The association of liver fat content and serum alanine aminotransferase with bone mineral density in middle-aged and elderly Chinese men and postmenopausal women

Study subjects

The present study consisted of 2224 participants of Shanghai Changfeng Study, a community-based prospective cohort study of chronic diseases in a middle-aged and elderly Chinese population, who were consecutively recruited from May 2010 to June 2011 [17]. Among the initial 2224 participants, 441 were excluded from study enrollment because they were individuals with viral hepatitis B or C (86 persons), renal failure (75 persons), history of thyroid dysfunction (17 persons), excessive alcohol consumption ≥10 g/d for women and ≥20 g/d for men (143 persons) [18], without BMD data (120 persons), and 124 premenopausal women were also excluded. Finally, a total of 1659 community participants (755 men and 904 postmenopausal women; age range 46–93 years; mean 63.2 ± 10.3 years) were included in the study. This study was approved by the Research Ethics Committee of the Shanghai Health Bureau, and all participants provided written informed consent.

DXA measurements of BMD and body composition analysis

BMD (g/cm²) at the lumbar spine (L1–L4), total hips and the whole body region and the body fat masses at the whole body, trunk, and limbs sites were measured using dual-energy x-ray absorptiometry (Lunar iDXA, GE Healthcare). Body fat distribution was represented by the ratio of trunk to appendicular fat mass (trunk-appendicular fat ratio) [19]. A single, trained technician at a single clinical center carried out all measurements.

Quantitative ultrasonography

Hepatic ultrasound examination was performed in all patients by an ultrasonographist (who was unaware of the clinical details of the participants). The ultrasound images were taken by a GE Logiq P5 scanner (GE Healthcare, Milwaukee, WI, USA), analyzed using NIH image software (ImageJ 1.41o, National Institutes of Health, Bethesda, MD, USA) and standardized using a tissue-mimicking phantom (Model 057; Computerized Imaging Reference Systems, Norfolk, VA, USA) to correct for the instrument differences. As detailed in our previous work [16], all the instrument settings, including “gain”, “depth”, and “time-gain compensation”, were calibrated using the tissue-mimicking phantom before measurement. The ultrasound images with both liver and right kidney clearly visualized in the sagittal liver/kidney view and the right liver lobe in right intercostal view at anterior axilla line were captured under the ultrasound machine and transferred to a personal computer installed with NIH image software. In the sagittal liver/right kidney view, a region of interest (ROI) of 1.5 × 1.5 cm (1296 pixels) in the liver parenchyma and another ROI of 0.5 × 0.5 cm (144 pixels) in the right renal cortex at the same depth was selected. In right intercostal view at anterior axilla line, two ROIs of 1.5 × 1.5 cm (1296 pixels) were also selected in liver homogeneous regions near the liver anterior margin (depth 4–6 cm) and the liver posterior margin, respectively. The echo intensity within the ROIs was measured, as well as the linear distance between the liver anterior and posterior ROIs. Then the LFC was obtained using the automatic calculator (Additional file 1) based on the following equation: LFC (%) = 62.592 × ultrasound hepatic/renal ratio + 168.076 × hepatic attenuation rate −27.863 [16]. The LFC by ultrasound quantitative method showed excellent correlation and agreement with that by proton magnetic resonance spectroscopy previously (r = 0.89, P < 0.001), and this method has been used to investigate the associations of LFC with carotid atherosclerosis and diabetes previously [20, 21]. The subjects were defined as NAFLD if their LFCs by quantitative ultrasonography ≥9.15 % [16, 21].

Anthropometric and biochemical measurements

Each participant underwent a clinical examination that consisted of an interview by a trained investigator, anthropometric measurements and serum biochemical examinations. Information regarding demographics, lifestyle, medical history and current medication was obtained through interview and questionnaires. The questionnaire regarding alcohol intake included items about the type of alcoholic beverage consumed, the frequency of alcohol consumption on a weekly basis and the usual amount of daily alcohol consumption. Participants were classified as non- or light-drinkers or as excessive drinkers when their average alcohol consumption was <140 g/week for men (<70 g/week for women) or >140 g/week for men (>70 g/week for women) [18], respectively. Smoking status was defined as non-smoker or smoker. Standing height and bodyweight were measured without shoes or outer clothing. The body mass index (BMI) was calculated as weight (kg) divided by height squared (m²). Blood samples were obtained after a fasting period of at least 12 h. Total cholesterol (TC), high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), triglycerides (TG), creatine, uric acid and liver enzymes were measured using a model 7600 automated bio-analyzer (Hitachi, Tokyo, Japan). The fasting blood glucose (FBG) and 2 h post load glucose levels following a 75 g oral glucose challenge for non-diabetics were measured using the glucose oxidase method. Serum osteocalcin, 25-hydroxyvitamin D [25(OH)D] and β-isomerized form carboxy-terminal telopeptide of type I collagen (β-CTx) were measured using electrochemiluminescence immunoassay. The glomerular filtration rate (eGFR) was calculated using the Modification of Diet in Renal Disease (MDRD) Study formula [22].

Statistical analysis

All statistical analyses were performed using SPSS software version 15.0 (SPSS, Chicago, IL, USA). Kolmogorov–Smirnov Test was used for test of normality. Data were presented as mean ± SD, except for skewed variables, which were presented as median with the interquartile range (25–75 %) given in parentheses. Anova or the Kruskal–Wallis test was used for intergroup comparisons of continuous data, whereas the Chi squared test was used for comparisons of categorical variables. We stratified all analyses by sex [23], and the skewed variables were log transformed to approximate normal distribution before analyses. Correlations of BMD with LFC, serum ALT as well as other metabolic parameters were investigated using the linear correlation analysis after adjustment for age and body weight [24–26]. Multiple linear regression models were used to examine the independent associations of LFC with spine, hip and whole body BMD, adjusting for age, body weight, total body fat percentage (BF %), trunk-appendicular fat ratio and the variables that displayed significant associations with BMD in the univariate correlation analysis. We also calculated the sample size for the multivariate linear regression analysis, and a sample size of 1184 would allow for the detection of small effects (r² = 0.02) in multiple regression analysis with all covariates in the analysis with 2-tailed alpha of 0.05 and power of 0.80 [27].

To determine if ALT levels and LFC had synergistic effects on age-related bone mineral loss, the average BMDs were compared across three groups: (1) subjects without NAFLD; (2) subjects with NAFLD and normal ALT (≤40 IU/L); (3) subjects with NAFLD and increased ALT level (>40 IU/L). All P values reported were two-tailed and P < 0.05 was considered significant.

Clinical characteristics of the study population

Association of LFC and serum ALT level with BMD

After age and body weight were controlled for, LFC was negatively associated with BMD in both men and women (Fig. 1). These associations were significant at lumbar spine (r = −0.116, P < 0.001), hip (r = −0.095, P = 0.014) and whole body site (r = −0.134, P < 0.001) in men. In postmenopausal women, significant negative associations were also observed at the whole-body site (r = −0.107, P = 0.002) and the lumbar spine (r = −0.093, P = 0.008).

Serum ALT is commonly measured clinically as a biomarker for liver injury. In men, there were also inverse correlations of serum ALT with BMD at the whole body (r = −0.164, P < 0.001), lumbar spine (r = −0.102, P = 0.005), and hip (r = −0.075, P = 0.041). In postmenopausal women, ALT showed no association with spine and whole body BMD, and only marginal significant association with hip BMD (r = 0.067, P = 0.046) (Fig. 2).

Association of body fat distribution and metabolic parameters with BMD

Significant associations between BMD and NAFLD-related metabolic parameters were also found in both men and postmenopausal women. BMDs at specific bone site were significantly associated with systolic blood pressure (SBP), serum TG and TC in men, and SBP, FBG and uric acid in postmenopausal women after adjustment for age and body weight. Cigarette smoking is also related to decreased BMD at all bone sites in men (Table 2).

Independent contribution of LFC and ALT to variability in BMD and bone metabolic biomarkers

Interactions between ALT and LFC as risk factors for bone mineral loss in men

Compared to individuals with normal LFC, NAFLD patients had significantly lower BMD at the lumbar spine and whole body sites (Fig. 3a, b). When both NAFLD and elevation of ALT level were present, there was a synergistic worsening of the BMD at all bone sites, as shown in Fig. 3. Serum osteocalcin was also decreased with elevated serum ALT and presence of NAFLD.