Markers of inflammation and their association with muscle strength and mass: A systematic review and meta-analysis

BACKGROUND
Chronic inflammation has been associated with sarcopenia and its components skeletal muscle strength and muscle mass. The aim of this systematic review and meta-analysis was to determine the relationship between systemic inflammation, muscle strength and/or muscle mass in adults.


METHODS
An electronic search using keywords such as 'acute phase proteins, cytokines and sarcopenia, muscle mass, muscle strength' was conducted via Pubmed, Web of Science and Embase from inception until the 30th of June 2020. A meta-analysis using correlation data was performed to determine the overall relationship between inflammation and muscle strength and muscle mass in adults.


RESULTS
Overall, 168 articles; 149 cross-sectional articles (n = 76,899 participants, 47.0 % male) and 19 longitudinal articles (n = 12,295 participants, 31.9 % male) met inclusion criteria. Independent of disease state, higher levels of C reactive protein (CRP), Interleukin (IL)-6 and Tumor necrosis factor (TNF)α were associated with lower handgrip and knee extension strength (CRP; r = -0.10, p < 0.001, IL-6; r = -0.13, p < 0.001, TNFα; r = -0.08, p < 0.001 and CRP; r = -0.18, p < 0.001, IL-6; r = -0.11, p < 0.001, TNFα; r = -0.13, p < 0.001 respectively) and muscle mass (CRP; r = -0.12, p < 0.001, IL-6; r = -0.09, p < 0.001, TNFα; r = -0.15, p < 0.001). Furthermore, higher levels of systemic inflammatory markers appeared to be associated with lower muscle strength and muscle mass over time.


CONCLUSION
Higher levels of circulating inflammatory markers are significantly associated with lower skeletal muscle strength and muscle mass.


Introduction
Skeletal muscle plays an integral role in maintaining homeostasis across organ systems. Skeletal muscle is plastic, changing dynamically in response to physical activity, load, injury, illness and ageing (Haddad et al., 2005). The age-related loss of skeletal muscle strength, muscle mass and physical performance (sarcopenia), has been associated with falls and fractures in older populations (Yeung et al., 2019a), and remains a largely undiagnosed condition (Fielding et al., 2011;Reijnierse et al., 2017;Yeung et al., 2019b). Beyond ageing, sarcopenia is associated with age-related diseases such as dementia, chronic obstructive pulmonary disease and cardiovascular disease (Pacifico et al., 2020). In older adults, several of these diseases coincide with decline in muscle mass and whether this is caused by ageing or disease is largely unknown (Cruz-Jentoft et al., 2019). However, a common feature underlying both conditions is inflammation (Chhetri et al., 2018;Chung et al., 2019;Souza et al., 2017).
Chronic inflammation, characterised by higher systemic cytokine and acute phase protein circulation (Ferrucci et al., 2005;Pedersen et al., 2003), is not only linked to ageing 'inflammaging' (Franceschi et al., 2000) but also muscle mass loss (Ali and Garcia, 2014). Tumor necrosis factor α (TNFα) released from diseased tissues has been shown to exert endocrine effects on skeletal muscle (Powers et al., 2016).
In vitro studies have shown that TNFα is a key endocrine stimulus for contractile dysfunction in chronic inflammation and that the muscle derived reactive oxygen species (ROS) and nitric oxide (NO) participate in depressing specific force of muscle fibre, which can lead to muscle atrophy (Li et al., 2000). Furthermore Interleukin (IL)-6, a key cytokine involved in low-grade chronic inflammation, has been shown to facilitate muscle atrophy via blunting muscle anabolism and energy homeostasis (Haddad et al., 2005). However, despite significant research within the area the extent of the association between systemic markers of inflammation such as, TNFα, IL-6 and clinical markers of skeletal muscle strength and muscle mass is unknown, particularly in populations with age-related diseases (Can et al., 2017;Kwak et al., 2018).
A systematic review and meta-analysis of the literature was undertaken to identify if systemic inflammatory markers are associated with lower skeletal muscle strength and muscle mass in adults sub grouped by inflammatory marker, population and sex.

Search strategy
This systematic review was conducted following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Liberati et al., 2009). The systematic search of the literature was conducted using the electronic databases: PubMed, EMBASE and Web of Science from inception until 30/06/2020. The search included search terms relating to inflammatory markers ("cytokines", "acute phase proteins") and muscle measures ("sarcopenia", "muscle mass", "muscle strength", "hand strength"). The complete search strategy for this systematic review can be found in Supplementary   Table 1.

Study selection
Publications returned from the databases were imported into Endnote, and duplicates removed. Titles and abstracts of selected articles were screened by two independent reviewers (LT, JK or CT) using Covidence software (Covidence systematic review software, Veritas Health Innovation, Melbourne, Australia). Articles were included if 1) study cohort mean or median age was ≥18 years, 2) data for at least one inflammatory marker was available, and 3) data for at least one muscle measure of either muscle strength or muscle mass was available. Studies were excluded if 1) in vitro or animal models were utilised; 2) participants had a genetic disease, such as progeroid disease or muscular dystrophies; 3) article not written in English and 4) conference abstracts, reviews, editorials, letter to the editor and case reports. Two independent reviewers (LT, JK or CT) screened the full-text articles using the above criteria and the additional criteria; 5) no statistical test assessing an association between inflammatory markers and muscle strength and/ or muscle mass was used. Disagreement of the eligibility of an article was resolved by the involvement of a third reviewer (CT).

Data extraction
Data extraction was performed by two independent reviewers (CT, LT). The following data were extracted: author, year of publication, study design (cross-sectional (C), longitudinal (L)), study population, time to follow up (L studies), sample size, age of participants, percentage of males, health status, inflammatory marker, muscle parameter and technique used to measure muscle strength or muscle mass.

Statistical analysis
The meta-analysis was performed using Comprehensive Meta-Analysis (CMA) Software (Version 3.0, Biostat, Englewood, New Jersey, USA). To determine the overall association between inflammatory marker and muscle mass and strength, data extracted from articles was entered using one of two formats. Where the article provided a correlation coefficient (r) for inflammatory marker and muscle parameter; r, sample size and direction of the association was used, when r was not provided, but the sample size, p value and the direction of the association was given, r was calculated. Z-scores were calculated from the correlation coefficients using Fisher's method and pooled under a random effects model. This method was used to reduce the risk of unknown factors responsible for variability even under homogeneity. Heterogeneity of the results was investigated using the I-squared test. In addition to the overall analysis, the meta-analysis was sub grouped by inflammatory marker, population and sex. Begg's funnel plot and Eggers Regression intercept were used to determine publication bias. Overall 168 articles were included in this systematic review. Table 1 provides a comprehensive overview of included articles, categorized by the study design; C = 149 and L = 19 articles. For a detailed description of each article included in this review refer to Supplementary Table 2. The total number of included participants for cross-sectional studies was 76,899 of which 47.0 % were male. Conversely, 12,295 participants, 31.9 % male, have been included in longitudinal studies; with follow-up ranging from 1 month to 16 years. Overall, community dwelling cohorts assessing the association between systemic inflammation and muscle strength and/or muscle mass were reported most frequently (C = 41/ 149, L = 9/19 articles). The association between inflammation and muscle strength and/or muscle mass have been assessed with 31 acute phase proteins and cytokines (Table 2). Muscle strength was reported more frequently than muscle mass (C = 121/149 vs. 82/149; L = 15/19 vs. 9/19 articles respectively). Handgrip strength was the most frequently used measure (C = 90/149, L = 9/19 articles) to determine muscle strength. To determine muscle mass, dual-energy X-ray absorptiometry (C = 35/149, L = 6/19 articles) and bio impedance analysis (C = 23/149, L = 2/19 articles) were the most frequently utilised techniques. A third of included articles (C = 53/149, L = 5/19) measured both muscle strength and muscle mass within the studied populations.

Cross-sectional analysis of inflammation and muscle strength
CRP, IL-6 and TNFα were the most frequently used markers to assess the relationship between inflammation with muscle strength (Supplementary Fig. 1). Fig. 2 describes the meta-analysis for the association between CRP and muscle strength sub grouped by population and sex and stratified by muscle strength measurement. CRP was inversely associated with handgrip strength (r = − 0.10, p < 0.001) and knee extension strength (r = − 0.18, p < 0.001), which was independent of sex. Higher IL-6 was associated with lower handgrip strength (r = − 0.13, p < 0.001) and knee extension strength (r = − 0.11, p < 0.001). IL-6 had a slightly stronger association with handgrip strength in some populations compared to others (menopause: r = − 0.18, p < 0.01, kidney disease: r = − 0.17, p < 0.001) and this relationship was strongest in cardiovascular disease (CVD) populations for IL-6 with knee extension strength (r = − 0.39, p < 0.001). When sub grouped by sex, higher IL-6 and lower handgrip strength (r = − 0.15, p < 0.001) and knee extension strength (r = − 0.31, p < 0.001) were observed in males (Fig. 3). TNFα was inversely associated with handgrip strength (r = − 0.08, p < 0.001) and knee extension strength (r = − 0.13, p = 0.01), which was dependent on the population. Cancer cohorts had the highest correlation of TNFα with handgrip strength (r = − 0.34, p < 0.001), and CVD cohorts had the highest correlation with TNFα and knee extension strength (r = − 0.39, p < 0.001). When stratifying by sex, higher TNFα and low knee extension strength was strongest in males (r = − 0.32, p < 0.001) (Fig. 4). Higher levels of Growth Differentiation Factor (GDF)15 and IL-8 were significantly associated with lower handgrip strength (r = − 0.32, p < 0.001 and r = − 0.33, p < 0.001 respectively) and higher levels of GDF15 was associated with lower knee extension strength (r = − 0.31, p < 0.001) ( Supplementary Fig. 2).

Publication bias
Publication bias was assessed for all outcomes via visually inspecting the asymmetry of the funnel plot ( Supplementary Fig. 5). The funnel plot including all data points demonstrated asymmetry. This, combined with Egger's regression intercept analysis, indicated publication bias towards positive findings (p = 0.001).

Discussion
Higher levels of circulating inflammatory cytokines are associated with lower skeletal muscle strength and muscle mass. While previous systematic reviews have demonstrated a relationship between inflammation and sarcopenia, this is the first review to directly analyse the relationship between systemic inflammation and muscle strength and muscle mass independently. The association between inflammation and muscle strength and mass is most often investigated using CRP, IL-6 and TNFα. However, this review has identified that the strength of this linear association maybe variable and dependent upon populations and sex.
CRP, a marker of both acute and chronic phase inflammation, is often used in clinical practice to evaluate the inflammatory state of an individual. CRP is known to be heightened in response to infection and injury. Furthermore, high levels of CRP are associated with a higher risk of the development of chronic diseases such as, type 2 diabetes, cardiovascular disease and sarcopenia, a combination of low muscle mass, strength and physical performance (Bano et al., 2017;Cesari et al., 2003;Freeman et al., 2002;Koenig et al., 2008). The wide and easily accessible availability of CRP may explain why it has been utilised most often by this research field to assess the relationship between systemic inflammation and lower muscle strength or muscle mass. Interestingly, despite observing a significant inverse association between CRP levels and muscle strength and muscle mass, it is not known if CRP directly acts upon muscle to cause muscle atrophy.
This analysis highlights a stronger association between CRP and muscle mass than IL-6 and muscle mass in community dwelling populations. Elevated CRP levels have also been associated with sarcopenia, but not with IL-6 and TNFα (Bano et al., 2017). It is possible that the observed association reflects a higher level of inflammation within the body however, when other chronic diseases are involved other inflammatory cytokines such as IL-6, may directly act upon muscle (Londhe and Guttridge 2015).
IL-6 has long been recognised as a key cytokine that promotes muscle anabolism or catabolism depending upon the in vivo environment (Belizário et al., 2016). As part of the inflammatory secretome, IL-6 is secreted by immune cells in response to tissue infection and/or tissue damage. Furthermore, chronic low-grade inflammation is associated with prolonged exposure to IL-6 signalling. IL-6 is also a known myokine, produced by skeletal muscle and secreted from active skeletal muscle during exercise and is likely to improve glucose tolerance and insulin sensitivity post exercise. As such a temporal higher level of IL-6 can be beneficial (Pedersen and Febbraio, 2006). However, prolonged exposure to IL-6 has been shown to facilitate muscle atrophy by blunting muscle anabolism and energy homeostasis and may also directly mediate muscle catabolism (Belizário et al., 2016).
In this analysis we observed a stronger association of IL-6 and lower muscle mass in populations with age-related diseases such as cardiovascular disease. Furthermore, this relationship was strongest in men. A meta-analysis investigating sex effects on the relationship between IL-6 and handgrip strength also found a stronger negative correlation between plasma IL-6 and handgrip strength in men compared to women (Mikó et al., 2018). The impact sex has on the relationship between inflammation and muscle is important to consider (Anderson et al., 2017;Gubbels Bupp, 2015) and may differ depending on the cytokine. Furthermore, it has previously been shown that the sole action of IL-6 is not enough to induce muscle wasting rather, the catabolic effect of IL-6 is dependent on the synergistic interaction with other factors mediating the inflammatory response (Belizário et al., 2016). This may explain why a significant but overall weak relationship between IL-6 and muscle strength was observed within community dwelling cohorts. The higher IL-6 level associated with ageing may not be enough to induce the catabolic effect required for muscle wasting. However, the concomitant activity of other cytokines such as TNFα may create the physiological environment required that ultimately leads to muscle atrophy.
TNFα released from diseased tissues has been shown to exert CRP, n (%) 89 (60) 12 (63)        endocrine effects on skeletal muscle (Powers et al., 2016). TNFα is a pro-inflammatory cytokine that directly acts upon skeletal muscle to depress specific force. This decline in specific force has been shown through various in vitro and in vivo experiments using exogenous TNF (Alloatti et al., 2000), systemic administration of TNF in dogs (Wilcox et al., 1994) and overexpression of TNF in mice (Wolkorte et al., 2015). As such, it is not surprising that higher systemic TNFα levels are associated with lower muscle strength and muscle mass. TNF signalling is triggered by ligand/receptor interactions at the cells surface. Skeletal muscle fibres constitutively express two TNF receptors, TNFR1 and TNFR2 (Powers et al., 2016). Interestingly, in vitro studies have demonstrated that muscle responses to TNF are triggered by activation of TNFR1 rather than TNFR2 (Hardin et al., 2008). However, in this analysis a stronger association was observed between systemic levels of TNFR2 and lower muscle strength/mass compared to TNFR1. Furthermore, a sex difference may exist with a stronger negative correlation between TNFα and muscle strength and mass observed in men. Therefore, the interplay between TNFα and IL-6 and its impact on muscle maybe sex specific. However, the male populations included in this analysis also often had chronic diseases. Further analysis is required to untangle the relationship interplay between chronic disease, sex, and muscle.
In addition to CRP, IL-6 and TNFα, other inflammatory markers investigated in this review showed an inverse relationship with muscle strength and muscle mass such as GDF15. GDF15 is a cytokine that has been associated with ageing (Fujita et al., 2016). However, whether GDF15 is beneficial or detrimental to skeletal muscle is contentious. In this analysis higher levels of GDF15 was associated with lower muscle strength and mass. GDF15 is able to induce muscle fibre apoptosis via phosphorylation of STAT3 (Tang et al., 2019) and this may explain why higher levels of GDF15 are associated with lower muscle strength and mass, however it is important to note that the analysis performed here does not demonstrate causality.
There appears to be sufficient evidence to conclude that higher systemic inflammation is associated with lower muscle strength and muscle mass in humans. Furthermore, longitudinal studies also confirm this relationship exists over time, but it is important to note that some of these associations while significant were weak. The cross-sectional association of single inflammatory markers with muscle strengths and mass and the less consistent longitudinal associations, questions the temporal relationship between inflammatory markers and muscle measures. Muscle strength and mass might decline prior to an increase in systemic inflammation. Furthermore, skeletal muscle can actively alter the pro-and anti-inflammatory immune system, regulating innate and adaptive immune responses. Chronic low-grade systemic inflammation may occur as a direct result of low muscle mass (Looijaard et al., 2020). However, the stronger relationship between higher inflammation and lower muscle mass reported here within specific populations such as, cardiovascular disease, would suggest an alternate theory. As such other pathophysiology associated with low muscle strength and mass might also contribute to high systemic inflammation.
This review is not without limitations. The overall meta-analysis for each cytokine demonstrated weak correlation coefficient that were statistically significant. Often when an effect size is small but statistically significant it is due to a large sample size. Given the combined population of the cross-sectional studies is 76,899 the impact of sample size on this analysis cannot be discounted. However, it is important to note that not all studies included in the systematic review could be included in the meta-analysis. Furthermore, this analysis highlights that a linear approach to examine the association between systemic inflammation and muscle may not reflect the true relationship between the two parameters, indeed some publications have described a quadratic relationship between systemic inflammation and muscle strength decline (Tay et al., 2019;Rosenberg et al., 2019). It was not possible to stratify the overall analysis by age. Higher chronological age is known to impact both chronic inflammation and muscle strength and mass (Beenakker et al., 2010). The association between inflammation and lower muscle strength or mass reported here could be stronger in older populations. The search strategy was designed to be broad and inclusive; all reported markers of inflammation were included; this has led to some articles contributing multiple associations within the meta-analysis. Furthermore, some articles report using the same cohort e.g. InCHIANTI (Bartali et al., 2012;Stenholm et al., 2010) but used different techniques to measure muscle strength such as handgrip strength (Stenholm et al., 2010) and knee extension (Bartali et al., 2012). This is likely to have a positive bias on the power and significance of the overall meta-analysis. Finally, when interpreting the findings reported here, publication and reporting bias must also be considered.
The findings of this review conclusively demonstrate a significant association between higher levels of circulating inflammatory markers and lower skeletal muscle strength and muscle mass and the strength of this relationship differs depending upon population and sex.

Declaration of Competing Interest
None declared.