Association between elevated expression of GDF15/GFRAL and sarcopenia risk

Summary

Glial cell line-derived neurotrophic factor family receptor α-like (GFRAL) is the exclusive receptor for growth differentiation factor 15 (GDF15) and plays distinct roles in various diseases. However, its impact on sarcopenia remains poorly understood. This study, including White British and Chinese participants, demonstrated that elevated serum levels of GDF15 and its receptor GFRAL are associated with an increased risk of sarcopenia, particularly among individuals with low physical performance. While GDF15 exerts effects across the whole population, GFRAL appears to be more detrimental in women. Complementary single-cell RNA sequencing and immunohistochemical analyses in animal models further support the involvement of the GDF15–GFRAL axis in mediating skeletal muscle damage. These findings highlight the clinical and preventive relevance of the GDF15–GFRAL axis and underscore the need for future mechanistic studies to elucidate its role in sarcopenia.

Graphical abstract

Highlights

• •Serum levels of GDF15 and GFRAL were associated with an increased risk of sarcopenia
• •Elevated GDF15 and GFRAL levels showed more pronounced adverse effects in women
• •These findings link the GDF15–GFRAL axis to skeletal muscle impairment

Health sciences

Introduction

Sarcopenia is a growing public health concern, characterized by diminished muscle strength, decreased muscle mass, and declined physical performance.¹ This condition significantly contributes to disability and frailty, increasing the risk of falls, osteoporosis, and a reduced quality of life and longevity.^1,2 Its current prevalence ranges from 5% to 13% among individuals aged 60–70 years, with a global total of 500 million cases by 2050.^1,2 Despite recent advances in research, the mechanisms underlying sarcopenia remain incompletely elucidated.

Circulating proteins are essential for skeletal muscle health and influence metabolic, immune, and neurological functions.³ Recently, growth differentiation factor 15 (GDF15), a novel pleiotropic hormone initially identified as a macrophage activation regulator, has been recognized for its involvement in energy and glucose homeostasis, diabetes, cancers, anorexia/cachexia syndromes, and cardiovascular and inflammatory diseases.^4,5,6 Emerging epidemiological evidence highlights GDF15’s strong association with aging,⁷ linking elevated GDF15 levels to increased chronic inflammation, muscle damage, and all-cause mortality, for which it is an independent predictor.^7,8 However, its impact on sarcopenia remains controversial and poorly understood.

The glial cell line-derived neurotrophic factor family receptor alpha-like (GFRAL) was initially identified as an orphan receptor for GDF15 in 2017.^6,9,10,11 Although GFRAL expression is primarily restricted to the brainstem,^6,9 the GDF15–GFRAL complex recruits the co-receptor RET, which is expressed in distal cells or tissues that otherwise lack GFRAL.⁶ The GDF15-GFRAL axis is well-established as a critical regulator of appetite, energy homeostasis, and metabolic regulation.^6,9,10,11,12 Currently, the GDF15-GFRAL axis is increasingly recognized for its systemic roles in cancers, diabetes, and cachexia,^9,10,11,12 and for its proposed influence on skeletal muscle via the muscle-brain axis.^12,13,14 However, fundamental aspects of its biology are poorly understood. It remains unclear whether elevated GDF15 is beneficial or detrimental in disease, and in particular, its roles in sarcopenia have scarcely been investigated.

In this study, we used the UK Biobank (UKB) cohort to examine the associations between serum GDF15 and GFRAL and sarcopenia risk and its phenotypes. An independent Chinese cohort was utilized for validation. Furthermore, we characterized GDF15 and GFRAL in mouse skeletal muscle using single-cell RNA sequencing and immunohistochemistry. Together, our results provide evidence that the GDF15-GFRAL axis is involved in sarcopenia, thereby informing future mechanistic and preventive research.

Results

Population characteristics

Table 1 summarized the characteristics of 44,736 participants from the UKB cohort. The average age was 57.17 years (standard deviation [SD] = 8.09), and 53.86% were women. Among them, 1,160 individuals (6.19%) had probable sarcopenia, and 237 individuals (1.26%) had sarcopenia in older adults. The baseline level of GDF15 showed an overall SD of 0.58 NPX. Compared with the non-sarcopenia group (0.06 ± 0.57 NPX), GDF15 levels were higher in both the confirmed sarcopenia group (0.68 ± 0.70 NPX) and the probable sarcopenia group (0.37 ± 0.67 NPX). Similarly, the baseline GFRAL level had an overall SD of 0.67 NPX and was elevated in individuals with confirmed sarcopenia (0.15 ± 0.73 NPX) and those with probable sarcopenia (0.11 ± 0.65 NPX) relative to the non-sarcopenia group (0.03 ± 0.66 NPX).

Table 1.

The baseline characteristics of the study participants

Characteristics	GDF15 (n = 44736)			GFRAL (n = 37050)
	Non-sarcopenia	Probable sarcopenia	Sarcopenia	Non-sarcopenia	Probable sarcopenia	Sarcopenia
Number of people, n (%)	42441(94.87)	1956(4.37)	339(0.76)	35191(94.98)	1592(4.30)	267(0.72)
Age (years), mean (SD)	56.97(8.10)	60.54(7.01)	62.60(5.92)	57.01(8.07)	60.56(7.02)	62.47(6.10)
Sex, n (%)
Female	22679(53.44)	1285(65.70)	133(39.23)	18850(53.56)	1064(66.83)	108(40.45)
Male	19762(46.56)	671(34.3)	206(60.77)	16341(46.44)	575(33.17)	63(59.55)
Townsend deprivation index, mean (SD)	−1.52(2.97)	−0.81(3.29)	0.17(3.53)	−1.52(2.97)	−0.73(3.33)	0.18(3.59)
Education, n (%)
Degree	13341(31.43)	439(22.44)	33(9.73)	11060(31.43)	362(22.74)	25(9.36)
No degree	29100(68.57)	1517(77.56)	306(90.27)	24131(68.57)	1230(77.26)	87(90.64)
Smoking, n (%)
Never	22889(53.9)	1074(54.91)	155(45.72)	18900(53.71)	872(54.78)	123(46.07)
Previous	15243(35.9)	675(34.51)	150(44.25)	12662(35.98)	552(34.67)	117(43.82)
Current	4309(10.2)	207(10.58)	34(10.03)	3629(10.31)	168(10.55)	27(10.11)
Alcohol intake, n (%)
Daily or almost daily	9079(21.39)	333(17.02)	51(15.04)	7627(21.67)	270(16.96)	36(13.48)
1-4 times per week	21565(50.81)	789(40.80)	147(43.36)	17812(50.62)	662(41.58)	121(45.32)
1-3 times per month	4662(10.98)	230(11.76)	43(12.68)	3821(10.86)	187(11.75)	32(11.99)
Occasional or never	7135(16.81)	595(30.42)	98(30.42)	5931(16.85)	473(29.71)	78(29.21)
Sarcopenia phenotypes
Grip strength, mean (SD)	33.63(10.76)	15.38(6.08)	17.22(6.44)	33.58(10.78)	15.28(6.07)	17.23(6.44)
ALST (kg), mean (SD)a	22.39(5.32)	20.65(5.00)	21.97(4.99)	22.37(5.32)	20.45(4.91)	21.88(5.14)
ALST/BMI, mean (SD)b	0.82(0.17)	0.76(0.16)	0.68(0.14)	0.82(0.17)	0.76(0.15)	0.68(0.14)
Slow pace, n (%)	3204(7.58)	490(25.48)	146(43.84)	2611(7.44)	385(24.63)	118(45.21)
GDF15 (NPX), mean (SD)c	0.06(0.57)	0.37(0.67)	0.68(0.70)	–	–	–
Female	−0.02(0.53)	0.30(0.64)	0.48(0.61)	–	–	–
Male	0.15(0.61)	0.50(0.72)	0.82(0.73)	–	–	–
≤60 years	−0.13(0.51)	0.16(0.69)	0.45(0.66)	–	–	–
>60 years	0.33(0.54)	0.51(0.63)	0.78(0.70)	–	–	–
GFRAL (NPX), mean (SD)d	–	–	–	0.03(0.66)	0.11(0.65)	0.15(0.73)
Female	–	–	–	0.08(0.66)	0.18(0.63)	0.34(0.73)
Male	–	–	–	−0.03(0.65)	−0.04(0.65)	0.03(0.70)
≤60 years	–	–	–	−0.05(0.66)	0.03(0.65)	0.08(0.74)
>60 years	–	–	–	0.14(0.64)	0.16(0.64)	0.19(0.72)

Values are means (SD) or n (%).

^aALST, appendicular lean soft tissue

^bBMI, body mass index

^cGDF15, growth differentiation factor 15

^dGFRAL, GDNF family receptor a-like; SD, standard deviation.

Associations of growth differentiation factor 15 level with sarcopenia risk

Higher levels of serum GDF15 were associated with increased risks of sarcopenia (Figure 1A) and its phenotypes (low muscle strength, low muscle mass, and low physical performance). A non-linear positive association was observed between GDF15 levels and the risk of sarcopenia (p for non-linearity = 0.0016) (Figure 1B), as well as the specific sarcopenia phenotype risks (low muscle strength: p = 0.0015; low muscle mass: p < 0.001; and low physical performance: p < 0.001) (Figures S2A–S2C). Furthermore, the risks of sarcopenia and its phenotypes were increased from quintile 1 to quintile 4 of the serum GDF15 levels (p trend <0.001), and these associations remained robust after adjustment for the covariates, including sex, age, education, TDI, smoking status, and alcohol consumption (Table S1).

Figure 1.

The non-linear dose-response relationships between serum protein levels of GDF15 and GFRAL and sarcopenia risk

(A) Serum GDF15 concentrations with age across three sarcopenia status groups; (B) The non-linear dose-response relationship between serum protein levels of GDF15 and sarcopenia risk.; (C) Serum GFRAL concentrations with age across three sarcopenia status groups; (D) The non-linear dose-response relationship between serum protein levels of GFRAL and sarcopenia risk.

Data are presented as fitted values with 95% confidence intervals (CIs). See also Figure S2.

When all participants were categorized as non-sarcopenia, probable sarcopenia, and confirmed sarcopenia, higher GDF15 levels were significantly associated with the increased risks of probable sarcopenia (OR [95% CI] = 1.73 [1.60 to 1.86], p < 0.001) and confirmed sarcopenia (OR [95% CI] = 2.23 [1.94 to 2.57], p < 0.001), compared to the non-sarcopenia group. We further found similar associations of elevated GDF15 levels with risks of low muscle strength (OR [95% CI] = 1.49 [1.42 to 1.56], p < 0.001), low muscle mass (OR [95% CI] = 1.40 [1.34 to 1.47], p < 0.001), and low physical performance (OR [95% CI] = 1.93 [1.85 to 2.00], p < 0.001) (Figure 2A). These results did not change after adjustment for additional covariates, including healthy diet, physical activity, diabetes, hypertension, and high cholesterol (Table S2). Similar patterns were observed in the PSM method (Table S3), and the main results were validated in the Chinese population (Figures S3A and S3B).

Figure 2.

Associations between serum protein levels of GDF15 and GFRAL and risks of sarcopenia and its phenotypes

(A) Association of serum protein levels of GDF15 with risks of sarcopenia and its phenotypes; (B) Association of serum protein levels of GFRAL with risks of sarcopenia and its phenotypes.

Model 0 was the crude model; Model 1 was adjusted for sex and age; Model 2 was further adjusted for Townsend deprivation index, educational attainment, smoking status, and alcohol consumption based on Model1. Abbreviations: CI, confidence interval. Data are presented as fitted values with 95% CIs. See also Figure S3 and Tables S2, S3, and S5.

Association of glial cell line-derived neurotrophic factor family receptor α-like with sarcopenia risk

Elevated GFRAL levels were positively correlated with sarcopenia risk (Figure 1C). A linear relationship was observed between GFRAL levels and the risks of sarcopenia, low muscle strength, and low muscle mass (all p < 0.05), and a non-linear relationship was observed between GFRAL levels and low physical performance (p for non-linearity = 0.0092) (Figures 1D and S2D–S2F). Similarly, the risks of sarcopenia, low muscle mass, and low physical performance were increased from quintile 1 to quintile 4 of serum GFRAL levels (p trend <0.05) (Table S4).

Additionally, when participants were categorized into non-sarcopenia, probable sarcopenia, and sarcopenia groups, statistical associations were observed between GFRAL levels and both probable sarcopenia and confirmed sarcopenia after adjusting for sex and age (both p < 0.05), and the same association still existed in confirmed sarcopenia after the additional adjustment for education, TDI, smoking status, and alcohol consumption. Moreover, a higher GFRAL level was associated with increased risks of low muscle mass and low physical performance (low muscle mass: OR [95% CI] = 1.13 [1.07 to 1.20], p < 0.001; low physical performance: OR [95% CI] = 1.10 [1.04 to 1.15], p = 0.013) (Figure 2B). The positive associations of serum GFRAL with low muscle mass and low physical performance remained robust after adjustment for the covariates and the validation using the PSM method (Table S5). Meanwhile, an elevated GFRAL level was also observed in Chinese individuals with sarcopenia (Figure S3C and Table S6).

Subpopulation analyses

As shown in Table 2, elevated GDF15 levels exert a more deleterious impact on the risk of low physical performance among older adults. A one SD increase in GDF15 levels was associated with an increased risk of low physical performance by 81% (95% CI: 70%–92%) in younger participants and 117% (95% CI: 105%–129%) in older adults. Moreover, higher levels of GFRAL posed varying degrees of risk for low muscle mass across women participants, and a more pronounced risk was observed in younger participants (sex: p for interaction = 1.19 × 10⁻⁴; age: p for interaction = 6.98 × 10⁻⁶). Specifically, a one SD increase in GFRAL levels was associated with an increased risk of low muscle mass by 34% (95% CI: 23%–47%) in women individuals, and 33% (95% CI: 23%–45%) in younger individuals. Elevated GFRAL levels were also found to be a risk factor for low physical performance in women individuals and younger individuals (sex: p for interaction = 0.024; age: p for interaction = 2.81 × 10⁻⁵). A one SD increase in GFRAL levels was associated with an increased risk of low physical performance by 16% (95% CI: 9%–24%) in women, similarly, the risk of low physical performance increased by 24% (95% CI: 16%–32%) in younger individuals.

Table 2.

Distribution of serum GDF15 and GFRAL levels on risk of sarcopenia in subpopulations

Characteristics	GDF15a		GFRALa
	OR and 95% CI per 1 SD increase	P	OR and 95% CI per 1 SD increase	P
Sarcopenia
≤60 years	1.70(1.51,1.91)	<2.00 × 10−16	1.24(0.96,1.59)	0.096
>60 years	1.57(1.41,1.74)	<2.00 × 10−16	1.19(1.01,1.41)	0.042
Men	1.72(1.51,1.96)	2.97 × 10−16	1.08(0.90,1.31)	0.406
Women	1.53(1.34,1.75)	2.21 × 10−10	1.32(1.05,1.67)	0.017
Low muscle strength
≤60 years	1.56(1.47,1.66)	<2.00 × 10−16	1.15(1.05,1.25)	2.85 × 10−3
>60 years	1.45(1.37,1.53)	<2.00 × 10−16	1.01(0.94,1.10)	0.718
Men	1.56(1.45,1.68)	<2.00 × 10−16	0.99(0.90,1.09)	0.846
Women	1.52(1.43,1.63)	<2.00 × 10−16	1.05(0.97,1.13)	0.219
Low muscle mass
≤60 years	1.55(1.46,1.64)	<2.00 × 10−16	1.33(1.23,1.45)	3.78 × 10−12
>60 years	1.37(1.30,1.44)	<2.00 × 10−16	1.08(1.01,1.16)	0.029
Men	1.37(1.29,1.45)	<2.00 × 10−16	1.04(0.97,1.11)	0.301
Women	1.54(1.43,1.66)	<2.00 × 10−16	1.34(1.23,1.47)	4.48 × 10−11
Low physical performance
≤60 years	1.81(1.70,1.92)	<2.00 × 10−16	1.24(1.16,1.32)	1.91 × 10−10
>60 years	2.17(2.05,2.29)	<2.00 × 10−16	1.04(0.97,1.11)	0.242
Men	1.78(1.69,1.87)	<2.00 × 10−16	1.04(0.97,1.11)	0.320
Women	1.96(1.87,2.06)	<2.00 × 10−16	1.16(1.09,1.24)	3.18 × 10−6

OR, odds ratio; CI, confidence interval.

^aAdjustment for age, sex, Townsend deprivation index, educational attainment, smoking status, and alcohol consumption.

Potential evidence linking the growth differentiation factor 15–glial cell line-derived neurotrophic factor family receptor α-like to skeletal muscle impairment

We investigated the links between GDF15 and GFRAL and skeletal muscle impairment by analyzing scRNA-seq data from tibialis anterior samples of frail and control mice (Figure 3A). Frail mice was assessed using core diagnostic indicators of sarcopenia, such as grip strength, given that sarcopenia is a key pathological basis of frailty.¹⁵ Our findings revealed significant upregulation of GDF15 in M2 macrophages in frail mice (Figure 3B), with enrichment in fibrosis-related pathways including cytokine activity, SMAD protein signal transduction, and TGF-β receptor binding (Figure 3C). Fibrosis gene scores, particularly in the AP-1 and macrophage pathways, were notably higher in frail mice (Figures 3D and 3E). Additionally, the RET-MAPK/ERK pathway was identified as an upstream signal of the AP-1 pathway. Previous research indicates that GDF15/GFRAL/RET can affect distant cells or tissues that do not express GFRAL, such as in amyotrophic lateral sclerosis.^6,9,10,11 As expected, GFRAL was not detected in skeletal muscle. Analysis of skeletal muscle samples collected from mice aged 17–31 months showed that GDF15 was distinctly upregulated in older mice, culminating in significantly elevated levels in frail mice compared with non-frail controls (Figure 3F). These findings suggest that increased GDF15–GFRAL signaling may contribute to skeletal muscle impairment in sarcopenia or frailty.

Figure 3.

The effects of elevated GDF15 and GFRAL levels on sarcopenia

(A) UMAP visualization of 11 annotated cell types in skeletal muscles from two frail mice and three control mice; (B) volcano plot showing differentially expressed genes in M2 macrophages from non-frail and frail muscle; (C) pathways involving GDF15 in M2 macrophages; (D–E) significantly up-regulated pathways of fibrosis-related genes between non-frail and frail muscle; (F) immunohistochemical staining for GDF15, with positive signals visualized as brown DAB precipitate. Scale bars, 200 μm.

Discussion

In the UKB cohort, we observed that elevated serum levels of GDF15/GFRAL were associated with an increased risk of sarcopenia. Additionally, our findings indicated that serum GDF15 levels significantly contributed to sarcopenia risk in the entire population, whereas GFRAL exhibited more pronounced adverse effects specifically in women. Single-cell RNA sequencing analysis of skeletal muscle tissues further corroborated that the upregulation of the GDF15 and GFRAL may be implicated in skeletal muscle impairment associated with sarcopenia or frailty.

Recent research identifies GDF15 as a broadly applicable biomarker for aging and skeletal muscle health.^16,17 A developing hormetic hypothesis suggests that GDF15 may play a beneficial role in skeletal muscle when expressed at elevated levels in response to acute stressors.¹⁸ However, it is emphasized that GDF15 has been shown to have maladaptive effects when chronically expressed in pathological conditions such as obesity and aging. Recent human studies and pre-clinical animal models have reported increased GDF15 expression in skeletal muscle during aging,¹⁹ which aligns with our findings in sarcopenia. Although there is substantial evidence linking GDF15 to maximal muscle power,⁴ muscle performance,²⁰ and mitochondrial dysfunction in the context of sarcopenia,^21,22 significant gaps remain in our understanding of the systemic effects of GDF15 on primary sarcopenia.²³ Our study is the first to examine the detrimental effects of serum GDF15 on sarcopenia across an entire population, accounting for confounding variables and PSM validations. Notably, this study underscores the significant role of serum GDF15 in contributing to low physical performance in sarcopenia, in line with the 30s sit-to-stand muscle power test for sarcopenia across the adult lifespan.⁴ A robust body of evidence indicates that elevated serum levels of GDF15 are positively associated with reduced muscle strength and physical motor dysfunction.²⁴ Overexpression of GDF15 in animal models can induce a reduction in skeletal muscle fibers and muscle atrophy in mice, while the systemic administration of GDF15 in mice can result in restricted running activities.^13,25 Thus, there is compelling evidence to suggest that GDF15 may serve as a potential molecular target for disease-modifying interventions in individuals with sarcopenia.

Recent studies highlight that the binding of GFRAL to GDF15 is essential for GDF15’s biological effects.²⁶ Our study thoroughly investigates the association of GFRAL with sarcopenia, muscle strength, muscle mass, and physical performance. Using various models (model0, model1, and model2) and the PSM method, we found a strong positive correlation between high GFRAL levels and increased sarcopenia risk. GFRAL, a high-affinity receptor for GDF15, is a pivotal member of the GDNF family and is involved in maintaining the nervous system. However, emerging evidence on GFRAL is increasingly recognized for roles beyond the nervous system, including in cachexia, cancer, kidney disease, and obesity, supporting our findings.^27,28 We used RCS to evaluate the dose-response effect of GFRAL on sarcopenia as a continuous variable in the overall population. This approach revealed a significant dose-response relationship and even non-linearity in the association with low physical performance. These findings were consistent with the observed association between serum GDF15 and sarcopenia. In our subpopulation analysis, we found that high serum GFRAL levels negatively impact women more, which aligns with findings in other diseases. Female GFRAL knockout mice show impaired GDF15–GFRAL signaling involved in compensatory food intake, suggesting a stronger effect of GFRAL on sarcopenia in women.²⁹ To our knowledge, this is the first study to implicate GFRAL in the pathogenesis of sarcopenia, and this association was consistently observed in both Chinese and White British populations. Comprehensive evidence indicates that elevated GFRAL levels exacerbate sarcopenia, underscoring the need to focus on its role in mediating GDF15 or its combined effect in this condition.

The potential biological role of the GDF15-GFRAL axis in sarcopenia is increasingly recognized.^6,9,10,11 GFRAL specifically binds to GDF15, triggering RET activation and phosphorylation of Erk, Akt, and PLCγ.^6,30 This signaling cascade is linked to metabolic disorders mediated by the GDF15/GFRAL/RET pathway.⁶ The GDF15-GFRAL axis has the potential to interact with the signaling co-receptor RET on remote cells or tissues that do not naturally express GFRAL, thereby initiating downstream signaling.⁶ Inhibiting GFRAL suppresses myogenin-related transcription and the expression of muscle atrophy genes, while GDF15-GFRAL signaling affects lipid metabolism in early amyotrophic lateral sclerosis.¹¹ In aging, GDF15 acts as a key regulator of muscle fibrosis, with its expression rising due to chronic inflammation and other triggers.⁸ Elevated GDF15 levels may reduce muscle mass by inducing apoptosis in skeletal muscle cells and promoting atrophy,³¹ with mitochondrial stress signals interacting through GFRAL signaling.³² GDF15, initially recognized as an autocrine factor for macrophage activation, can be secreted by M2-like macrophages within lipid microenvironments.^8,33 As a member of the TGF-β superfamily, it interacts with TGF-β receptors to influence pro-fibrotic gene expression, regulated by GFRAL signaling.³⁴ For example, the GDF15-induced upregulation of Smad7 is mediated by GFRAL, with Smad7 associated with muscle fibrosis.³⁴ Additionally, GDF15 signaling promotes an immunosuppressive environment, regulates M2 macrophage activity, and worsens aging tissue issues such as cellular fibrosis and sarcopenia.³⁵ Thus, this study offers new insights into sarcopenia, though the role of the GDF15-GFRAL axis in this disease remains to be fully elucidated.

In conclusion, our findings from two large-scale cohorts identify elevated GDF15 and GFRAL levels as a potential risk factor and biomarker for sarcopenia. The data also reveal a stronger adverse effect on physical performance in women. Supporting evidence from animal models indicates that GDF15-GFRAL signaling may contribute to skeletal muscle impairment in frailty or sarcopenia. Ultimately, these findings underscore the clinical and preventive relevance of this axis while highlighting the need to elucidate its precise mechanistic role in sarcopenia pathogenesis in the future.

Limitations of the study

This study has several limitations. The lack of repeated measurements of GDF15 and GFRAL may lead to regression dilution. Although the findings were consistent, slight differences in baseline characteristics exist between the Chinese cohort and the UK Biobank cohort. Moreover, further investigation is required to elucidate the origin and specific function of circulating GDF15 and GFRAL, the clinical applicability of our findings, and the underlying biological mechanisms in sarcopenia.

Resource availability

Lead contact

Requests for further information and resources should be directed to and will be fulfilled by the lead contact, Jun Wu (wujun@gdmu.edu.cn).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •Single-cell RNA-seq data have been deposited at GEO with the accession number and are publicly available from the date of publication. Data from the UK Biobank are available on application at www.ukbiobank.ac.uk (ID: 62663).
• •This article does not report original code.
• •Additional information required to reanalyze the data reported in this article is available from the lead contact upon request.

Acknowledgments

We are deeply grateful for the contributions of all the participants, their families, and the investigators and staff involved in this study. This work was supported by the National Natural Science Foundation of China (82273709); the Guangdong Provincial Medical Research Foundation (B2025656); Non-funded scientific and technological research project in Zhanjiang (2024B01217); Undergraduate innovation and entrepreneurship education base for the prevention and control of chronic non-communicable diseases (2JD24026); The innovation training project of Guangdong Medical University (S202510571050); and Guangdong medical university combined clinical and basic sciences technology innovation program (GDMULCJC202416).

Author contributions

Conceptualization, J.W. and J.D.N.; methodology and analysis, J.W. and Z.S.Z.; Investigation, W.Y.Z. and H.Q.Z.; writing – original draft, J.W., P.D.F., and Z.S.Z; writing – review and editing, J.W., P.D.F., and Z.S.Z.; funding acquisition, J.W., X.L.Y., and J.D.N.; resources, J.W., X.L.Y. and J.D.N.; supervision, X.L.Y. and J.D.N.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
GDF15 (for immunohistochemical staining)	Proteintech	Cat No. 27455-1-AP
Biological samples
Serum blood samples	Guangdong medical university	Ethics No: YJYS2020142
Deposited data
Single cell sequencing data	GSE316316	https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE316316
Protein data	UKB-PPP	www.ukbiobank.ac.uk
Critical commercial assays
10X Genomics Chromium Single-Cell 3' kit (V3)	10x Genoinics	Cat No.1000157
Olink Explore 384 Inflammation	olink	Cat. No.97500
Olink Explore 384 Neurology	olink	Cat. No.97800
Olink Explore 384 Oncology	olink	Cat. No.97600
Olink Explore 384 Cardiometabolic	olink	Cat. No.97700
Experimental models: Organisms/strains
Muscles of C57BL/6J wild-type mice	Animal center of Guangdong medical university	License mumber: SCK (Zhe) 2019-0001
Software and algorithms
R software	The R Foundation	R version 4.2.2

Experimental model and study participant details

Serum blood samples

A total of 44,736 White British individuals were recruited, and their serum samples were collected from 22 centers in England, Wales, and Scotland between 2006 and 2010 in the UKB Population-Based Proteomics (PPP) project. Detailed information regarding the UKB protocol is available online (https://www.ukbiobank.ac.uk/).³⁶ The UKB study received ethical approval from the North West Research Ethics Committee (06/MRE08/65), and all participants provided written informed consent. Serum samples from Chinese older adults were collected from the Dongguan Longitudinal Aging Study (DLAS), as previously described.³⁷ Blood samples were collected by individual venipuncture into tubes without EDTA, and then centrifuged at 4°C to obtain the serum, which was subsequently stored at −80°C. The study was approved by the ethics committee of Guangdong University (YJYS2020142). The sex and age of the human subjects are reported in the current manuscript (Table 1).

Experimental animals

Experiments were performed on C57BL/6J aged 17, 23, and 31 months. Male and female mice were analyzed. The mice used in this investigation were obtained from the Animal Center of Guangdong Medical University. The ethics approval for the mouse experiment was obtained from the experimental animal ethics committee of Guangdong Medical University (GDY2202123).

Method details

Study population

Adults aged 37 to 73 years were recruited from 22 centers in England, Wales, and Scotland between 2006 and 2010 in the UKB cohort. Detailed information regarding the UKB protocol is available online (https://www.ukbiobank.ac.uk/).³⁶ Within the UK Biobank Population-Based Proteomics (PPP) project, a total of 44,736 white British individuals with Olink proteomics measurements were selected for analyses (Figure S1). Participants were excluded based on the following criteria: (1) Non-white British ethnic backgrounds (n = 6,828); (2) GDF15/GFRAL data collected outside the 2006 to 2010 timeframe (n = 2,074); (3) Missing values of covariates (n = 816). The UKB study received ethical approval from the North West Research Ethics Committee (06/MRE08/65), and all participants provided written informed consent. The validation population was derived from the Dongguan Longitudinal Aging Study (DLAS) in China. As previously described,³⁷ DLAS recruited over 50,000 individuals aged 60 years or older between 2018 and 2024. Seventy participants were enrolled in the independent replication study, which received ethical approval from the Ethics Committee of Guangdong University (YJYS2020142), and each participant signed the informed consent.

Laboratory measurement of GDF15 and GFRAL

Blood samples were collected from all participants into EDTA anticoagulant tubes. After centrifugation, the plasma supernatant was collected and stored at −80°C for subsequent proteomic analysis. The concentrations of GDF15 and GFRAL in the plasma were quantified using the Olink Proximity Extension Assay (PEA) technology, as reported previously.³⁸ This method uses a pair of antibodies to each target protein, with each antibody conjugated to a unique DNA single-strand marker. Upon binding to the target protein, the two DNA single strands, come into close proximity, hybridize and extend to form a double-stranded DNA template, which is then quantified by qPCR or NGS technology.

Single-cell RNA-seq using 10× genomics chromium

The tibialis anterior muscles from 26-month-old female C57BL/6 mice (license number: SCK (Zhe) 2019-0001) underwent enzymatic digestion, followed by passage through a 30–70 μM cell sieve. Erythrocytes were removed, and the sample was further filtered using MS Columns to eliminate debris and dead cells. The resultant cell suspension was resuspended in 100 μL of 1×PBS containing 0.04% BSA. Cell counting was conducted using a hemocytometer or the Countess II Automated Cell Counter, resulting in a concentration of 700–1200 cells/μL. The single-cell suspension was processed using a 10× Chromium chip in accordance with the 10X Genomics Chromium Single-Cell 3’ kit (V3) protocol. cDNA amplification and library construction were performed following standard procedures. Considering that sarcopenia underlies frailty and exhibits similar changes in skeletal muscles, frail mice were designated as the case group based on the 31-item frailty index (FI) evaluation, with an FI score ≥ 0.18, while mice with FI scores < 0.18 were classified as controls.³⁹ After quality control filtering, two frail mice and three control mice were included in the single-cell sequencing analysis. The ethics approval for the mouse experiment was obtained from the experimental animal ethics committee of Guangdong mMedical uUniversity (GDY2202123).

Immunohistochemical staining

Immunohistochemical staining for GDF15 was performed on paraffin-embedded gastrocnemius tissues from female C57BL/6J mice aged 17, 23, and 31months. After dewaxing and antigen retrieval in citrate buffer (pH 6.0), sections were sequentially incubated with anti-GDF15 (27455-1-AP, Proteintech; 1:300; overnight at 4°C) and HRP goat anti-rabbit secondary antibody (1:200; 50 mins, 37°C), followed by hematoxylin counterstaining, dehydration, and mounting. Positive signals were detected as brown DAB precipitate.

Quantification and statistical analysis

Assessment of sarcopenia patients

Sarcopenia was defined according to the European Working Group on Sarcopenia in Older People 2 (2019 EWGSOP2) and the Asian Working Group for Sarcopenia (2019 AWGS). Probable sarcopenia was diagnosed in participants with low muscle strength, while sarcopenia was diagnosed in those with both low muscle strength and low muscle mass, or reduced physical performance.

In the UKB study, low muscle strength was determined by grip strength (less than 27 kg for men and less than 16 kg for women), measured with a Jamar J00105 hand dynamometer.⁴⁰ Low muscle mass was defined by an adjusted appendicular lean soft tissue (ALST) to body mass index (BMI) ratio below 0.84 for men and 0.55 for women, assessed using a Tanita BC-418MA body composition analyzer. ALST was calculated using the formula: ALST (kg) = (0.958 × appendicular FFM (kg)) - (0.166 × S) + 0.308, where S is 0 for women and 1 for man.⁴¹ Recent studies suggest that adjusting ALST for BMI is more suitable than height² adjustment.⁴¹ Individuals self-reporting a slow walking speed were classified as having low physical performance in UKB.⁴¹ In the DLAS cohort, the appendicular skeletal muscle index (ASMI) was employed to assess low muscle mass, as determined by bioelectrical impedance analysis (BIA) with thresholds set at <7.0 kg/m² for men and <5.7 kg/m² for women. Low muscle strength was assessed using handgrip strength, with cutoffs established at < 28 kg for men and <18 kg for women. Low physical performance was defined as a walking speed of <1.0 m/s at a normal pace without acceleration or deceleration.

Covariate measurements

The covariates adjusted for in this study included age, sex, educational attainment (degree or no degree), Townsend deprivation index (TDI), smoking status (categorized as never, previous, and current smoking), alcohol consumption (categorized as daily or almost daily, 1–4 times per week, 1–3 times per month, or occasional/never). Additionally, sensitivity analyses considered covariates related to diet, physical activity, and health conditions such as diabetes, hypertension, and high cholesterol. Data for these variables were obtained through questionnaire surveys and laboratory assessments.

Observational analyses of associations between GDF15/GFRAL levels and sarcopenia risk

A multivariable linear regression model was used to estimate the association of GDF15/GFRAL expression with sarcopenia and its phenotypes, while adjusting for covariates. Subsequently, restricted cubic spline (RCS) models were developed to elucidate the non-linear relationships between GDF15/GFRAL and the risk of sarcopenia and its phenotypes. Multivariate logistic regression models were used to calculate the odds ratio (OR) and 95% confidence interval (CI) for the association between GDF15/GFRAL and sarcopenia risk. Three models were constructed: Model 0 (unadjusted), Model 1 (adjusted for age and sex), and Model 2 (fully adjusted for TDI, education, smoking, and alcohol consumption). We performed subgroup analyses based on sex and age, employed propensity score matching (PSM) utilizing logistic regression, and conducted sensitivity analyses with further adjustments for variables including diet, physical activity, and health conditions such as diabetes, hypertension, and high cholesterol to ensure robust results.

10× Genomics chromium library and sequencing bioanalyses

Single-cell suspensions were loaded onto a 10× Chromium chip to capture 8,000 cells, and over 40,000 cells passed quality control for scRNA-seq data. Cell types were identified based on known marker genes from the literature and Gene Ontology (GO) analysis. Differentially expressed genes (DEGs) were identified using thresholds of |log₂FC| > 0.5 and adjusted p < 0.05.. Fibrosis-related pathways were analyzed using gene set enrichment analysis (GSEA).

All analyses were conducted using R software version 4.2.2, with two-sided p < 0.05 indicating statistical significance.

Published: February 16, 2026