Introduction
With a worldwide population aging, chronic musculoskeletal disorders have imposed a tremendous burden on society [
1]. Among them, osteoarthritis (OA) and sarcopenia (SP) are two common diseases. OA is characterized by the destruction and loss of articular cartilage as its main pathological feature, but all joint tissues and even extra-articular structures are involved in some form [
2,
3]. The two most common types in clinical practice are knee osteoarthritis (KOA) and hip osteoarthritis (HOA). Sarcopenia was originally defined in 1989 as the age-related loss of muscle mass, but gradually expanded to include muscle strength, muscle mass and physical performance [
4,
5]. As two age-related diseases, the prevalence of OA and SP is annually increasing [
6,
7]. Unfortunately, at present, we do not have an effective treatment for these two diseases [
8,
9]. Moreover, the coexistence of these two conditions: sarcopenic OA, which is frequently seen in clinical practice, could exacerbate the risk of falls and compromise the quality of life [
10,
11].
Preliminary clinical studies have shown a strong correlation between OA and SP and suggested that one condition can increase the possibility of developing the other, especially in the OA of the lower limbs [
12‐
14]. Initially, OA and SP are interconnected by biomechanical factors represented by muscle strength [
15‐
17]. With the in-depth study of bone-muscle crosstalk, the researchers found that the relationship between OA and SP can also be mediated biologically [
18‐
20]. The balance between lean mass and fat mass in muscle would lead to the dysregulation of multiple myokines. These myokines consequently affect cartilage gene expression in terms of formation and homeostasis [
21‐
23]. On the other hand, several molecules released by bone structures can also modulate muscles, such as Indian hedgehog and undercarboxylated osteocalcin [
24,
25]. However, none of the previous studies demonstrated the causality relationship between OA and SP. Evaluating the causality between OA and SP may provide new strategies for prevention, diagnosis, and treatment of OA, SP and sarcopenic OA.
Recent studies have extensively applied the Mendelian randomization (MR) approach to provide evidence for the causal relationships between exposures and outcomes [
26‐
28]. The identification of single nucleotide polymorphisms (SNPs) associated with common complex diseases has been greatly facilitated by genome-wide association studies (GWAS) [
29]. By applying SNPs as instrumental variables (IVs), the MR method can obtain a robust causal estimate independent of postnatal lifestyle or environmental factors [
30]. Several studies have identified causal factors leading to OA or SP [
26‐
28], but no MR study has examined causal relationships between OA and SP. Therefore, the aim of our study was to investigate the causal associations between OA of lower limbs and SP via a bi-directional MR approach. We assumed that SP and OA have a significant casual effect on each other.
Discussion
This study aimed to explore the causal relationships between SP and OA via a bi-directional two-sample MR approach. To the best of our knowledge, our MR study is the first to evaluate the bi-directional causal link between SP and OA comprehensively. Based on our results, there is a causal effect of SP on OA, while conversely, we did not observe a significant causal effect of OA on SP. Moreover, our findings indicated a causal relationship between SP and OA through the mediation of ALM.
Previous clinical studies have demonstrated the positive correlation between SP and OA. In a cross-sectional study, Suh et al. found that high fat mass and low lower extremity muscle mass were correlated with the presence and intensity of KOA [
17]. The similar results were also found in the HOA, which were featured by the proportionally higher fat mass, and lower lean body mass [
38,
39]. Furthermore, with better awareness of the importance of SP on OA, multiple long follow-up, large sample cohort studies were conducted to observe the relationship between them. In a cohort of healthy older population with no clinically diagnosed, symptomatic KOA and knee pain, ALM and GS was associated with the development of KOA and knee pain 5 years later, respectively [
40]. In another large longitudinal cohort study, the body composition of fat and muscle mass was also associated with KOA risk within a 60-month follow-up period [
41]. Meanwhile, compared to the isolated OA, the quality of life was more compromised and the likelihood of patients being referred for surgery was higher in the presence of SP [
14,
42]. However, there was still a lack of high-level evidence-based evidence for the causality of SP for OA, for instance, randomized controlled studies. Our MR study found a significant causal effect of SP on OA, which provided additional evidence to support causality between them.
A diagnosis of SP can be confirmed by the presence of both low muscle strength and low muscle quantity or quality, as one alone is not sufficient [
43]. Consistent with previous observational studies mentioned above, our study showed that four types of OA were causally associated with a lower ALM. However, OA is not causally affected by low GS, unlike what is expected. This may be due to several reasons. First, in the summary-level data of GS, the low GS was defined as grip strength < 30 kg for males and < 20 kg for females, which was recommended in 2010 by The European Working Group on Sarcopenia in Older People (EWGSOP) [
44]. However, EWGSOP2 updated the cutoff values of low GS in 2018: < 27 kg for males and < 16 kg for females [
43]. While in the Foundation for the National Institutes of Health Sarcopenia Project, the cutoff values of low GS are < 26 kg for males and < 16 kg for females [
45]. The assessment of the causal effect of low GS on OA may have been influenced by the differences in the criteria for low GS in SP. Furthermore, although higher muscle strength can better maintain joint stability to protect the joint, some clinical studies also indicated that greater muscle strength is not always protective. Chaisson et al. reported that men and women with higher GS were both associated with a greater risk of developing incident radiographic OA at hand [
46]. Similar results were also reported by Sharma et al. in KOA, in which they found greater quadriceps strength was associated with increased likelihood of OA progression in malaligned knees and lax knees [
47]. Further research is required to determine the impact of muscle strength on OA.
On the other hand, muscle composition changes are believed to be the primary contributor to OA in SP. Accumulating evidence suggested muscle as a paracrine and endocrine organ which can secrete a variety of myokines to modulate the bone, including irisin, insulin-like growth factor-1, myostatin, and interleukin 6 [
48]. The dysregulation of these myokines were proven to be relevant in the development and progress of OA [
49,
50]. Simultaneously, the increased fat mass (decreased lean mass) is another important factor for explaining the potential causal association between SP and OA [
51]. Fat tissue can secrete a variety of deleterious cytokines for both cartilage and muscle, including inflammatory cytokines, leptin, and adiponectin. The cumulative release of these cytokines eventually leads to systemic low-grade inflammation, which has also been considered the basis of OA [
23,
52]. Our study results suggest that ALM has a significant causal effect on OA. Similar findings were reported by Liu et al. who found a positive causal relationship between ALM and bone mineral density. However, there was no evidence of a causal association between low GS and bone mineral density [
53]. These findings suggest that changes in muscle composition may play a crucial role in the muscle–bone crosstalk mechanism.
In the reverse direction, although we found a nominally significant result for the relationship between HOA and GS, the rest results did not support the causal effect of OA on SP. Compared to studies on the influence of SP on OA, fewer studies have focused on the impact of OA on SP. Most recently, Francesco et al. conducted a systematic review and meta-analysis of the prevalence of SP in KOA [
54]. It is noteworthy, though, that they only included four cross-sectional studies in their analysis, even though their results indicated a greater prevalence of SP in patients with KOA than in non-KOA patients. Besides, from the perspective of bone-muscle crosstalk, there was limited research regarding the possible direct influence of chondrocytes on muscular cells in vitro and in animal research. Some researchers argued that atrophy of muscles in osteoarthritis would be caused more by the functional impairment caused by pain than by direct biomolecular factors inhibiting muscle development [
50]. Therefore, well-designed epidemiological, molecular mechanism and MR experiments are required to determine the causality of OA on SP in the future.
Previously, studies have tried to analyze the causal relationship between SP and OA, but no MR analysis has been conducted to investigate bi-directional causal link between SP and OA. As a result of the MR approach, we were able to avoid most confounding factors and reverse causality associated with traditional observational studies, which always bias these studies. In the present study, we selected IVs based strictly on the three hypotheses of the MR study, which makes our results more reliable. Furthermore, we adopted various methods as well as sensitivity tests to assess the causal relationship between SP and OA, which can generate more robust results. Nevertheless, there were some limitations in our study. First, the results from other MR methods, including MR-Egger and weighted median did not fully align with the IVW method in the MR analysis. However, based on the principle of method selection, IVW estimated results can be preferred if there is no pleiotropy present. Second, only summary statistics were collected, so it was not possible to assess the effect of age or gender separately. Moreover, the selected datasets were restricted to European ancestry, it is unknown if similar results would be obtained from other ancestries. Lastly, although confounding has been addressed in this study, the potential impact of third-party factors cannot be ruled out. Therefore, the results may not follow a linear pattern and require additional verification.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.