Skeletal muscle loss was associated with the risk of diabetes in non non-alcoholic fatty liver disease Chinese male middle-aged and elderly population, the Shanghai Changfeng Study

Previous studies have presented skeletal muscle loss was associated with diabetes mellitus (DM) and non-alcoholic fatty liver disease (NAFLD). However, whether the presence of NAFLD could inuence the association between skeletal muscle mass and DM was still unknown. The aim of the present study was to investigate the relationship of skeletal muscle mass with diabetes in Chinese middle-aged and older community population, and whether the association could be effected by NAFLD. A cross-sectional study of 5,626 residents aged 45 and above in Changfeng community in Shanghai were conducted. Skeletal muscle mass (SMM) was detected by dual-energy X ray absorption (DXA) and calculated as ASM% [appendicular skeletal muscle mass (ASM) (kg) /body weight*100%]. Liver fat content (LFC) was measured using a quantitative ultrasound method. Multivariate logistic regression analyses were conducted to investigate the association between ASM% quartiles with DM.

With the age increasing, the progressive decreasing in skeletal muscle mass and strength, which was called sarcopenia has signi cantly increased [1] . Skeletal muscle loss reduced the mobility of the elderly, increased the risk of fractures and falls, and meanwhile was closely related to metabolic disorders, tumors and other chronic diseases [2][3][4] . With the aggravation of the aging society, sarcopenia has become an important worldwide public health problem.
As the largest non-fat component of the human body, skeletal muscle accounts for about 40% of body weight and is responsible for most of the postprandial glucose disposition. As an important insulin target organ for glucose uptake and utilization, skeletal muscle loss with insulin resistance would lead to systemic metabolic disorder, which was closely related to diabetes [5][6][7] . Compared with non-diabetics, diabetic patients had lower muscle mass and higher prevalence of sarcopenia [8] . Conversely, reduced skeletal muscle mass may also increase the risk of diabetes [9] . The results in previous few studies about the relationship between skeletal muscle mass and the risk of diabetes were not consistent [9,10] , which may due to different ethnic, gender, age and other population characteristics.
NAFLD is a chronic liver disease caused by abnormal accumulation of fat in the liver. Previous studies have shown type 2 diabetes often coexists with the occurrence and progression of NAFLD. In China, about 28-70% of type 2 diabetes patients have NALFD, and meanwhile 22.5% of NAFLD patients suffered from type 2 diabetes [11,12] . The risk of diabetes also increased with the progression of liver disease [13] . All the above suggested type 2 diabetes was closely related to NAFLD. As muscle and liver are both important target organs of insulin action and effect signi cantly on maintaining glucose homeostasis, which indicated skeletal muscle had an important role on the development of both diabetes and NAFLD. Indeed, a few previous studies have shown that age-related skeletal muscle mass reduction was associated with NAFLD, NASH (non-alcoholic steatohepatitis), and related liver brosis [14][15][16] .
However, so far no research was conducted to explore the relationship between skeletal muscle mass and the risk of diabetes in Chinese community population. Furthermore, whether NAFLD, which was closely related to both skeletal muscle mass and diabetes, would effect signi cantly on this relationship remains unknown. Therefore, it is necessary to carry out a large-scale community population study to investigate the association of skeletal muscle mass, DM and NAFLD.
In the present study, we recruited participants aged 45 and above of Changfeng Community in Shanghai to investigate the relationship between skeletal muscle mass and diabetes, as well as the role of NAFLD involved in the relationship, in order to provide new evidences for the prevention and treatment of NAFLD and diabetes.

Methods
Participants we enrolled a total of 5626 residents aged 45 and above of Changfeng community in Shanghai from May 2010 to December 2012 [17,18] . Participants according with the following criteria were excluded: 1) lacking biochemical and liver fat content data; 2) lacking DXA data; 3) viral hepatitis and excessive alcohol consumption. Finally, 3969 subjects were included in the analysis (1370 men and 2599 women).
All participants were informed of the research details and signed the informed consent. The study was approved by the Ethics Committee of Zhongshan Hospital of Fudan University.

Data collection
Trained researchers interviewed all participants and obtained their medical history by a standard questionnaire. Then, standing height and body weight were measured without shoes and outer clothing.
Body mass index (BMI) was calculated as weight divided by height squared (kg/m 2 ). Resting blood pressure (BP) including systolic blood pressure (SBP) and diastolic blood pressure (DBP) were measured three times with an electronic blood pressure monitor (OMRON Model HEM-752 FUZZY, Omron Co., Dalian, China), and the average was calculated. All the above has been previously described elsewhere [17,18] .
Blood samples were collected after a fasting period of at least 10 hours overnight. Biochemical indexes including FBG, total cholesterol (TC), triglycerides (TG), high-density lipoprotein-cholesterol (HDL-C) were measured with an automated bio-analyzer (HITACHI 7600, Tokyo, Japan). Low-density lipoprotein cholesterol (LDL-C) was calculated using the Friedewald equation. 2hBG was tested following a 75g oral glucose tolerance for non-diabetics or a 100g steamed bread meal for diagnosed diabetes. Electrochemiluminescence immunoassay was used to measure the serum insulin concentrations. HOMA-IR was calculated by multiplying the FBG (mmol/L) times fasting insulin (mU/L) and dividing by 22.5.
Hepatic ultrasonography scanning was performed by an experienced technician who was blinded to the participants' details using a GE Logiq P5 scanner (GE Healthcare, Milwaukee,USA) with a 4-MHz probe.
The liver ultrasound images were analysed with imaging software from the National Institutes of Health (ImageJ 1.41o, National Institutes of Health, Bethesda, MD) and standardised using a tissue-mimicking phantom (Model 057; Computerized Imaging Reference Systems, Norfolk, VA). We measured LFC according to the method described in detail elsewhere [19] .
Body composition including lean mass and fat mass (FM) were measured using dual-energy X-ray absorptiometry (Lunar iDXA, GE Healthcare). All measurements were carried out by a single, trained technician at a single clinical center. Manual DXA analysis software was used to analyze all of the DXA scans. The percentage of fat mass (FM%) was calculated as FM divided by total body mass. SMM was calculated as following: weight adjusted by appendicular skeletal muscle mass (ASM% = appendicular skeletal muscle mass (kg)/body weight (kg) × 100%) [9,20] .
De nitions FBG≧7.0mmol/l or 2hBG≧11.1mmoo/l based on OGTT by WHO 1999 criteria [22] or a previous diagnosis or self reported current hypoglycemic treatment. NAFLD was diagnosed when LFC by ultrasonography exceeded the cut-off value of 9.15%, excluding excessive alcoholic intake and virus hepatitis [19] .
Spearman analysis was performed to assess the relationships between ASM% and blood glucose as well as other clinical parameters. Multivariate logistic regression analyses were conducted to investigate the association between ASM% quartiles with DM after successively adjusting for age, cigarette smoking, diabetes family history, fat mass, interaction between fat mass and ASM% quartiles, the presence of obesity, blood pressure, serum triglyceride, HDL-C, and HOMA-IR. The interaction between ASM% and fat mass was included in the multiple regression models because there were signi cant correlations between ASM%, fat mass and blood glucose. To further investigate whether NAFLD effected the relationship between SMM and DM, subgroup analysis were performed according to the presence of NAFLD and LFC quartiles. P value < 0.05 was considered to be statistically signi cant.

Characteristics of subjects
Of the total 3969 subjects, the mean age was 63.3 years, and the mean BMI was 24.1 kg/m 2 . Details of the subject characteristics are shown in Table 1. All subjects with lower ASM% were older and had higher body weigh and blood pressure, with higher BMI, FM, FM%, SBP and DBP. The lipid disorders were aggravated in subjects with lower ASM%, presenting higher TC, TC and LDL-C, and lower HDL-C. The most noteworthy was that FBG, 2hBG, HOMA-IR and LFC increased gradually as well as the prevalence of DM and NAFLD, with ASM% decreasing in both male and female participants (all P < 0.001).   To further investigate whether low ASM% was associated with the risk of DM, we performed logistic stepwise regression analysis, with ASM% quartiles as independent variate and the presence of DM as dependent variate. As shown in Table 3  Effect of NAFLD on the relationship between skeletal muscle mass and DM As we know, NAFLD increased the prevalence and risk of type 2 diabetes, we also found ASM% was negative associated with liver fat content in our study. Thus, we further conducted logistic analysis to observe the effect of NAFLD on the relationship between skeletal muscle mass and the risk of DM. As shown in Table 4  As NAFLD was diagnosed by LFC in the present study, which was displayed as continuous variable. We further strati ed the population by LFC quartiles from low to high. The results showed that before adjustments, in the rst three LFC quartiles in male and the rst two LFC quartiles in female, the ASM% quartiles were negatively correlated with the risk of DM. After adjustments, the relationship remained signi cant in the rst and second quartile in male, however, the correlation no longer existed after adjustments in the third and forth quartile in male and in all quartiles in female. (Table 5)

Discussion
As far as we know, several studies explored the relationship between skeletal muscle mass with diabetes. The cross-section research by Srikanthan et al. revealed skeletal muscle mass accessed by BIA was negatively associated with HOMA-IR and glycated hemoglobin A1c (HbA1C) in American middle aged population [23] . The cohort study conducted by Korean researchers also showed lower ASM% could increased the risk of DM in young and middle-aged population [24,25] . Although these studies showed results similar to our ndings, they didn't showed the gender difference of the relationship between muscle loss and DM. In our study, we found that skeletal muscle mass measured by ASM% was negatively associated with blood glucose, and skeletal muscle loss may increase the risk of DM only in men. The dissociation of skeletal muscle mass loss and DM in women is noteworthy, especially after adjusted by fat mass and lipid parameters. The exact reason for this interesting phenomenon is still unknown. The subjects in the present study were older., and previous studies showed with age increasing, the percentage of body fat increased gradually, which was more pronounced in older [24,25] and women [26] . This could result increasing insulin resistance and lipid disorders more signi cant in women, which might neutralize the effect of skeletal muscle reduction on diabetes. Another possible explanation is the difference of body fat distribution between genders. Although the body fat percentage of women was higher than the match aged men, and with aging the accumulation of intramuscular and intermuscular fat was more signi cant in women than in men [27] , however, women had more type I muscle bers than men, contributing stronger oxidative function of skeletal muscle [28] , which could reduce the risk of diabetes. On the other hand, hormones especially estrogen could in uence the metabolism of triglyceride and free fatty acids [29] . Estrogen decreased as aging, especially in postmenopausal women, which may result in the reduction of triglyceride, and was associated with the reduced risk of diabetes [30] . In addition, in the process of aging, the decline of skeletal muscle mass was more remarkable in male than in female, which also contributed to the more signi cant effect of skeletal muscle loss on the risk of diabetes in male. The results suggest gender strati ed management of diabetes according to skeletal muscle needs to be considered in the future. Increasing skeletal muscle mass might have a more bene cial effect on improving glucose metabolism in male population.
NAFLD is an important risk factor of diabetes, and several previous studies have demonstrated that low skeletal muscle mass was also independently associated with NAFLD [14][15][16] . In the present study, the skeletal muscle mass was also negatively associated with LFC, which was similar to the previous results.
Whether liver fat content would in uence the relationship between sarcopenia and diabetes is still unknown. Our results showed the relationship between skeletal muscle mass and diabetes existed in the non-NAFLD male population and disappeared in the NAFLD population. It indicated that as an important risk factor for diabetes, excessive liver fat accumulation which could lead to insulin resistance, mitochondrial dysfunction, hyperlipidemia and so on [31] , could lessen the effect of skeletal muscle reduction on diabetes. The results indicated that reducing liver fat maybe more important when referred to improving diabetes in NAFLD population. While in the non-NAFLD population, skeletal muscle mass enhancement might be helpful in diabetes treatment. Further analysis in our study revealed an interesting result was that, the association of skeletal muscle and DM remained in the male population whose LFC less than 5.52%, which was similar to the histopathologic diagnosis of fatty liver, The ndings indicated that in men with liver fat less than 5.52%, increasing skeletal muscle mass may help prevent diabetes.
The mechanism underlying the relationship between low skeletal muscle mass and diabetes are still not fully understood. It has been known that insulin resistance and systemic in ammation played an important role on the development of both skeletal muscle reduction and diabetes [6,32] . As an important target organ of insulin action, skeletal muscle plays an important role in maintaining the basic metabolism and glucose metabolism stability [33] . Decreased skeletal muscle mass, which was often accompanied by intermuscular fat accumulation, could increase macrophage in ltration, mitochondrial dysfunction and in ammatory factors releasing, contributing to insulin resistance and reduced glucose uptake and utilization [34][35][36] . While increasing skeletal muscle mass could improve insulin sensitivity and glucose metabolism [23] . But our current study also found that in male population, age-related skeletal muscle mass loss was independently associated with the risk of diabetes after adjustment for obesity, HOMA-IR and all components of metabolic syndrome, which suggested that there may be other mechanisms to explain this association. Although it is unclear whether skeletal muscle loss is the cause or the consequence of diabetes, a direct crosstalk between skeletal muscle and glucose metabolism has been uncovered. Previous studies have shown that skeletal muscle can secrete a variety of cytokines, such as IL-6 and Irisin, which can regulate insulin sensitivity, promote glucose uptake by skeletal muscle cells, reduce liver gluconeogenesis, and improve glucose metabolism by acting on adipose tissue, liver and other tissues [37,38] . The impairment of muscle secretary function due to muscle loss may contribute to the development of diabetes.
To our knowledge, the current study might be the rst large-scale community population research to access the in uence of NAFLD on the association between diabetes with gender-and age-related skeletal muscle mass which was measured by DXA, a more accurated method of body compostion recommended by the guidelines [39] .Our ndings might develop a new perspective for preventing diabetes, especially in male non-NAFLD population. But there are also several limitations in our study. First, it was a crosssectional study which cannot reveal the causal relationship between skeletal muscle mass and diabetes. So it is necessary to further verify our ndings in a prospective cohort study. The second, the association of skeletal muscle loss and diabetes only existed in the rst and second LFC quartiles, and the cut-off point of LFC should be further conformed. Second, several serum myokines were not detected in our current study, which might help explore the mechanisms underlying the relationship between low skeletal muscle mass and diabetes.

Conclusions
Skeletal muscle loss was associated with increasing risk of diabetes in male non-NAFLD middle-aged and elderly community population. Our results suggested a new practical strategy to facilitate personalised intervention of diabetes by increasing skeletal muscle mass in male non-NAFLD patients.