The relationship between serum lipid with sarcopenia: Results from the NHANES 2011 – 2018 and bidirectional Mendelian randomization study

Background: The relationship between serum lipids and sarcopenia remains unclear due to conflicting results in previous studies. Objective: To explore the associations and potential causality between serum lipids, including high-density li-poprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides (TG), and total cholesterol (TC), and sarcopenia. Methods: Data from the National Health and Nutrition Examination Survey (NHANES) were analysed using multivariable regression and restricted cubic splines (RCSs) to assess the associations between serum lipids and sarcopenia. Bidirectional Mendelian randomization (MR) was employed to investigate the causal relationships with sarcopenia-related traits such as appendicular lean mass (ALM), hand grip strength, and usual walking pace. Results: Serum HDL-C and TG levels were inversely associated with ALMBMI, with each 1-unit increase linked to a 0.13 % and 1.32 % decrease, respectively. Elevated TG, but not HDL-C, LDL-C, or TC levels, was significantly associated with an increased risk of sarcopenia (P for trend = 0.001). RCS analysis revealed a log-shaped dose-response relationship between TG and sarcopenia risk (P overall < 0.001, P non-linear < 0.001), with a cutoff value of 92.75 mg/dL. Genetically predicted HDL-C, LDL-C, and TG were associated with ALM. Conversely, ALM showed an inverse causal relationship with all four serum lipids. Additionally, genetically predicted usual walking pace influenced HDL-C and TG levels ( P < 0.001). Conclusion: The study reveals a nonlinear association between TG levels and sarcopenia risk, and a bidirectional association between lipid profiles and muscle mass, underscoring the need for further research to elucidate these mechanisms.


Introduction
Sarcopenia, initially defined by Rosenberg (1997) as the loss of muscle mass with aging, is a concept that has continued to evolve over time.The European Working Group on Sarcopenia in Older People (EWGSOP) has expanded this definition to include the loss of muscle strength, with or without a decline in muscle function (Cruz-Jentoft et al., 2019), which can be further exacerbated by factors such as inflammation, lack of exercise, malnutrition, and chronic diseases (Coletta and Phillips, 2023).In the context of global aging, sarcopenia has become an important public health challenge.Sarcopenia has been reported to predict adverse outcomes, including falls (Zhang et al., 2020), disability (Su et al., 2022), increased risk of osteoporosis (Petermann-Rocha et al., 2021), diabetes (Mesinovic et al., 2019), cardiovascular diseases (Xia et al., 2021), and even death (Cruz-Jentoft and Sayer, 2019).Currently, sarcopenia still lacks specific therapeutic approaches substantiated by robust evidence.Thus, investigating its risk factors would be advantageous for a more comprehensive understanding of its pathogenesis, as well as for identifying potential targets for prevention and treatment.
Skeletal muscle plays dual roles not only as an executor of motion but also as an endocrine organ intimately involved in hormone regulation.Sarcopenia is histologically characterized by a reduction in the number and size of muscle fibres and fat infiltration in the muscles (Wiedmer et al., 2021).Current research has identified a strong interplay between lipid metabolism and muscle function, yet the precise underlying mechanisms remain elusive (Al Saedi et al., 2022), suggesting the potential of lipids as biomarkers.Serum lipid analysis, a routine, inexpensive, and straightforward clinical test, is commonly employed for early cardiovascular and cerebrovascular disease screening.Clinically, several studies have endorsed an association between lipids and sarcopenia.For instance, a meta-analysis (Bi et al., 2024) demonstrated a correlation between dyslipidaemia and sarcopenia in older adults.Ana et al. (Coto Montes et al., 2017) identified serum lipid levels, dyslipidaemia, and the use of lipid-lowering drugs as independent risk factors for sarcopenia.However, the evidence is not universally consistent.Du et al.'s meta-analysis (Du et al., 2018) indicated that only triglycerides, high-density lipoprotein cholesterol, and total cholesterol were associated with sarcopenia and that there are potential sex-specific differences.In addition, observational studies are subject to several limitations, such as insufficient adjustment for confounding factors, small sample sizes, and reverse causation.Therefore, it is difficult to provide definitive evidence of an independent association between serum lipids and sarcopenia.
The NHANES is a cross-sectional, nationally representative survey designed to assess the health and nutritional status of nonhospitalized U. S. residents and adopts a stratified multistage sampling design.Therefore, the NHANES can provide high-quality, large-sample, and nationally representative data to evaluate the correlation between the level of serum lipids and sarcopenia risk.Mendelian randomization (MR), a robust and effective method that uses genetic variants to infer causal relationships between exposure and outcomes, assumes that genetic variation, specifically single-nucleotide polymorphisms (SNPs), is randomly assigned at conception (Sekula et al., 2016).This intrinsic nature of genetic variation ensures that MR analysis minimizes susceptibility to confounding factors and reverses causation bias, which are common in observational studies (Davies et al., 2018).Therefore, by combining a large-scale observational study using NHANES data and a two-sample bidirectional MR analysis, we aimed to explore the associations and potential causality between four commonly assessed clinical lipid measurements, including high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), triglycerides (TG), and total cholesterol (TC), and sarcopenia.The research design is depicted in Fig. 1.

Data sources
In phase 1, participants who underwent complete blood laboratory Fig. 1.Study design for relationship between serum lipid with sarcopenia.
In phase 2, GWAS summary data regarding HDL-C, LDL-C, and TG were obtained in one meta-analysis, which included 393,193 to 441,016 individuals with a mean age of 56.9 years (range 39-73 years), 54.2 % of whom were female.The mean (standard deviation) lipid concentrations were 3.57 (0.87) mmol/L for LDL cholesterol and 1.45 (0.38) mmol/L for HDL cholesterol, and the median triglyceride concentration was 1.50 (IQR = 1.11) mmol/L.For TC, data were obtained from the MRC Integrated Epidemiology Unit (IEU) through its open GWAS database (https://gwas.mrcieu.ac.uk/).Ultimately, genetic associations with ALM were obtained from 450243 UK Biobank participants.ALM was measured by determining the sum of the fat-free mass of the arms and legs via bioelectrical impedance analysis.Low grip strength was derived from a meta-analysis of GWAS summary data conducted on 256,523 Europeans aged 60 years and older across 22 cohorts (n = 48,596 patients).Low grip strength was defined according to the EWGSOP criteria (male: <30 kg, female: <20 kg), given all included European races.Summary statistics for genetic polymorphisms associated with walking pace were extracted from corresponding GWASs of up to 331,285 Europeans.

Data collection for observational study
The diagnostic criteria for sarcopenia recommended by the Foundation for the National Institutes of Health (FNIH) (Studenski et al., 2014), given that the population under investigation consisted of American citizens.Specifically, sarcopenia was assessed by calculating the sum of BMI-corrected Appendicular Lean Mass (ALM), known as ALMBMI.The cutoff values for the diagnosis of sarcopenia were set at <0.789 for males and <0.512 for females.In the NHANES 2011-2018, all participants underwent DEXA scanning to obtain full-body data.Pregnant individuals or those weighing >136 kg or measuring >196 cm were excluded to ensure safety.This methodology has been documented in similar research and is acknowledged for its effectiveness in sarcopenia assessment (Chen et al., 2022;S. Li et al., 2023).
Regarding serum lipids, all blood samples were collected, processed and stored at the Mobile Examination Center before being transported to a remote laboratory for biological analysis.Among these tests, only individuals who fasted for 8.5 to 24 h and whose blood was drawn in the morning underwent LDL-C testing.
Covariables in this phase included age, sex (male and female), race (non-Hispanic white, non-Hispanic black, Mexican American, other), education (below high school, high school graduate, above high school), body mass index (BMI) categories (≤24.9, 25-29.9,≥30), smoking status (never, ever, current), alcohol consumption (never, ever, current), and income groups using the ratio of family income to poverty (PIR), low-income group (<1.3), middle-income group (1.3-3.5), and highincome group (>3.5).The leading contributors to chronic disease burden in older people were assessed using the self-reported question "Have you ever been told by a doctor or other health professional that you had …".For this analysis, we identified the following disorders: heart failure, coronary heart disease, diabetes mellitus, cancer, and arthritis.Each variable was dichotomized (yes/no).Depression was diagnosed by the Patient Health Questionnaire (score ≥ 10).Statin use (including atorvastatin, cerivastatin, fluvastatin, lovastatin, pitavastatin, pravastatin, rosuvastatin, and simvastatin) and steroid use (cortisone) were defined by interviewer pill bottle review.

Instrumental variable selection for bidirectional Mendelian randomization
The following screening steps were performed to select instrumental variables (IVs) to meet the three MR hypotheses (Emdin et al., 2017).First, the independent SNPs associated with each outcome were selected (P < 5 × 10 − 8 , LD clumping r 2 threshold <0.001 and distance = 10,000 kb).Second, to ensure the accuracy of the results, palindromic and ambiguous SNPs and those with intermediate allele frequencies were deleted.Third, to further avoid the bias introduced by pleiotropic outliers, MR pleiotropy residual sum and outlier (MR-PRESSO) and MR-Radial methods were used to detect and remove potential outliers.Fourth, SNPs exhibiting weak instrument bias (F value <10) were discarded.The strength of each instrument was measured by calculating the F-statistic: . Finally, the proportion of variance in the phenotypic variance explained (PVE) was calculated using the following equation: PVE (SNP)=2*MAF*(1 − MAF)*beta 2.

Statistical analysis
In phases 1, complex weighted analysis was conducted on the extracted data using the "survey" package in accordance with CDC guidelines.Missing covariables were imputed using chained multiple imputation with the "jomo" package, following previous literature (Quartagno et al., 2019).To ensure independence among the generated imputed datasets, a Gibbs sampling process was used with 1000 initial iterations followed by 10,000 updates to generate five imputed datasets.Independent samples t-tests were used to compare continuous variables, and Rao & Scott adjusted χ2 tests were used to compare categorical variables for intergroup differences.Blood lipid levels were divided into four groups based on quartiles, with quartile 1 (Q1) serving as a reference.A weighted multivariable logistic regression model was used to analyse odds ratios (ORs) and their corresponding 95 % confidence intervals (CIs) between blood lipids and sarcopenia.Restricted cubic splines (RCSs) fitted dose-response relationships between blood lipids and sarcopenia.The selection of the number and location of knots in the RCS was guided by the Akaike information criterion (AIC) to strike a balance between optimal fit and overfitting (Johannesen et al., 2020).
The significance level was set at α = 0.05 for two-tailed tests.
In phase 2, all analyses were carried out using "TwoSampleMR", "MRPRESSO", and "RadialMR" packages.The sensitivity analysis consisted of three parts.First, heterogeneity was detected using Cochran's Q statistic and is indicated by the value of I 2 .In cases where there was no significant heterogeneity (P > 0.05), a fixed-effects model was applied; otherwise, a multiplicative random-effects model was utilized.Second, pleiotropy was assessed by the MR-Egger intercept test and MR pleiotropy residual sum and outlier (MR-PRESSO) method.A MR-Egger regression intercept term that exhibits a significant departure from zero can be considered a dependable indicator for estimating causal effects influenced by horizontal pleiotropy, and a p value <0.05 in the MR-PRESSO method could be considered indicative of a pleiotropic effect.Third, the leave-one-out analysis was further employed for sensitivity analysis of the statistically significant MR results, and any results that failed to meet the criteria were excluded.We used three Mendelian randomization (MR) methods: inverse-variance weighting (IVW), MR-Egger and weighted median (WM).IVW assumes that all instrumental variables are valid, while MR-Egger can detect and correct for potential bias.The weighted median provides robustness against invalid instruments.We chose IVW as our primary approach but also considered outcomes from the other two methods to verify the results.Bonferroni correction was applied for multiple comparisons, with a p value threshold of 0.05/2*(3*4) ≈ 0.002.P values between 0.002 and 0.05 were considered to indicate nominal significance.

Characteristics of the NHANES population
A total of 4636 subjects were included, representing 54,645,402 (4,104,338 cases) nonhospitalized residents.The baseline characteristics are presented in Table 1.Among these patients, 415 (7.51 % of whom were weighted) fulfilled the diagnostic criteria for sarcopenia.Notably, compared with the control group, the sarcopenia group exhibited higher LDL-C, TG and TC levels and lower HDL-C levels (P < 0.05), details were shown in Table 1.

Associations between serum lipids and sarcopenia according to the observational study
The associations between serum lipid levels and BMI-corrected ALM, as well as the incidence of sarcopenia after adjusting for all covariables, were presented in Table 2.The results indicate that each 1-unit increase in HDL-C and TG are associated with decreases of 0.13 % and 1.32 %, respectively, in ALM BMI .Moreover, compared to the Q1 group, only an increase in TG levels was associated with a greater risk of sarcopenia (P for trend =0.001).No significant correlations were detected between HDL-C, LDL-C, or TC levels and the risk of sarcopenia (P > 0.05).
The nonlinear relationship between lipid levels and the risk of sarcopenia was analysed using RCS after adjustment for all covariates, as shown in Fig. 2. The results showed a non-linear log-shaped doseresponse relationship between TG and the risk of sarcopenia (P of overall <0.001, P of non-liner <0.001).Within the physiological range, when TG > 92.75 mg/dL (1.047 mmol/L), OR > 1 was a risk factor for sarcopenia.

Causal effects of serum lipids on sarcopenia according to three MR analyses
A total of 7 to 520 strong associated IVs were identified using MR-Radia after excluding outliers.Additionally, all IVs had F-statistics exceeding 10, indicating no weak instrument bias.These IVs collectively accounted for variations in phenotypes ranging from 0.05 % to 9 %.The detailed results are presented in Table 3. Sensitivity analysis revealed no heterogeneity or pleiotropy.Moreover, leave-one-out tests indicated that the associations were robust (Supplementary files).

Discussion
The relationship between lipid metabolism and sarcopenia has garnered significant attention in the past decade.This study analysed the correlation between four serum lipids and sarcopenia, determining the threshold of their effect through RCS analysis; furthermore, by leveraging the genetic characteristics of gene variation, we also explored the causal relationships between serum lipids and the three major traits of sarcopenia, thereby providing more comprehensive insights into determining the diagnostic value of lipid indices in sarcopenia.
In alignment with previous studies (Huang et al., 2024), our findings presented in Table 1 confirm the prevalence of sarcopenia at approximately 7.51 %.This statistic underscores the significance of sarcopenia as a public health concern.Further analysis, as delineated in Table 2, revealed that TG (beta = − 0.0132) and HDL-C (beta = 0.0013) were negatively and positively associated with ALM BMI , respectively, which is in line with the results of numerous other studies (Wang et al., 2020;Yang et al., 2023).Although the exact biological mechanism by which TG and HDL-C affect muscle health is still unclear, TG and HDL-C have a certain biological correlation, and high TG levels and low HDL-C levels often coexist in clinical practice.Studies have shown that TG/HDL-C is closely related to insulin resistance (Giannini et al., 2011;Nur Zati Iwani et al., 2022), and it is recommended to be used as a predictor of cardiovascular and cerebrovascular diseases (Che et al., 2023;Sato et al., 2022), diabetes (Liu et al., 2022), and periodontitis (Lee et al., 2022).Therefore, a few scholars have proposed that the TG/HDL-C ratio could also be employed as a clinical indicator for assessing sarcopenia status (Chung et al., 2016;Fu et al., 2023).
Our further analysis of Fig. 2 delineates a potential nonlinear dose-response relationship between TG levels and sarcopenia risk.This explanation addresses the inconsistencies previously observed in studies regarding the association between TG and the risk of sarcopenia (Hwang and Park, 2022;Park et al., 2022).This means that TG may play a direct role in the occurrence and development of sarcopenia and, compared with low TG levels (< 92.75 mg/dL), the risk of sarcopenia may increase with increasing TG levels.However, weighted logistic regression showed that HDL-C, LDL-C, and TC were not significantly associated with sarcopenia risk (P > 0.05).The RCS results showed that although the nonlinear relationship was not significant (P non-linear > 0.05), there may be an overall association between these three lipids and the development of sarcopenia (all P < 0.05).This may be related to the complex biological characteristics of blood lipids, and this finding provides a direction for further research.
Consistent with the observational findings, our forward MR analysis presented in Table 3 and Fig. 3A further revealed that genetic-level lipids, which are significant risk factors for sarcopenia, could be used for early screening.In addition, we have also demonstrated that lipids primarily affect muscle mass (ALM) rather than functional performance (grip strength and gait speed); therefore, future attention should focus on changes in the muscles themselves.This finding aligns with previous research findings that lipids and fatty acids may lead to muscle atrophy (Lipina and Hundal, 2017), potentially through the regulation of the PKB/Akt (De Larichaudy et al., 2012;Mahfouz et al., 2014) and/or mTORC1 pathways (Shad et al., 2015;You et al., 2014).For instance, an animal study by Turpin et al. (2009) revealed that diacylglycerol (a lipid substitute) increased in the gastrocnemius muscle of high-fat diet-fed mice, along with an increase in caspase-3 activity.
Our reverse causal analysis presented in Table 4 and Fig. 3B revealed that ALM functions as a protective factor for the four lipid measures.On the one hand, this novel finding offers a potential explanation for the observed phenomenon in observational studies where all lipid measures, except for TG, exhibited P values >0.05 for nonlinear associations but <0.001 in the RCS analysis.On the other hand, the bidirectional causal relationship between dyslipidaemia and the risk of sarcopenia, particularly muscle mass decline, may be partly explained by the 'Metabaging Cycle' proposed by Li et al. (Li et al., 2022), which, posits that the accumulation of adipose tissue within skeletal muscle leads to mitochondrial dysfunction, the promotion of reactive oxygen species (ROS) production, and subsequent insulin resistance.These changes may also contribute to adipose tissue atrophy, thereby fostering the development of chronic inflammation and perpetuating a vicious cycle involving dysregulated lipid metabolism, insulin resistance, and systemic inflammation.Furthermore, our reverse MR study revealed that the usual walking pace is a determinant of lipid levels, especially HDL-C and TG levels, highlighting the potential role of physical activity in lipid management and sarcopenia prevention.Walking is a simple and convenient form of exercise, and it is one of the most common and effective forms of physical activity for elderly people.Wang and Xie (2023) also reported that maintaining the habit of walking for a long time can also contribute to the health benefits of pre-elderly women.A randomized controlled study by Kitajima et al. (2023) revealed that 60 min of rapid walking exercise per week for 20 weeks could effectively improve lipid metabolism in elderly patients with diabetes (TG: Cohen's d = − 0.55).As a result, existing research once again highlights the potential role of physical activity in managing lipid levels and preventing sarcopenia.
However, there are certain limitations that should be acknowledged.First, while the observational study suggested a nonlinear relationship between TG levels and sarcopenia risk, MR analysis revealed only linear relationships.Thus, further exploration is needed.Second, caution must be exercised regarding potential confounding effects arising from population heterogeneity.The observational data encompassed diverse ethnic populations in the U.S., whereas the MR data were primarily from individuals of European ancestry.To reduce confounding effects due to Fig. 2. Restricted cubic spline plots between serum lipids and the risk of sarcopenia.Adjusted for age, sex, race, marital status, education level, income group, BMI, alcohol consumption, smoking status, chronic diseases, depression, statin use, and steroids use.
racial differences, we employed different sarcopenia diagnostic criteria based on racial groups.However, physiological and environmental differences across various ethnicities could still influence the generalizability of our findings.Future research needs to involve more diverse populations to verify the accuracy and applicability of the results.Lastly, while the study endeavored to control for potential confounders, the inherent limitations of the available data may mean that not all factors were adjusted for.
Despite certain limitations, our study presents significant value in elucidating the relationship between blood lipids and sarcopenia, offering insights that serve as a reference for understanding clinical thresholds crucial to maintaining muscle health.Our methodological strengths include a substantial sample size, which enhances the statistical power and generalizability of our findings.Furthermore, by examining three distinct sarcopenia-related traits, as illustrated in Fig. 1, we adopt a multidimensional approach that allows for a more nuanced exploration of the condition.Importantly, MR analysis reduces potential confounding biases, providing a more comprehensive understanding of observational research findings and thus reinforcing the reliability and validity of our conclusions.These insights contribute to the broader discourse on muscle health and can potentially guide future research and inform clinical practices aimed at preventing or managing sarcopenia.

Conclusion
In conclusion, our observational study indicated a strong association between TG and sarcopenia.Furthermore, our MR study indicated a bidirectional causal link between serum lipids and genetically predicted ALM.Therefore, based on the present findings, we underscore the importance of early screening and intervention for dyslipidaemia and sarcopenia, especially decreased muscle mass, in high-risk populations.

Fig. 3 .
Fig. 3. Bidirectional Mendelian randomization analysis of serum lipid and sarcopenia traits.(A) Serum lipid serves as the exposure and three sarcopenia traits act as the outcomes; (B) three sarcopenia traits serve as the exposures and serum lipid act as the outcomes.The analysis is performed using three different MR methods, with the results from the Inverse Variance Weighted (IVW) method highlighted in red.

Table 1
Baseline characteristics of participants by sarcopenia status.

Table 2
Associations between the serum lipid level and BMI-corrected ALM and the incidence of sarcopenia.Beta coefficients are presented for the analyses of continuous outcomes (i.e., BMI-corrected ALM).b ORs are presented for the analyses of binary outcomes (i.e., sarcopenia).Adjusted for age, sex, race, marital status, education level, income group, BMI, alcohol consumption, smoking status, chronic diseases, depression, statin use, and steroids use. a

Table 3
Causal effects of serum lipids on sarcopenia according to three MR analyses after outlier deletion.

Table 4
Causal effect of sarcopenia on serum lipids according to three MR analyses after outlier deletion.