Sterol metabolism and protein metabolism are differentially correlated with sarcopenia in Asian Chinese men and women

Abstract Objectives Our aim was to investigate the prevalence and predictive variables of sarcopenia. Methods We recruited participants from the Peking Union Medical College Hospital Multicenter Prospective Longitudinal Sarcopenia Study (PPLSS). Muscle mass was quantified using bioimpedance, and muscle function was quantified using grip strength and gait speed. Logistic regression revealed the relationships between sarcopenia and nutritional, lifestyle, disease, psychosocial and physical variables. Results The prevalence of sarcopenia and sarcopenic obesity was 9.2%‐16.2% and 0.26%‐9.1%, respectively. Old age, single status, undernourishment, higher income, smoking, low physical activity, poor appetite and low protein diets were significantly associated with sarcopenia. Multiple logistic regression analysis showed that age was a risk factor for all stages of sarcopenia, and participants above 80 years were greater than fivefold more susceptible to sarcopenia, while lower physical activity was an independent risk factor. The optimal cut‐off value for age was 71 years, which departs from the commonly accepted cut‐off of 60 years. Female participants were greater than twofold less susceptible to sarcopenia than male participants. The sterol derivative 25‐hydroxyvitamin D was associated with fourfold lower odds of sarcopenia in male participants. Several protein intake variables were also correlated with sarcopenia. Based on these parameters, we defined a highly predictive index for sarcopenia. Conclusions Our findings support a predictive index of sarcopenia, which agglomerates the complex influences that sterol metabolism and nutrition exert on male vs female participants.


| INTRODUC TI ON
Rapidly ageing populations around the world are experiencing an increase in muscle wasting syndromes. Sarcopenia is a progressive skeletal muscle wasting disorder that is associated with an increased likelihood of adverse outcomes, including falls, fractures, physical disability and mortality. 1 Sarcopenia has been formally recognized as a muscle disease with an ICD-10-MC Diagnosis Code in some countries. 2 In 1998, following the recommendation by Baumgartner et al, 3 sarcopenia was defined with a cut-off of a skeletal muscle index (SMI; appendicular skeletal muscle mass [ASM]/height 2 ) that is more than two standard deviations (SD) below the mean for young and healthy adults. Subsequently, several regions including Europe, USA and Asia incorporated decreased physical performance as a diagnostic criterion. 4 In 2010, the European Working Group on Sarcopenia in Older People (EWGSOP) published a sarcopenia definition that is now used worldwide. 5 In 2014, the Asian Working Group on Sarcopenia (AWGS) further developed the EWGSOP-based consensus by specifying cut-off points for the diagnostic variables in Asians. 6 Based on these criteria, the prevalence of sarcopenia was estimated to be 11.3% in women and 9.7% in men. 7 Despite these efforts, reports on the prevalence of sarcopenia continue to vary widely between studies (10%-50%), and they are difficult to compare because of the wide variance depending on the country of origin, the methods used and the diagnostic criteria. 8 Several factors can influence muscle mass and strength, including muscular disuse and age-related alterations in sex hormones, protein synthesis, proteolysis, neuromuscular integrity, endocrine function, nutritional balance and intramuscular fat content. 9 Moreover, few studies have systematically surveyed the interactions between sarcopenia and all nutrient groups holistically, and even fewer studies focus on old adults. 10 Sarcopenia has become the focus of intense research aiming to translate current knowledge about its pathophysiology into improved diagnosis and treatment, with particular interest in the development of biomarkers, nutritional interventions and drugs to become part of routine. 11 Designing effective preventive strategies that people can apply during their lifetime is of primary concern. Hence, there is an urgent need to collect and report comprehensive data according to the best consensus criteria, to determine the cut-off points for the appropriate diagnostic variables in Asian Chinese.
To address these limitations, our first aim was to determine the prevalence of sarcopenia in Asian Chinese male and female, using different established diagnostic criteria for skeletal muscle mass, namely EWGSOP and AWGS. Secondly, this study aimed to evaluate the association(s) between sarcopenia and common chronic illnesses, lifestyle factors, psychosocial well-being and dietary nutrition patterns (including protein intake and sterol metabolism), in order to identify risk ≤0.8 m/s. Pre-sarcopenia was defined as low muscle mass, 5 probable sarcopenia was defined as low muscle strength, 1 and severe sarcopenia was defined as the presence of reduced muscle mass, strength and performance. 1,5 Muscle mass was measured by using a segmental multifrequency bioelectrical impedance analysis (M-BIA) instrument that operated at frequencies of 1, 5, 50, 250, 500 and 1000 kHz (H-Key350, Beijing Seehigher Technology Co., Ltd). Hand grip strength was measured by using an electronic hand dynamometer (CAMRY MODEL EH101, HaNDCReW). Physical function was assessed by the average walking speed over a 4-m distance. 5 The details of muscle mass and function measure were referenced as our previous study method part. 12 The abdominal circumference (AC) was measured midway between the lateral lower rib margin and the superior anterior iliac crest at the end of a gentle expiration phase. CC was measured on the left leg in a seated position with the knee and ankle at right angles, feet resting on the floor. Mid-upper arm circumference (MAC) was measured with anon-stretchable measuring tape at a point equidistant between the acromion process of the left scapula and the olecranon process of the left ulna.

| Data collection
Face-to-face interviews were conducted to complete a standardized, structured questionnaire to obtain information. The questionnaire used in the cross-sectional study was developed specifically based on the Korea National Health and Nutrition Examination Survey (KNHANES) 13 and combined with multidisciplinary expert discussion. The reliability, validity and acceptability of the questionnaire were analysed by a pilot study. The alpha coefficient was 0.6, the recovery was 96%, and the response rate was 95%. The time taken to complete the data collection ranged from 18.0 to 29.0 minutes depending on the participant's capacity to complete measurements, with an average of 15.0 ± 7.0 minutes across all subjects.
Demographic characteristics and lifestyle data were ascertained by an interviewer who administered the questionnaire at baseline.
Occupations were classified into several major categories according to labour intensity and level of education. We defined participants to have a smoking habit if they had smoked more than 100 cigarettes and still smoked one pack (20 cigarettes) at least per month for more than 6 months. Alcohol intake was assessed by asking participants whether they were non-drinkers, drank once a month, drank once a week and drank every day. The International Physical Activity Questionnaire (IPAQ) was used to evaluate the level of physical activity for all participants. 14 The medical history, including the presence of diabetes, hypertension, hyperlipidaemia, was assessed by referring to the self-reported physician's diagnosis. Activities of daily living were assessed using the Barthel index, 15 and nutritional status was evaluated using the Mini-Nutritional Assessment (MNA). 16 Information on quality of life was obtained using the 5-dimensional EuroQol (EQ-5D). 17 A trained interviewer asked each participant to report the frequency and the usual amount of consumption of each food item over the past year.

| Vitamin D and testosterone measurements
Serum levels of 25-hydroxyvitamin D (25OHD, including 25OHD 2 and 25OHD 3 ) and testosterone were measured at the Department of Clinical Laboratory (PUMC Hospital, China). The level of serum 25OHD was measured using liquid chromatography-tandem mass spectrometry (LC-MS/MS) system according to the previous reports. 18 Total testosterone and sex hormone binding globulin (SHBG) levels were measured using an automated chemiluminescence immunoassay analyser (Beckman Coulter UniCel DXI 800, Beckman Coulter) using the corresponding reagents, calibration materials and quality control materials. The level of albumin (ALB) was measured using an automated chemistry analyser (Beckman Coulter AU5800, Beckman Coulter).

| Data analysis
Data were analysed by using the statistical software EPIDATA 3.0. Analyses were performed by using SAS21.0.1 (SAS Institute).
Continuous variables were summarized as means ± SD or medians (25th, 75th percentiles), and categorical variables were summarized as counts and percentages. Prevalence was based on a proportion of cases of sarcopenia among total study population. Subgroup analyses were conducted on the prevalence of sarcopenia based on demographics, lifestyle factors, and functional and clinical variables. The comparisons between groups were analysed using the chi-squared test, Fisher exact test and Mann-Whitney U test, where appropriate. We performed analysis of covariance to verify interrelationships between reduced muscle mass and related changes in physical function, and analysis of associations between nutritional parameters and BMI and muscle strength using Spearman's rank correlation. Multiple comparisons were made by the Nemenyi test.
Conditional forward stepwise multiple logistic regressions were used to analyse the factors associated with the risk components.
Most of the variables were categorized into two levels based on the median, while levels were subdivided into three levels based on the upper and lower quartile, to obtain the appropriate likelihood statistical power. The highest level was regarded as the reference group.
The models included demographic variables, lifestyle variables, chronic conditions, anthropometric variables, dietary and nutritional variables. Non-significant variables were omitted from models of the multiple logistic regression analyses to obtain the odds ratio (OR) and 95% confidence interval (CI). Receiver operating characteristic (ROC) analysis was performed to explore the cut-off values of AC, MAC, CC, fat mass, hand grip strength and BMI for men and women, and to verify the predictive validity for sarcopenia. To eliminate the multicollinearity in establishing predictive model at the greatest extent, correlation analysis including variance inflation factor, tolerance, system of eigenvalues and Spearman's rank correlation was performed before the multivariable analysis. Conditional forward stepwise multiple logistic regressions were used again to establish the predictive model. Finally, Hosmer and Lemeshow tests were used to evaluate the exact of two predictive models. Differences were considered significant at P < .05.
For sample size calculations, we took previous AWGS-based consensus sarcopenic prevalence estimates of 7.3% from a study of Chinese participants, 19 with an error of 0.15P and an α level of 5% for the main variable, and it was estimated that 2260 adults would be required for this study. With allowance for a dropout for 20%, >2712 adults would meet the demand for sample size.

| Participant inclusion criteria
The flow chart for participant inclusion and exclusion in the study is shown in Figure 1. In total, 3586 participants were recruited during the data collection, of which 211 participants were considered ineligible to participate (73.5% subsequently refused to participate or failed to obtain guardian consent, 7.6% had cognitive dysfunction, and 3.8% had a pacemaker). Of the 3375 participants who finished the baseline examination and registration in PPLSS, 27 had communicable disease, 62 had received major surgery within the past 6 months, and 76 were diagnosed with Parkinson's disease, rheumatism or other diseases that might influence the results of the study.
Of these, 3210 participants had finished body composition analysis and physical function evaluation. The records from 3090 participants were eventually considered complete, eligible and suitable for further analysis.

| Baseline characteristics
The mean age of study population was 69.3 ± 7.7 years and ranged from 60 to 94 years. The BMI ranged between 15.1 and 43.0 kg/ m 2 . Relative skeletal muscle mass index (RSMMI) ranged from 5.1 to 9.9 kg/m 2 in male and from 2.5 to 8.5 kg/m 2 in female. The hand grip strength ranged from 9.2 to 67.2 kg in male and from 5.1 to 56.4 kg in female. The walking speed was 0.95 ± 0.35 m/s. 97.3% of the participants were completely independent, 1.7% were slightly dependent, 0.1% were moderately dependent, and 0.7% were severely dependent. 47.6% of the participants had normal nutritional status, 49.6% were at risk of malnutrition, and 2.8% were malnourished.

| Prevalence of sarcopenia and sarcopenic obesity
We considered three clinical definitions of sarcopenia in our study: (a) the Baumgartner definition, (b) the EWGSOP and AWGS (2014) cut-off points and (c) the EWGSOP2 (2019) cut-off points. According to the Baumgartner definition, sarcopenia is present in subjects whose muscle mass fall more than two SD below the young adults' mean values ( According to the EWGSOP and AWGS (2014) definition and cut-off points, sarcopenia is present in subjects with reduced muscle mass and low muscle function (strength or performance).
Hence, 11.6% of our study population presented with sarcopenia, in which 10.3% subjects were men and 12.4% subjects were women (Table 2). We further observed that 10.1% participants had pre-sarcopenia, and 4.7% participants had severe sarcopenia. Men suffered pre-sarcopenia and severe sarcopenia more frequently than women.
According to the EWGSOP2 (2019) cut-off points, only 5.7% of our study population had sarcopenia.
We also considered the prevalence of sarcopenic obesity using four different definitions of sarcopenia (Table 3). According to the Baumgartner definition, the prevalence of sarcopenic obesity was 4.1%, and 5.8%, respectively, based on two definitions of obesity: P 60 of fat percentage and WHO reference fat percentage cut-off points. 20 According to the EWGSOP and AWGS (2014) definition, the prevalence of sarcopenic obesity was 6.0%, and 9.1% respectively, based on the two definitions of obesity. According to the EWGSOP2 (2019) definition, the prevalence of sarcopenic obesity was 3.6%, and 5.8%, respectively. The prevalence of sarcopenic obesity, as defined by BMI, approached zero in both male and female, suggesting that BMI might not be appropriate for defining sarcopenic obesity. The most robust definition for sarcopenic obesity appeared to be based on body fat percentage, ranging from 3.6% to 9.1% for various definitions of sarcopenia. The EWGSOP and AWGS (2014) definition gave the highest percentage of participants with sarcopenic obesity.

| Demographic risk factors for sarcopenia
We chose the EWGSOP and AWGS (2014) definition for further analysis of the risk factors for sarcopenia, because its cut-off points have been optimized with Asians and could most robustly identify The P value indicated the significance between sex in each age group. b The significance was presented among three age group. c The P value indicated the significance between young and middle age adults, the significance was adjusted by .0167. d The P value indicated the significance between young and elderly adults, the significance was adjusted by .0167. Sarcopenic participants tended towards higher incomes (P = .008), living alone without families (P = .001), being single (P < .001) and suffering from malnutrition risk (P < .001), although these associations (Table 4) became less significant after adjustment for other parameters in the multivariate analysis (Table 5).
Although smoking was not associated with sarcopenia in general (Tables 4 and 5), it was more frequent in sarcopenic women (P = .013).
Similarly, while coronary heart disease and hypertension were not associated with sarcopenia in general (Table 4), they were more frequent in sarcopenic men (P = .004) and sarcopenic women (P = .017), respectively. Hyperlipidaemia was associated with non-sarcopenia in general (P = .007), especially in women (P = .002). Osteoporosis and fracture risks were also associated with sarcopenia in general (P = .001), especially in women. Cancer was more frequent in sarcopenic women (P = .016). Exercise intensity (IPAQ) and daily activity level (ADL) also showed similarly curious gender-specific associations.
Interestingly, when adjusted for other parameters in multivariate analysis (Table 5), female participants were greater than twofold less susceptible to sarcopenia than men (OR = 0.589, 95% CI [0.400, 0.868], P = .008). This is well reflected in the prevalence of pre-sarcopenia and severe sarcopenia ( Table 2). These results suggest complex interactions between sarcopenia and gender.

| Gender-associated serum risk factors for sarcopenia
Complex gender-specific associations with sarcopenia behoved us to examine the sex-related sterol hormones more deeply. While elderly women tend to have very low oestradiol levels in general due to menopause, elderly men experience a more gradual drop in testosterone levels at a rate of ~8.2% every 10 years after the age of 30, similar to the rate of muscle decline. Indeed, our study population also reflected a steady decrease in free testosterone with age in men ( Figure 2A). Free testosterone levels were significantly correlated with grip strength (r = .441, P < .001) and muscle mass (r = .375, P = .004) ( Figure 2B,C). These correlations were even stronger if we considered total testosterone, instead of free testosterone ( Figure 2D-F), even though the free (bioactive) testosterone makes up only ~2% to 3% of total testosterone, while the inactive remainder is bound to SHBG or albumin. This suggests that serum testosterone deficiency is more likely to be an effect than a cause for sarcopenia.
Given that another sterol derivative, cholecalciferol or vitamin D, is known to influence SHBG and testosterone levels, we also exam-

| Nutritional and dietary risk factors for sarcopenia
To broadly understand the role of nutrition in sarcopenia, we surveyed the participants' appetite and intake of various food groups, oil, salt, caffeine and vitamins (Table 4). In general, sarcopenic participants had poor appetite (P = .016), lower total and animal protein (P = .033 and .044, respectively), lower nut frequency (P = .008), lower poultry (only women), vegetable (only women) and nut intake (P = .024, P = .001 and P = .015, respectively) than non-sarcopenic participants (Table 4). There were again many gender-specific associations, but both sarcopenic men and women ate less meat and beans ( Figure 4). Sarcopenic women tended to have poor appetite (P = .009), lower total protein intake (P = .005), animal protein intake (P = .018), fish intake (P = .024), poultry intake (P = .038), vegetable intake (P = .003) and nut frequency (P = .032

| Associations between body fat and sarcopenia
Next, we aimed to capture the associations between lean mass, fat mass and other related body composition parameters with sarcopenia ( Table 4). As expected, there were significant correlations ( Figure 5A,B) between RSMMI vs hand grip strength (r = 0.465, P < .001), and walking speed (r = 0.117, P < .001). There was also a significant correlation between hand grip strength and walking speed (r = .225, P < .001; Figure 5C). In contrast, there were significant inverse correlations between fat percentage and hand grip strength (r = −.397, P < .001), walking speed (r = −.161, P = .002) and RSMMI (r = −.218, P < .001; Figure 5D-F). The association between BMI and muscle parameter showed significant difference in RSMMI (r = .465, P < .001) and walking speed (r = −.059, P = .021; Figure 5G,I). The association between BMI and hand grip did not show significance ( Figure 5H). After adjustment for demographic and socioeconomic status (

| Risk factor cut-off points for sarcopenia
For improved diagnosis of sarcopenia, we aimed to find easily measurable anthropometric variables that could be used to replace mus-  Figure S1 ), was actually 71 years (sensitivity, 59.7%; specificity, 73.3%). We compared grip strength, fat mass, BMI, AC, MAC and CC to confirm gender-and age group-specific cut-off points ( Figure 6 and Table 6).
For all elderly men above 60 years, the best cut-off points for

| PUMCHS index for predicting sarcopenia in men and women
Based on the above comprehensive analysis for sarcopenia, gender appeared to play an important role in diagnosis and pathogenesis.
Hence, the predictive model was calculated separately for men and women. In the univariate analysis, the following parameters were In the multivariate analysis using forward conditional stepwise   The P 1 value indicated the significance between different sex in each age group.

TA B L E 2 Prevalence of sarcopenia based on different diagnostic criteria
b The P 2 value indicated the significance of the three sarcopenia stages among different age groups. BMI was significantly correlated with meat intake, meat frequency, milk frequency, poultry intake, vegetable protein, bean intake, nut frequency, fruit intake, grain intake, salt intake and oil intake (all P < .05; Table 7 and Figure 4), to very different degrees for male vs female. In contrast, grip strength was significantly correlated with total protein, animal protein, vegetable protein, meat frequency, meat intake, fish intake, poultry intake, bean frequency, bean intake, vegetable intake, nut frequency, nut intake, fruit intake, grain intake and salt intake (all P < .05; Table 7 and Figure 4), to differing degrees for men vs women. CC was significantly correlated with vegetable protein, meat frequency, milk frequency, bean frequency, grain intake, milk intake, fruit intake, vegetable intake, meat intake, bean intake, salt intake and oil intake (all P < .05, Table 7), to very different degrees for men vs women. Combined with the daily distribution of protein intake and age, the PUMCHS index is a simple yet powerful model to predict the risk for sarcopenia, based on the differential effects of nutrition on elderly men and women.

| D ISCUSS I ON
Among the geriatric participants from the BELFRAIL study, more than half had both reduced grip strength and limited walking speed, but with normal muscle mass. 8  b Almost every day means the frequency more than four times; less once every day means the frequency less than twice every month.
c Almost every day means the frequency more than four times; less once every day means the frequency less than twice every month.
d Almost every day means everyday intake milk; Almost every week means intakes more than three times.  However, determination of the optimal quantity, frequency and subtype of protein intake to preserve muscle mass and function has even greater practical value. Our present findings showed significant associations between sarcopenia and protein intake, especially meat and bean intake, but to varying degrees among men vs women. Meat intake less than once per week doubled the risk for sarcopenia, and sarcopenia risk further doubled with zero meat intake, which is a common practice of vegetarians in China. Interestingly, our data further indicated that sufficient protein intake (1.2 g/kg body weight/day) concentrated in one meal per day reduced the risk of sarcopenia by over threefold, likely through stimulating insulin/IGF-PI3K-mTOR signalling and increasing protein synthesis to conserve muscle mass in older adults. 24 Surprisingly, our data also showed that bean intake has a protective effect on preserving muscle and preventing sarco- nutrition and incorporated correlations with intake of total protein, meat, fish, poultry, bean, vegetable, nut, fruit, grain, salt and fat percentage (all P < .05), but to differing degrees for men vs.
However, as this was a cross-sectional study, the exact causal relationships between muscle mass/function and the various variables could not be precisely determined. While our analyses clearly indicate that several extrinsic variables, such as dietary factors related to the intake of protein, fat, sterols, vitamin D (cholecalciferol) and caffeine, can differentially influence the risk for sarcopenia in men and women, outcome-based clinical trials and human cell experiments will still be needed to confirm the effects of these metabolic perturbations and dietary interventions. It should also be noted that the study subjects excluded from the analyses for various technical reasons were generally older with worse health conditions, which could mean we underestimated the prevalence of sarcopenia in Asian Chinese as a result.
Nevertheless, among the confusing array of muscle parameters used for clinical definitions of sarcopenia, our epidemiological data analyses suggest that gender, sterol metabolism, BMI, grip strength, calf circumstance and age stand out as the most important predictors to consider for improving the accurate standardization of sarcopenia assessment for future studies.

ACK N OWLED G EM ENTS
This material is based upon work supported by the Nutrition