010-136). Moreover, this study complied with the Code of Ethics of the World Medical Association (Declaration of Helsinki) for research involving humans.
Acknowledgments
We would like to acknowledge all the SOC investigators and staff at participating centres.
References
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.
Footnotes
HZ and LZ are joint first authors.
YS and CZ are joint senior authors.
Handling editor Josef S Smolen
Collaborators Changqing Zhang - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of ChinaQirong Qian - Department of Orthopaedics, Shanghai Changzheng Hospital, Shanghai, People’s Republic of China, Zhenjun Yao - Department of Orthopaedics, Zhongshan Hospital Fudan University, Shanghai, People’s Republic of China, Guodong Li - Department of Orthopaedics, Shanghai Tenth People’s Hospital, Shanghai, People’s Republic of China, Yong Feng - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Yigang Huang - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Chenyi Jiang - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Weitao Jia - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China. Guangyi Li - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Xiaochun Peng - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Hao Shen - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Huipeng Shi - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Yunchao Shao - Department of Orthopaedics, Zhongshan Hospital Fudan University, Shanghai, People’s Republic of China, Kun Tao - Department of Orthopaedics, Shanghai Tenth People’s Hospital, Shanghai, People’s Republic of China, Bo Wang - Department of Orthopaedics, Shanghai Changzheng Hospital, Shanghai, People’s Republic of China, Xuetao Xie - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Zongping Xie - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Ting Yuan - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Xiaowei Yu - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Xiaoyu Yan - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Yi Yang - Department of Orthopaedics, Zhongshan Hospital Fudan University, Shanghai, People’s Republic of China, Zhenzhong Zhu - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Yanyan Song - Department of Biostatistics, Clinical Research Institute, Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Qianying Cai - Shanghai Institute of Microsurgery on Extremities, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Shengbao Chen - Shanghai Institute of Microsurgery on Extremities, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Kai Fu - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Kaiyuan Liu - Department of Orthopaedics, Shanghai Tenth People’s Hospital, Shanghai, People’s Republic of China, Huan Wang - Department of Orthopaedics, Shanghai Changzheng Hospital, Shanghai, People’s Republic of China, Shengcheng Wan - Department of Orthopaedics, Zhongshan Hospital Fudan University, Shanghai, People’s Republic of China, Hongyi Zhu - Department of Orthopaedics, Shanghai Sixth People’s Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, People’s Republic of China, Yiqin Zhou - Department of Orthopaedics, Shanghai Changzheng Hospital, Shanghai, People’s Republic of China.
Contributors HZ and LZ contributed equally to this work. HZ, LZ, YS and CZ designed the study. YS was responsible for statistical analysis and methodology. All authors were involved in analysing and interpreting the data. HZ, LZ and QC drafted the manuscript. QW, FY, HJ and YC critically revised the manuscript for important intellectual content. HZ and QC did the statistical analysis. HZ and CZ obtained funding and provided administrative support. CZ supervised the study. The corresponding author attested that all listed authors meet authorship criteria and that no others meeting the criteria have been omitted. HZ, LZ, YS and CZ are the guarantors. LZ and QC have directly accessed and verified the underlying data reported in the manuscript. All authors critically reviewed this manuscript and provided final approval for publication. The corresponding author had full access to the data and the final responsibility to submit for publication.
Funding The study was funded by Shanghai Shen Kang Hospital Development Centre, Clinical Research Plan of SHDC (grant number: SHDC2020CR6019); National Natural Science Foundation of China (No. 82002324).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.