Association Between Advanced Glycation End Products and Sarcopenia: The Mediating Role of Osteoporosis

Abstract

Context

Advanced glycation end products (AGEs) are a group of molecules formed through nonenzymatic reactions. These compounds are associated with several age-related diseases, including sarcopenia and osteoporosis.

Objective

This work aimed to investigate the relationships between AGEs, osteoporosis, and sarcopenia in community-dwelling older adults.

Methods

This cross-sectional study included 1991 older adults aged 72.37 ± 5.90 years from China. AGE levels were measured by the AGE Reader device. Bone mineral density was assessed using dual-energy X-ray absorptiometry, and osteoporosis was diagnosed based on a T score of less than −2.5. Sarcopenia was defined as loss of muscle mass plus loss of muscle strength and/or reduced physical performance. Presarcopenia was defined as low muscle mass with normal muscle strength and normal physical performance.

Results

The prevalence of sarcopenia was 18.5%, and that of osteoporosis was 40.5%. Compared to the lowest AGE quartile, the highest AGE quartile showed a significant association with sarcopenia (odds ratio [OR] 2.42; 95% CI, 1.60-3.66) (P for trend <.001), but not with presarcopenia. Per-SD increase in AGE was associated with higher odds of sarcopenia (OR 1.44; 95% CI, 1.26-1.66). Additionally, in the mediation analysis, when AGEs were treated as a continuous variable (the mediation effect is denoted by Z_a*Z_b = 18.81; 95% CI, 8.07-32.32]—the 95% CI does not contain zero, representing a significant mediating effect) or a categorical variable (the mediating effect is expressed as Z_mediation = 3.01 > 1.96, which represents a significant mediating effect), osteoporosis played a partial mediating role in the association between AGEs and sarcopenia.

Conclusion

Elevated AGEs are associated with sarcopenia but not with presarcopenia. This association was partially mediated by osteoporosis.

Sarcopenia is a syndrome characterized by a progressive and generalized loss of skeletal muscle mass and strength (1). The prevalence of sarcopenia varies in different populations around the world due to different diagnostic criteria. Among older adults aged 60 to 89 years in eastern China (2), the prevalence of sarcopenia is 21.7% in women and 12.9% in men. Sarcopenia is associated with an increased risk of adverse outcomes, including physical disability (3), poor quality of life (4), and high prevalence among individuals with cardiovascular diseases (5), dementia (6), and osteoporosis (7). It has been reported that the prevalence of sarcopenia is 29.7% in individuals with osteoporosis, whereas it is 9.0% in those without osteoporosis (7). However, early screening and intervention for sarcopenia and osteoporosis are often insufficient among older adults residing in Chinese communities.

Advanced glycation end products (AGEs) are a group of molecules generated nonenzymatically by attaching sugars to proteins, lipids, or nucleic acids, leading to the modification and cross-linking of proteins (8). The accumulation of AGEs in tissues has been associated with various age-related diseases, including sarcopenia and osteoporosis. Previous studies have reported that accumulated AGEs could contribute to reduced muscle mass and strength, leading to sarcopenia in patients with type 2 diabetes mellitus (T2DM) (9, 10) as well as in older adults in a community with a northern European background (11). Specifically, AGEs might be related to the pathogenesis of sarcopenia by binding to receptor for advanced glycation end products (RAGE) (12) and cross-linking between collagen molecules (13). Additionally, intracellular accumulation of AGEs can also induce osteoblast apoptosis via endoplasmic reticulum stress (14) and disrupt the functions of osteoblasts by inducing cell ferroptosis (15), thus contributing to osteoporosis. However, the association of AGEs with sarcopenia and osteoporosis, as well as whether AGEs can serve as an early screening tool for these conditions, remains uncertain among community-dwelling older adults in China.

To the best of our knowledge, osteoporosis and sarcopenia frequently occur together. This coexistence has been recognized as a syndrome known as osteosarcopenia (16). The evidence of shared pathophysiological mechanisms between sarcopenia and osteoporosis (17) indicates a closely intertwined relationship between these two conditions. A bidirectional mendelian randomization study (18) indicated that osteoporosis might be a risk factor for the development of sarcopenia. Consistent with the aforementioned research, a 4-year follow-up study (19) suggested that the presence of osteoporosis significantly increased the risk of future sarcopenia, whereas the presence of sarcopenia did not increase the risk of future osteoporosis. However, another cross-sectional study (20) demonstrated a significant relationship between osteoporosis and sarcopenia, suggesting that they act as risk factors for each other. Therefore, the relationship between osteoporosis and sarcopenia is complex and requires further exploration.

In conclusion, we believe that there is a relationship among AGEs, osteoporosis, and sarcopenia to some extent. However, no study has comprehensively elucidated the relationship among them. Therefore, the aim of this study was to examine the associations among AGEs, osteoporosis, and sarcopenia in community-dwelling older adults in China, as well as to explore potential mediating effects.

Material and Methods

Participants

The data used in this study were collected from the Adult Physical Fitness and Health Cohort Study (APFHCS [ChiCTR1900024880]) conducted between June 2019 and August 2021. APFHCS is a large, prospective, dynamic cohort study focusing on older adults aged 65 years or older living in Shanghai, China. Participants were recruited for comprehensive annual health examinations and were asked to complete detailed questionnaires regarding their lifestyle and medical history. Trained investigators evaluated all clinical data. The inclusion criterion was voluntary participation in this study. Individuals were excluded based on the following criteria: (1) inability to perform the handgrip strength test or the walking speed test; (2) inability to stand for measurement of body composition; (3) inability to communicate with the study staff or provide informed consent; (4) severe hearing impairment or blindness; and (5) severe dementia.

A total of 2308 individuals participated in our study. Of these, 87 were missing AGE data, 38 were missing bone mineral density (BMD) data, 4 had abnormal AGE values, 47 were missing grip-strength tests, 12 were missing speed data, and 127 were missing body composition data. Therefore, we excluded these individuals and ultimately included 1991 participants for analysis (Fig. 1). This study was conducted in accordance with the Declaration of Helsinki, and participants provided full and informed written consent to participate in the study; ethical approval was obtained from the ethics committee of Shanghai University of Medicine and Health Sciences (protocol code: 2019-WJWXM-04-310108196508064467; April 10, 2019).

Figure 1.

Flowchart of participants included in the study.

Assessment of Osteoporosis

BMD was measured using dual-energy x-ray absorptiometry (DXA), which is widely regarded as the gold standard for diagnosing osteoporosis. Specifically, we employed the EXA-3000 bone resorption measurement system (Osteosis Co Ltd) to measure BMD (g/cm²) in the distal third of the ulna and radius of the patient's nondominant forearm, which was prewarmed and calibrated prior to testing. The fixation procedure was performed by a doctor certified in international clinical bone density measurements. Osteoporosis is defined as having a T score less than −2.5, as recommended by the World Health Organization criteria (21).

Assessment of Sarcopenia

Sarcopenia was defined according to the diagnostic criteria set forth by the Asian Working Group for Sarcopenia in 2019 (1). Presarcopenia was defined as having a low appendicular skeletal muscle index (ASMI) with normal handgrip strength and normal gait speed (22). Muscle mass was measured using a direct segmental multifrequency bioelectrical impedance analysis (BIA, In-Body720; Biospace Co Ltd). Low muscle mass was defined as having an ASMI lower than 7.0 in men and 5.7 in women. ASMI was calculated by dividing the appendicular lean mass by the square of body height (kg/m²). Muscle strength was assessed using grip strength measured with a hydraulic hand dynamometer (GRIP-D; Takei Ltd). Participants were asked to exert maximum effort twice using their dominant hand, and the average values of the 2 attempts were used for analysis (23). Walking speed was measured over a 4-m distance at the participants’ usual pace (24). Photocells were used to calculate the time between the activation of the first and second photocells, and the average time from 2 trials was used for analysis.

Measurement of Advanced Glycation End Products With Skin Autofluorescence

AGEs were measured using an AGE Reader (DiagnOptics Technologies BV). Skin autofluorescence (SAF) is considered a marker of long-term AGE burden due to the long half-life of AGEs (25), and the accuracy of this noninvasive technique has been validated by comparing it to the tissue levels of carboxymethyl-lysine (CML) and pentosidine, which are 2 major AGE molecules (26). Briefly, a small area of forearm skin, approximately 4 cm², was irradiated with an AGE Reader using a peak wavelength of 370 nm. The AGE Reader estimates AGEs by converting the emission and reflection spectra of the skin into values reported in arbitrary units (AUs) through software programs. The automation software in the AGE Reader ensures that skin reflectivity values fall between 6% and 10% of the SAF values and excludes participants with skin reflectivity below 6%. SAF values beyond the range of mean + 4 SDs were considered abnormal and were excluded from the analysis (27). The age-adjusted AGE value was used for analysis (28) because the level of AGEs is significantly affected by age.

Covariates

All participants were invited to attend a face-to-face interview. Data on lifestyle factors were obtained through a standardized questionnaire and included age, sex, education level, living situation, smoking habits, drinking habits, marital status, and depression. Medical history was also recorded, including diabetes, hypertension, stroke, dyslipidemia, and kidney disease. These diseases were defined based on self-reporting of a physician's diagnosis or the use of prescribed medications. Diabetes was assessed using a clinically standardized test (fasting blood glucose ≥126 mg/dL) and diagnosed by a clinician. Participants were considered to have hypertension if they reported taking antihypertensive medication or had an average blood pressure of 140/90 mm Hg or higher during the clinical examination. Total cholesterol, triglycerides, high-density lipoprotein cholesterol, and low-density lipoprotein cholesterol levels were measured using a Roche Modular P Chemistry Analyzer (Roche Diagnostic Company). Height and weight were recorded with participants in a standing position without shoes, and body mass index (BMI) was calculated. Depression was assessed using the 30-item Geriatric Depression Scale; a score of 11 or greater is considered indicative of depression. Physical activity was assessed using the short form of the International Physical Activity Questionnaire (IPAQ). Nutritional status was assessed through the Mini-Nutritional Assessment. Based on Mini-Nutritional Assessment scores, nutritional status was categorized as follows: well-nourished (> 24 points), at risk of malnutrition (17-24 points), and malnutrition (<17 points).

We used both theory-driven and data-driven approaches to identify potential confounding factors. First, in the theory-driven part, we identified 6 potential confounding factors, including age, sex, BMI, smoking, diabetes, and nutritional status, based on the literature and directed acyclic graphs (29). Second, in the data-driven part, we used regression analysis to identify potential confounding factors that showed statistical significance. Specifically, we used univariate linear regression and univariate logistic regression to explore the relationship between variables with group differences and AGEs and sarcopenia. Variables that had a statistically significant effect (P < .05) on both AGEs and sarcopenia and showed no interaction were considered potential confounding factors (Supplementary Table S1) (30). Using this approach, we identified 7 potential confounding factors: age, BMI, nutritional status, IPAQ, hyperlipidemia, and mild cognitive impairment (MCI). Age and sex were not included in the regression model as confounding factors, as the values of AGEs were already adjusted for sex stratification and age. In summary, we identified 7 variables as potential confounding factors, including BMI, smoking, diabetes, nutritional status, IPAQ, hyperlipidemia, and MCI. Model 1 was adjusted for BMI, smoking, diabetes, and nutritional status, while model 2 was adjusted for all identified confounding factors.

Statistical Analysis

Histograms and the Shapiro-Wilk test were used to assess the normality of the variables. Normally distributed variables were expressed as the mean ± SD, while nonnormally distributed variables were expressed as the median (interquartile range). Categorical variables were presented as absolute values and percentages (%) of the total. For normally distributed variables, one-way analysis of variance was used to compare the mean values among different groups. For nonnormally distributed variables, the Mann-Whitney Wilcoxon test was used. The chi-square test was used to compare categorical variables. Dummy variables were created for the AGE level groups, with the lowest group defined as the reference category. Binary logistic regression was performed to investigate the association between AGEs and sarcopenia. P for trend was calculated using the median of the quartile group as a quasicontinuous variable in the model. Moreover, to assess the effect of high AGEs on the prevalence of individual sarcopenia, we calculated the population attributable fraction (PAF). The PAF represents the proportionate reduction in the population incidence that would have occurred if all participants’ AGEs were reduced to the Q1 level (<1.36 AU).

We conducted a series of sensitivity analyses to test the robustness of our results. First, we acknowledged that physical activity plays a crucial role in influencing AGEs and sarcopenia. Increased physical activity levels can lower AGEs (31, 32) and enhance muscle mass and strength to mitigate sarcopenia (33, 34). To address possible reverse causality, we examined whether the association would change when participants with lower levels of physical activity were excluded. Second, we considered that diet-derived AGEs significantly contribute to the body's AGE pool (35) and dietary protein intake is an important measure for the treatment of sarcopenia (36). Consequently, we reexamined the association by selecting only participants with good nutritional status. Third, due to the importance of sex and diabetes as factors associated with AGEs (11, 37), we stratified the participants based on sex and diabetes status. Fourth, we performed multiple imputation by chained equations on the missing covariates, estimating 5 imputation values for each missing value to form 5 data sets, and then using each of the 5 data sets to perform the logistic regression analysis of AGEs and sarcopenia. Finally, the results of the 5 regressions were pooled. Additionally, we conducted propensity score matching to assess outcomes between the high-AGE and low-AGE groups. One-to-one matching to the nearest neighbor was performed based on BMI, smoking, diabetes, nutritional status, IPAQ, hyperlipidemia, and MCI.

To examine the mediating effect of osteoporosis on the relationship between AGEs and sarcopenia, AGEs were included in the logistic regression model as both continuous and categorical variables. When AGEs were treated as continuous variables, we used the product of the regression coefficients of path a and path b (denoted as a*b) to assess the mediation effect. This involved inputting the standardized regression coefficients and SEs of path a and path b into the “RMediation” software package (38), and calculating the estimated value and 95% CI of a*b. A significant value of a*b (95% CI excluding zero) indicated the presence of a mediating effect. Alternatively, when AGEs were considered as a categorical variable, we employed Iacobacci's method (39) to test the mediating effect of osteoporosis. Statistical analyses were conducted using SPSS v 25.0 (SPSS Inc) and RStudio (R version 4.1.2), and P values less than .05 were considered to indicate statistical significance.

Results

Characteristics of the Participants

The demographic and skeletal muscle-specific characteristics of the total population are presented in Table 1, categorized into normal, presarcopenia, and sarcopenia groups. The average age of the 1991 participants was 72.37 ± 5.90 years, with males accounting for 42.9% of the population. The mean value of age-adjusted AGE levels was 1.67 ± 0.44 AU. The prevalence of osteoporosis was 40.5% (n = 807). The prevalence of presarcopenia was 8.3% (n = 165), and sarcopenia was found in 18.5% (n = 368) of the participants. With the progression of sarcopenia, participants showed trends of higher AGE values, older age, increased prevalence of osteoporosis and depression, and decreased ASMI and BMD. Additionally, a high prevalence of hypertension (64.9%) and hyperlipidemia (39.5%) was observed among the participants.

Table 1.

Characteristics for our total cohort and based on sarcopenia status

Variables	Total (N = 1991)	Normal (N = 1458)	Presarcopenia (N = 165)	Sarcopenia (N = 368)	P
Age, y	72.37 ± 5.90	71.59 ± 5.40	71.60 ± 4.88	75.77 ± 6.92a,b	<.001
Sex, %
Male	854 (42.9)	637 (43.7)	68 (41.2)	149 (40.5)	.488
Female	1137 (57.1)	821 (56.3)	97 (58.8)	219 (59.5)
BMI	23.78 ± 3.33	24.80 ± 3.01	20.40 ± 2.35a	21.27 ± 2.45a,b	<.001
Widowed, %	313 (15.7)	223 (15.3)	21 (12.7)	69 (18.8)	.145
Living alone, %	273 (13.7)	196 (13.4)	20 (12.1)	57 (15.5)	.491
Farmer, %	565 (28.4)	407 (27.9)	45 (27.3)	113 (30.7)	.539
Smoking, %	239 (12.0)	175 (12.0)	25 (15.8)	38 (10.3)	.204
Drinking, %	572 (28.7)	433 (29.7)	46 (27.9)	93 (25.3)	.237
Fall, %	328 (16.5)	239 (16.4)	25 (15.2)	64 (17.4)	.802
Family income, RMB/mo, %					.966
<3000	730 (37.7)	536 (37.8)	61 (37.4)	133 (37.5)
3000-8000	1189 (61.4)	868 (61.3)	101 (62.0)	220 (62.0)
>8000	16 (0.8)	13 (0.9)	1 (0.6)	2 (0.6)
Education level, %					.009
Primary	775 (37.9)	546 (37.4)	50 (30.3)	159 (43.2)b
Lower	577 (29.0)	433 (29.7)	59 (35.8)	85 (23.1)a,b
Intermediate	362 (18.2)	259 (17.8)	38 (23.0)	65 (17.7)
Higher	297 (14.9)	220 (15.1)	18 (10.9)	59 (16.0)
Nutritional status, %					<.001
Well-nourished	1645 (85.9)	1282 (91.5)	115 (72.3)	248 (69.9)a
Risk of malnutrition	238 (12.4)	102 (7.3)	40 (25.2)a	96 (27.0)a
Malnutrition	32 (1.7)	17 (1.2)	4 (2.5)a	11 (3.1)a
AGE, AU	1.67 ± 0.44	1.63 ± 0.42	1.68 ± 0.39	1.81 ± 0.51a,b	<.001
AGE quartiles					<.001
Q1	501 (25.2)	405 (27.8)	31 (18.8)a	65 (17.7)a
Q2	502 (25.2)	384 (26.3)	47 (28.5)	71 (19.3)a
Q3	492 (24.7)	346 (23.7)	44 (26.7)	102 (27.7)
Q4	496 (24.9)	323 (22.2)	43 (26.1)	130 (35.3)a
BMD, g/cm2	0.39 ± 0.09	0.40 ± 0.09	0.38 ± 0.08a	0.36 ± 0.09a,b	<.001
T-score	−2.10 ± 1.12	−1.95 ± 1.08	−2.31 ± 1.06a	−2.58 ± 1.14a,b	<.001
ASMI	6.72 ± 0.99	7.05 ± 0.86	5.94 ± 0.64a	5.76 ± 0.73a,b	<.001
Grip strength, kg	23.87 ± 8.22	24.95 ± 8.32	26.21 ± 6.65a	18.52 ± 6.03a,b	<.001
Gait speed, m/s	1.06 ± 0.23	1.07 ± 0.23	1.2 ± 0.15a	0.93 ± 0.23a,b	<.001
TUGT, s	7.94 (6.84-9.44)	7.87 (6.81-9.19)	7.44 (6.41-8.12)	8.79 (7.37-11.41)	<.001
SPPB, score	12 (10-12)	12 (11-12)	12 (11.5-12)	11 (9-12)	<.001
IPAQ, %					.001
Lower	654 (34.2)	454 (32.5)	47 (29.4)	153 (43.3)a,b
Intermediate	645 (33.8)	478 (34.2)	56 (35.0)	111 (31.4)
Higher	611 (32.0)	465 (33.3)	57 (35.6)	89 (25.2)a,b
Laboratory examination
Glucose, mmol/L	5.91 ± 1.63	5.95 ± 1.73	5.75 ± 1.38	5.83 ± 1.28	.293
SBP, mm Hg	130 ± 19	130 ± 19	127 ± 18a	132 ± 21b	.015
DBP, mm Hg	74 ± 11	74 ± 11	72 ± 12	74 ± 13	.278
TCH, mmol/L	5.28 ± 1.07	5.29 ± 1.06	5.40 ± 1.16	5.19 ± 1.09	.174
TGs, mmol/L	1.53 ± 1.04	1.58 ± 1.08	1.48 ± 0.89	1.36 ± 0.92a	.008
HDL-C, mmol/L	1.39 ± 0.33	1.37 ± 0.33	1.43 ± 0.32	1.43 ± 0.35	.017
LDL-C, mmol/L	3.05 ± 0.87	3.06 ± 0.85	3.11 ± 1.01	2.97 ± 0.86	.243
Chronic conditions, %
Osteoporosis	807 (40.5)	512 (35.1)	81 (49.1)a	214 (58.2)a	<.001
Diabetes	404 (20.3)	296 (20.3)	26 (15.8)	82 (22.3)	.223
Hypertension	1293 (64.9)	969 (66.5)	95 (57.6)	229 (62.2)	.037
Hyperlipidemia	777 (39.5)	585 (40.5)	70 (43.2)	122 (34.0)	.048
Depression	259 (13.0)	172 (11.8)	20 (12.1)	67 (18.2)a	.005
MCI	187 (9.4)	127 (8.7)	14 (8.5)	46 (12.5)	.077
Stroke	407 (20.4)	296 (21.0)	23 (14.2)	88 (24.9)b	.022
CHD	517 (26.0)	391 (27.7)	34 (21.0)	92 (26.0)	.172
Kidney disease	125 (6.5)	98 (7.0)	6 (3.7)	21 (5.9)	.253
Medicine use					.001
0	389 (20.3)	257 (18.4)	51 (31.7)a	81 (22.9)
1	479 (25.0)	345 (24.7)	42 (26.1)	92 (26.1)
2	402 (21.0)	296 (21.2)	29 (18.0)	77 (21.8)
≥3	643 (33.6)	501 (35.8)	39 (24.2)a	103 (29.2)

Data are presented as mean ± SD, median (interquartile range), and number (%).

Abbreviations: AU, arbitrary units; ASMI, appendicular skeletal mass index; BMI, body mass index; CHD, coronary heart disease; DBP, diastolic blood pressure; HDL-C, high-density lipoprotein cholesterol; IPAQ, International Physical Activity Questionnaire; LDL-C, low-density lipoprotein cholesterol; MCI, mild cognitive impairment; SBP, systolic blood pressure; SPPB, Short Physical Performance Battery; TCH, total cholesterol; TGs, triglycerides; TUGT, Timed Up and Go Test.

^a P less than .05 compared with normal.

^b P less than .05 compared with presarcopenia.

AGEs and sarcopenia are highly influenced by aging (37), and the AGE values were significantly higher in males than in females (1.81 ± 0. 48 AU vs 1.57 ± 0.38 AU; P < .001). Therefore, we conducted a sex-stratified, age-adjusted analysis of AGEs and presented the participants’ characteristics in quartiles, as shown in Supplementary Table S2 (30). The mean AGE value, mean age, prevalence of hyperlipidemia, and MCI of the participants increased gradually from the first quartile to the fourth quartile, while ASMI, grip strength, and BMD level decreased gradually.

Association of Advanced Glycation End Products With Sarcopenia and Presarcopenia

Table 2 presents the results of the binary logistic regression analysis, describing the association between AGEs and the occurrence of presarcopenia and sarcopenia. After adjusting for potential confounders such as BMI, smoking, diabetes, and nutritional status, AGEs were found to be significantly associated with sarcopenia (odds ratio [OR] 1.44 per 1 SD; 95% CI, 1.26-1.66), but not with presarcopenia (OR 1.05; 95% CI, 0.85-1.30). When comparing the highest AGE quartile (Q4) to the lowest AGE quartile (Q1), a significant association with sarcopenia was found (OR 2.42; 95% CI, 1.60-3.66), whereas no significant association was observed with presarcopenia (OR 1.39; 95% CI, 0.76-2.51). Additionally, we included the median of each AGE category as a continuous variable in the multivariable regression model to test for linear trends. A linear trend was observed between AGEs and sarcopenia (P for trend <.001), but not between AGEs and presarcopenia (P for trend = .337). Additionally, we calculated the PAF for sarcopenia cases was significantly associated with high AGE values. We observed that 29.8% of sarcopenia cases would theoretically be prevented if the AGE values of all participants decreased to less than 1.36 AU (Q1), assuming a causal relationship.

Table 2.

Logistic regression analysis of advanced glycation end products and presarcopenia and sarcopenia

AGEs, AU	Presarcopenia						Sarcopenia
	Crude		Model 1		Model 2		Crude		Model 1		Model 2
	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P	OR (95% CI)	P
Q1 (<1.36)	Reference		Reference		Reference		Reference		Reference		Reference
Q2 (1.36-1.61)	1.60 (0.99-2.57)	.053	1.33 (0.76-2.33)	.313	1.36 (0.77-2.40)	.284	1.15 (0.80-1.66)	.446	1.02 (0.67-1.55)	.946	1.11 (0.72-1.73)	.633
Q3 (1.61-1.92)	1.66 (1.03-2.69)	.039	1.39 (0.79-2.43)	.255	1.35 (0.77-2.39)	.299	1.84 (1.30-2.59)	.001	1.55 (1.04-2.29)	.030	1.61 (1.06-2.44)	.025
Q4 (>1.92)	1.74 (1.07-2.82)	.025	1.49 (0.83-2.65)	.178	1.39 (0.76-2.51)	.282	2.51 (1.80-3.49)	<.001	2.31 (1.57-3.41)	<.001	2.42 (1.60-3.66)	<.001
P for trend		.038		.199		.337		<.001		<.001		<.001
Per 1 SD increase	1.13 (0.96-1.33)	.145	1.09 (0.89-1.34)	.409	1.05 (0.85-1.30)	.626	1.45 (1.30-1.62)	<.001	1.46 (1.28-1.67)	<.001	1.44 (1.26-1.66)	<.001

1 SD of AGEs = 0.44 AU. Model 1: adjusted for body mass index, smoking, diabetes, nutritional status; model 2: adjusted for body mass index, smoking, diabetes, nutritional status, International Physical Activity Questionnaire, hyperlipoidemia, and mild cognitive impairment.

Abbreviations: AGEs, advanced glycation end products; AU, arbitrary unit; OR, odds ratio.

Sensitivity Analysis

We conducted sensitivity analysis by excluding participants with low IPAQ scores to examine the association between AGEs and presarcopenia, as well as sarcopenia. The results were comparable to those obtained in the full cohort population. In model 2, the OR of AGEs for sarcopenia showed a slight increase from 2.42 (95% CI, 1.60-3.66) to 2.53 (95% CI, 1.49-4.28) (Supplementary Table S3) (30). In the sensitivity analysis, which included only participants with well-nourished status, AGEs remained significantly associated with sarcopenia (OR 2.80; 95% CI, 1.79-4.39), while no significant association was found with presarcopenia (OR 1.87; 95% CI, 0.96-3.66). The results were consistent, and no significant changes were observed (Supplementary Table S4) (30). In the stratified analyses by sex and diabetes, the associations of AGEs with presarcopenia and sarcopenia remained consistent (Supplementary Table S5) (30). The OR value of AGEs for sarcopenia was higher in men (3.45 vs 1.99) and patients with diabetes (2.86 vs 2.31). To address missing data in covariates, we employed multiple imputation with chained equations. The imputed pooled results were found to be in alignment with the main results (Supplementary Table S6) (30). In the subsequent sensitivity analysis, we employed the median of AGEs as a cutoff to categorize participants into high- and low-AGE groups. Subsequently, we conducted one-to-one matching between the high-AGE and low-AGE groups, considering factors such as BMI, smoking, diabetes, nutritional status, IPAQ, hyperlipidemia, and MCI (N = 1528). The characteristics of the participants before and after matching are presented in Supplementary Table S7 (30). Separate logistic regression analyses for AGEs and presarcopenia, as well as sarcopenia, were performed before and after matching, respectively. Both prematching and postmatching results exhibited similarity to those observed in the full cohort population, with a slight upward trend in the OR value in the postmatching regression analysis (1.68 vs 1.64) (Supplementary Table S8) (30).

Analysis of Mediating Effect

Additionally, we conducted a mediation analysis to explore the influence of osteoporosis on the association between AGEs and sarcopenia. Table 3 presents the results of the mediation effect. When AGEs were treated as a continuous variable, we observed a statistically significant indirect effect of AGEs on the risk of sarcopenia through osteoporosis (Z_a*Z_b = 18.81; 95% CI, 8.07-32.32) (as the CI does not include zero), with a mediating proportion of 39.1% for osteoporosis. When AGEs were considered a categorical variable, osteoporosis still played a partial mediating role (Z_mediation = 3.01 > 1.96; P < .05) in the association between AGEs and sarcopenia, and the proportion of the mediating effect was 45.7%.

Table 3.

Mediation of osteoporosis on the association between advanced glycation end products and sarcopenia

Dependent variable (Y)	Mediator (M)	Independent variable (X)	X →Y (c path)	X + M → Y (c′path)	X→M (a path)	M→Y (b path)	Zmediation	95% CI	Proportion of effect mediated
Sarcopenia	Osteoporosis	AGEs	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
		Continuous variable	0.828 (0.160)	0.768 (0.160)	0.532 (0.118)	0.609 (0.146)	18.81	8.07-32.32	39.1%
		Q1 (<1.36)	Reference	Reference	Reference	Reference
		Q2 (1.36-1.61)	0.107 (0.224)	0.087 (0.226)	0.122 (0.148)	0.602 (0.146)	0.79
		Q3 (1.61-1.92)	0.476 (0.212)	0.405 (0.214)	0.541 (0.146)	0.602 (0.146)	2.71
		Q4 (>1.92)	0.884 (0.211)	0.801 (0.214)	0.671 (0.148)	0.602 (0.146)	3.01		45.7%

The table shows regression coefficients and SEs for each step of mediation analysis after adjusting for body mass index, smoking, diabetes, nutritional status, International Physical Activity Questionnaire, hyperlipidemia, and mild cognitive impairment; bold means P less than .05. c path reports regression coefficients of AGEs with sarcopenia (not adjusted for osteoporosis), representing the total effect of AGEs on sarcopenia; c′ path reports coefficients of AGEs to sarcopenia (adjusted for osteoporosis), representing the direct effect of AGEs on sarcopenia; a path reports coefficients of AGEs to osteoporosis (mediator under examination); b path reports coefficients of osteoporosis (mediator) to sarcopenia. When AGEs were analyzed as a continuous variable, the product of regression coefficients of paths a and b was used to test the mediation, namely, the standardized regression coefficients and SE of path a and path b were written into the RMediation package to calculate the estimates of a*b and 95% CIs. The product of a*b indicates the magnitude of the mediating effect when the value of a*b is statistically significant (the 95% CIs do not contain a zero). When AGEs were analyzed as a categorical variable, Iacobacci's method was applied to examine the mediating effect of osteoporosis. Z_mediation statistics exceeding 1.96 were statistically significant at .05 levels.

Abbreviations: AGEs, advanced glycation end products; Q, quartile.

Furthermore, owing to variations in the pathogenesis of osteoporosis between male and female populations, we conducted a sex-stratified mediation analysis for osteoporosis. The results indicated that the relationship between AGEs and sarcopenia was partially mediated by osteoporosis both in males and females, mirroring the outcomes observed without sex stratification (Table 4). Specifically in the male population, when AGEs were examined as a continuous variable to assess its mediating effect, the value of Z_a*Z_b was 11.94, with a 95% CI of 2.28 to 23.87. Importantly, the CI did not include 0, signifying a significant mediating effect. Osteoporosis contributed to 49.8% of the mediation effect between AGEs and sarcopenia. When AGEs were treated as a categorical variable and compared with the lowest quartile, the mediation effect of osteoporosis in the association between AGEs and sarcopenia remained statistically significant, with Z_mediation = 2.01, which exceeds the threshold of 1.96, and the proportion of the mediating effect was 44.2%. The results in women paralleled those observed in men, with AGEs mediating effects as a continuous variable and categorical variable accounting for 44.2% and 52.3%, respectively.

Table 4.

Sex-stratified analysis of the mediating effect of osteoporosis on the association between advanced glycation end products and sarcopenia

Dependent variable (Y)	Mediator (M)	Independent variable (X)	X →Y (c path)	X + M → Y (c′path)	X→M (a path)	M→Y (b path)	Zmediation	95% CI	Proportion of effect mediated
Sarcopenia	Osteoporosis	AGEs	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)	Coefficient (SE)
Male		Continuous variable	0.939 (0.229)	0.857 (0.230)	0.841 (0.174)	0.556 (0.225)	11.94	2.28-23.87	49.8%
		Q1 (<1.47)	Reference	Reference	Reference	Reference
		Q2 (1.47-1.75)	0.331 (0.357)	0.328 (0.361)	0.027 (0.250)	0.556 (0.225)	0.10
		Q3 (1.75-2.08)	0.842 (0.337)	0.785 (0.341)	0.696 (0.238)	0.556 (0.225)	1.83
		Q4 (>2.08)	1.253 (0.333)	1.173 (0.338)	0.878 (0.236)	0.556 (0.225)	2.01		39.0%
Female		Continuous variable	0.930 (0.247)	0.842 (0.251)	0.748 (0.186)	0.550 (0.201)	11.00	2.63-21.96	44.2%
		Q1 (<1.30)	Reference	Reference	Reference	Reference
		Q2 (1.30-1.52)	0.014 (0.296)	−0.016 (0.298)	0.188 (0.190)	0.592 (0.200)	0.89
		Q3 (1.52-1.77)	0.268 (0.284)	0.187 (0.287)	0.460 (0.191)	0.592 (0.200)	1.81
		Q4 (>1.77)	0.691 (0.283)	0.606 (0.286)	0.611 (0.197)	0.592 (0.200)	2.09		52.3%

The table shows regression coefficients and SEs for each step of mediation analysis after adjusted for body mass index, smoking, diabetes, nutritional status, International Physical Activity Questionnaire, hyperlipidemia, and mild cognitive impairment; bold means P less than .05. c path reports regression coefficients of AGEs with sarcopenia (do not adjust for osteoporosis), representing the total effect of AGEs on sarcopenia; c′ path reports coefficients of AGEs to sarcopenia (adjusted for osteoporosis), representing the direct effect of AGEs on sarcopenia; a path reports coefficients of AGEs to osteoporosis (the mediator under examination); b path reports coefficients of osteoporosis (mediator) to sarcopenia. When AGEs was analyzed as a continuous variable, the product of regression coefficients of paths a and b was used to test the mediation, namely, the standardized regression coefficients and SE of path a and path b were written into the RMediation package to calculate the estimates of a*b and 95% CIs. The product of a*b indicates the magnitude of the mediating effect when the value of a*b is statistically significant (the 95% CIs do not contain a zero). When AGEs were analyzed as a categorical variable, Iacobacci's method was applied to examine the mediating effect of osteoporosis. Z_mediation statistics exceeding 1.96 were statistically significant at .05 levels.

Discussion

The present study revealed a significant association between AGEs and the occurrence of sarcopenia, but not presarcopenia, among older adults residing in Chinese communities. Additionally, we investigated the potential mediating effect of osteoporosis on this association. The mediation analysis revealed that osteoporosis partially mediated the relationship between AGEs and sarcopenia. To the best of our knowledge, this is the first study to explore the mediating effects of osteoporosis on the association between AGEs and sarcopenia in community-dwelling older adults in China.

Various epidemiological studies conducted in Asian countries have reported the prevalence of sarcopenia to range from 5.5% to 25.7% (5.1%-21.0% in men and 4.1%-16.3% in women) (3). In a cross-sectional study conducted in a rural area of eastern China, the prevalence of sarcopenia was found to be 21.7% in women and 12.9% in men (2). The prevalence of sarcopenia in our study was 18.5% in the total population, 17.4% in men, and 19.3% in women, which is consistent with previous studies.

Our study found that after adjusting for potential confounding factors, such as BMI, smoking, diabetes, and nutritional status, the risk of sarcopenia increased with higher AGE quartiles. AGEs were identified as an independent risk factor for sarcopenia (OR 2.42; 95% CI, 1.60-3.66; P < .001) but not for presarcopenia (OR 1.39; 95% CI, 0.76-2.51; P = .282). These findings are consistent with the majority of earlier research.

In a previous study (10), elevated serum pentosidine (a typical and representative AGE) was independently associated with a higher prevalence of sarcopenia (OR 1.01; 95% CI, 1.01-1.02; P < .001) in Chinese middle-aged and elderly men with T2DM. Likewise, another study (9) demonstrated that accumulated AGEs measured by the SAF method (SAF-AGE) were an independent factor for sarcopenia in patients with T2DM. However, it is worth noting that the aforementioned two studies were specifically conducted in patients with T2DM. A recent cross-sectional study investigated the association between SAF-AGE and presarcopenia and sarcopenia in older adults with a northern European background. The study included 2744 participants with a mean age of 74.1 years (11). The study found a significant association between AGEs and sarcopenia, which aligns with our own observation. However, the study also revealed that SAF-AGE was associated with higher odds of presarcopenia (OR 1.75; 95% CI, 1.16-2.63; P < .05), in contrast to our research findings. Two reasons could account for the disparate results. First, the two studies employed different methods for assessing muscle mass (DXA vs BIA). Second, the 2 studies adopted distinct criteria to define low muscle mass (Revised European consensus on Sarcopenia in Older People vs Asian Working Group for Sarcopenia 2019 Consensus). In conclusion, the AGE value is an independent risk factor for sarcopenia in older Chinese individuals living in the community, and SAF may serve as a simple, effective, and noninvasive tool to assess the risk of sarcopenia.

The association between AGEs and sarcopenia can be explained by several hypothetical underlying mechanisms. First, under nonenzymatic conditions, AGEs can form cross-links with free amino groups or other functional groups of macromolecular substances, such as proteins, resulting in the formation of irreversible complexes that lead to changes in protein structure and function. The covalent cross-linking of AGEs leads to the occurrence and development of sarcopenia by hardening the extracellular matrix components of skeletal muscle (40, 41), causing a reduction in the viscoelastic properties of muscle (13, 42), and dysregulating the structure of the basement membrane (43, 44). Second, AGEs have been implicated in the pathophysiology of sarcopenia through their binding to RAGE. This binding stimulates the phosphorylation of nuclear factor-kappa B (NF-κB), upregulates the production of extracellular matrix genes (such as type I collagen) and inflammatory cytokines (45), induces inflammation, and causes endothelial dysfunction through NF-κB transcription, which results in the loss of myocytes and satellite cells (46). Moreover, binding to RAGE activates nicotinamide adenine dinucleotide phosphate oxidase (NADPH oxidase), resulting in an increased level of circulating reactive oxygen species. Subsequently, inflammatory cytokines and reactive oxygen species activate the ubiquitin–proteasome system, accelerating the degradation of muscle proteins (47), thus contributing to the occurrence and development of sarcopenia.

PAF, a statistical indicator, quantitatively characterizes the health effect of a specific risk factor on a population. In essence, it denotes the proportion of total disease (or mortality) within a population that can be linked to a particular factor (48). Recognizing the significant influence of AGEs on sarcopenia, we computed the PAF for sarcopenia that can be attributed to AGEs. Our findings indicate that 29.8% of sarcopenia cases could potentially be prevented if the AGE values of all participants were reduced to less than 1.36 AU (Q1). Consequently, addressing methods to decrease AGE levels in the human body emerges as a pivotal topic with potential benefits for sarcopenia prevention.

As far as our current knowledge goes, dietary sources of AGEs play a significant role in determining the levels of AGEs in the body (35). Furthermore, studies have demonstrated an association between dietary AGE intake and sarcopenia (49). A previous intervention study, focusing on a low AGE dietary intake, reported that a 12-week AGE-restricted diet can reduce serum AGE levels (50). Consequently, reducing dietary AGE intake, possibly through the adoption of a low-temperature cooking diet, represents an effective strategy to decrease AGE levels in the human body. Second, exercise has also demonstrated the ability to decrease AGE levels. Bi-weekly Tai Chi exercise has been reported to significantly lower plasma AGE levels after 12 months (31). Animal experiments have also shown that treadmill training, initiated in late middle age, can reduce the accumulation of AGEs in rats (32). In conclusion, current research indicates that dietary and exercise interventions may represent effective strategies for reducing AGE levels.

Moreover, we conducted mediation analyses and observed that osteoporosis mediated the relationship between AGEs and sarcopenia. In line with our research findings (path a in Fig. 2), previous studies have shown an inverse association between CML (a representative AGE) and lumbar BMD among postmenopausal outpatients (OR 0.84; 95% CI, 0.76-0.93; P < .01) (51), as well as a negative correlation between SAF-AGE and lumbar BMD in patients with T2DM (52). Moreover, SAF-AGE has been found to be positively associated with the prevalence of osteoporosis in patients with T2DM (28). AGEs have been implicated in the pathophysiology of osteoporosis through several different mechanisms. These mechanisms include inducing osteoblast apoptosis via endoplasmic reticulum stress (14) and disrupting the functions of osteoblasts by inducing cell ferroptosis (15). Furthermore, our study indicated an association between osteoporosis and sarcopenia (path b in Fig. 2). In conjunction with our research, a 4-year follow-up study (19) (ROAD study) estimated the incidence of sarcopenia and found that the presence of osteoporosis significantly increased the future risk of sarcopenia (OR 2.99; 95% CI, 1.46-6.12; P = .003), but the presence of sarcopenia did not increase the future risk of osteoporosis (OR 2.11; 95% CI, 0.59-7.59; P = .253). However, another cross-sectional study (20) demonstrated a significant relationship between osteoporosis and sarcopenia, indicating that each is a risk factor for the other. Possible explanations for the inconsistent results are as follows. On the one hand, the mean age of participants in the cross-sectional study was much higher than that in the ROAD study. The mean age of participants in the ROAD study was 70.3 and 72.1 years at the baseline and second surveys, respectively, whereas the average age of the participants in the cross-sectional study was 86.6 years. Osteoporosis and sarcopenia may be more closely associated with increasing age. On the other hand, the diagnosis of sarcopenia may vary due to the use of different cutoff values for grip strength and walking speed, potentially leading to differences in the prevalence of sarcopenia. To our knowledge, findings from various studies (18, 53), including a meta-analysis (54), suggest that sarcopenia represents a risk factor for osteoporosis. Therefore, we believe that there is a bidirectional relationship between sarcopenia and osteoporosis.

Figure 2.

Mediation procedures of osteoporosis between advanced glycation end products (AGEs) and sarcopenia.

Previous studies have identified AGEs and osteoporosis as independent risk factors for sarcopenia and a bidirectional relationship between osteoporosis and sarcopenia (11, 18, 19, 53, 54). However, the specific mechanism by which AGEs affect sarcopenia through osteoporosis and how sarcopenia affects osteoporosis remain unclear. Several potential explanations exist. First, similarities in the pathogenesis of osteoporosis and sarcopenia may arise from the common origin of bone and skeletal muscle from mesenchymal stem cells. In cultured human preadipocytes, adipogenesis has been associated with increased levels of AGEs and RAGE (55). The presence of adipocytes can alter the microenvironment, affecting myogenesis and osteogenesis and producing local adipokines, free fatty acids, and lipids, leading to local lipotoxicity, reduced bone formation, and increased bone resorption (56). Additionally, the secretion of prostaglandin E2 and Wnt3a by osteocytes, as well as osteocalcin and insulin-like growth factor-1 by osteoblasts, along with the secretion of sclerostin by both cell types, may have an effect on skeletal muscle cells and thus lead to sarcopenia (57). Second, patients with osteoporosis are more prone to falls and fragility fractures, resulting in reduced mobility such as bed rest. Physical inactivity and positive energy balance may promote chronic low-grade inflammation, shifting the mesenchymal stem cell profile toward adipogenesis and away from myogenesis. This leads to adipose infiltration into muscle tissue and progressive replacement of myocytes with adipocytes (58). Consequently, osteoporosis may partially mediate the relationship between AGEs and sarcopenia through these pathways. Sarcopenia's effect on osteoporosis may be attributed to skeletal muscle's role as an endocrine organ, influencing bone anabolism through nonmechanical pathways (59). Skeletal muscle releases a range of actin proteins (eg, myostatin, interleukin-6, insulin-like growth factor-1, irisin) through autocrine, paracrine, or endocrine signaling, thereby modulating osteocyte metabolism and ultimately playing a role in osteoporosis pathogenesis (53). In the future, we will continue to conduct research on AGEs, osteoporosis, and sarcopenia to explore the exact mechanism among them.

It is noteworthy that the coexistence of osteoporosis and sarcopenia has been defined as osteosarcopenia (16). Patients with osteosarcopenia face a substantially higher risk of severe morbidity (60). Furthermore, overlapping pathophysiological evidence in osteoporosis and sarcopenia suggests the potential for shared therapeutic approaches (17). Therefore, we believe that the concept of osteosarcopenia is a topic that cannot be ignored. Importantly, how to identify the risk population of osteoporosis combined with sarcopenia simply and effectively is an important measure to prevent osteosarcopenia, and the SAF detection of AGEs provides the possibility for this purpose. Therefore, we propose the inclusion of SAF screening in community-based health checkups for older individuals to assess the potential risk of osteoporosis and sarcopenia. In the future, a large-scale longitudinal study will investigate the correlation between AGEs and osteosarcopenia, as well as the public health implications of AGEs in the early screening of osteoporosis, sarcopenia, and osteosarcopenia.

The strengths of our study lie in its large sample size and comprehensive collection of health information, which enabled the identification of sufficient confounders to minimize bias. However, the present study does have some limitations. First, a cross-sectional design was employed in this study, which precluded the inference of causal relationships between AGEs and sarcopenia. Our conclusions should be considered hypothesis-generating and require confirmation in longitudinal and experimental studies. However, to confirm the robustness of our findings, we conducted a series of sensitivity analyses that included different populations, different disease states, and different statistical methods. The results showed that the findings remained consistent throughout the analyses. Second, SAF-AGE represents the long-term level of AGEs, but it measures the skin's fluorescence only within a specific wavelength, which includes fluorescent AGE compounds and excludes nonfluorescent AGEs. Additionally, we did not directly measure the circulating levels of AGE molecules, such as CML and pentosidine, for comparison with SAF-AGE. Third, we did not assess participants’ dietary habits, which might influence their AGE levels. Fourth, while DXA is the standard method for measuring body composition, the BIA method was employed in this study, which may introduce certain limitations in the analysis of muscle mass. Fifth, despite considering numerous important confounders, we could not fully exclude the possibility of residual confounding. Sixth, in the mediation analysis of category variables involved in this study, all variables are assumed to be manifestly variable, and the mediation effect of explicit variables will be underestimated because the measurement error was ignored. Last, due to the voluntary nature of our study and the exclusion of patients with severe medical conditions who were unable to complete the physical fitness test, there might be some selection bias. Therefore, our study population may not have been sufficiently comprehensive, possibly leading to an underestimation of osteoporosis and sarcopenia prevalence, thus potentially affecting the study results.

Conclusion

Our study investigated the relationship between AGEs and sarcopenia in community-dwelling older adults in China. We found a significant association between AGEs and the occurrence of sarcopenia. Additionally, osteoporosis was found to play a mediating role in this association. Early screening for AGEs and osteoporosis may help to mitigate the future incidence of sarcopenia in older adults.

Abbreviations

AGE

advanced glycation end product

ASMI

appendicular skeletal muscle index

BIA

bioelectrical impedance analysis

BMD

bone mineral density

BMI

body mass index

CML

carboxymethyl-lysine

DXA

dual-energy x-ray absorptiometry

IPAQ

International Physical Activity Questionnaire

MCI

mild cognitive impairment

PAF

population attributable fraction

Q

quartile

RAGE

receptor for advanced glycation end product

SAF

skin autofluorescence

T2DM

type 2 diabetes mellitus

Funding

This work was supported by the Capacity Building Project of Local Colleges of Shanghai Science and Technology Commission (grant No. 23010502800), the Shanghai Sailing Program (grant No. 22YF1417900), and the Shanghai Municipal Health Commission (grant No. 202240367).

Author Contributions

X.Z. and X.C. wrote the manuscript. X.Z., S.L., M.G., and P.H. conceived the concept and design of the study. X.Z., S.L., M.G., L.C., J.G., Q.T., and J.Z. collected and assembled the data. X.C., D.L., and L.Q. analyzed and interpreted the data. Q.G. revised the article critically for important intellectual content and provided administrative support. All authors approved the final version.

Disclosures

The authors have nothing to disclose. No competing financial interests exist.

Data Availability

Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.