A case-control study was performed. Cases were Caucasian obese children (BMI above the 95^th percentile for age and gender) seen at the Hepatology outpatient Clinic of the Department of Pediatrics, Sapienza University of Rome, Italy. The diagnosis of NAFLD was based on magnetic resonance imaging (MRI) with high hepatic fat fraction (HFF ≥ 5%). Other causes of chronic liver disease, including hepatic virus infections (hepatitis A-E and G, cytomegalovirus, and Epstein-Barr virus), autoimmune hepatitis, metabolic liver disease, α-1-antitrypsin deficiency, cystic fibrosis, Wilson’s disease, hemochromatosis, and celiac disease were ruled out with appropriate tests. Exclusion criteria were also smoking habits, and history of type 1 or 2 diabetes, renal disease, total parenteral nutrition, use of hepatotoxic medications, and chronic alcohol intake. Finally, children were excluded for conditions that could have adversely influenced BMD including glucocorticoid therapy, hypothyroidism, Cushing’s disease; history of long bone fractures; indwelling hardware; and abnormality of the skeleton or spine[30,31].

Controls were selected from Caucasian obese children with normal levels of aminotransferases, and without MRI evidence of fatty liver (HFF < 5%) as well as of other causes of chronic liver diseases (see above). Controls were also excluded if they had smoking habits, history of type 1 or 2 diabetes, renal disease, chronic alcohol intake, and any condition known to influence BMD[30,31]. Controls were then matched (1- to 1-basis) with the cases on age, gender, pubertal stage and as closely as possible on BMI-SD score (SDS).

The research protocol was approved by the Hospital Ethics Committee, and informed consent was obtained from subjects’ parents before assessment.

All participants underwent physical examination including measurements of weight, standing height, BMI and determination of the stage of puberty, and laboratory tests. The pubertal stage was categorized into two groups (prepubertal: boys with pubic hair and gonadal stage I, and girls with pubic hair stage and breast stage I; pubertal: boys with pubic hair and gonadal stage ≥ II and girls with pubic hair stage and breast stage ≥ II). The degree of obesity was quantified using Cole’s least mean-square method, which normalizes the skewed distribution of BMI and expresses BMI as SDS[32]. Blood samples were taken, after an overnight fast, for estimation of glucose, insulin, alanine aminotransferase (ALT), aspartate aminotransferase (AST), HSCRP, leptin, and adiponectin.

Analyses of glucose, insulin, ALT, AST, and HSCRP were conducted by COBAS 6000 (Roche Diagnostics). Insulin concentrations were measured on cobas e 601 module (Electrochemiluminescence Technology, Roche Diagnostics), while the remaining analytes on cobas e 501 clinical chemistry module (Photometric Technology), according to the instructions of the manufacturer. The degree of insulin resistance was determined by a homeostasis model assessment of insulin resistance (HOMA-IR)[33]. Scores were calculated as the product of the fasting serum insulin level (μU/mL) and the fasting serum glucose level (mmol/L), divided by 22.5. A RIA was used to measure human (total) leptin (DRG Diagnostica, Marburg, Germany; detection limit, 0.5 ng/mL; inter- and intra-assay CVs, 3.0%-6.2% and 3.4%-8.3%, respectively), and adiponectin (DRG Diagnostica, Marburg, Germany; detection limit, 1 ng/mL; inter- and intra-assay CVs, 6.9%-9.2% and 1.8%-6.2%, respectively).

The amount of hepatic fat content (% HFF) was measured by MRI using the two-point Dixon method as modified by Fishbein[34], as previously described and validated[35]. MRI results were interpreted by an experienced radiologist who was blinded to clinical and laboratory findings.

Anteroposterior lumbar spine (L₁-L₄), and whole body scans were obtained from all cases and controls using a Hologic QDR-4500W (Waltham, MA, United States) in the fan beam mode with a multidetector system. All subjects were measured on the same machine. The measurements were performed by using standard positioning techniques. Quality control was performed daily using the Hologic anthropomorphic spine, and weekly with the whole body phantom. In our department, the precision error for BMD measurements is less than 1% for the spine phantom, and less than 2.5% for the whole body phantom. The data were analyzed using the software version 11.2. Spine scans were analyzed with low-density software[36]. BMD Z-scores for whole body (WB) and for lumbar spine (LS) were calculated using race and gender specific LMS curves[37]. Whole body DXA results (BMD, fat mass and lean mass) shown in this study represent values excluding the skull[38].

The clinical indication for biopsy was either to assess the presence of NASH and degree of fibrosis or other likely independent or competing liver diseases. Percutaneous needle liver biopsy was performed as previously described[35]. The main histologic features of NAFLD were scored according to the scoring system developed by the NASH Clinical Research Network (CRN)[39]. Features of steatosis, lobular inflammation, and hepatocyte ballooning were combined to obtain the NAFLD activity score. As recommended by a recent NASH CRN article[40], a microscopic diagnosis, based on overall injury pattern (i.e., steatosis, hepatocyte ballooning, and inflammation), as well as the presence of additional lesions (e.g., zonality of lesions, portal inflammation, and fibrosis), has been assigned to each case[41]. Accordingly, biopsies were subdivided into not-NASH and definite NASH subcategories[41].

Statistical analyses were performed using the SPSS package. The data are expressed either as frequencies or as means with 95%CI. Insulin, leptin and adiponectin levels were distributed with a long tail to the right (positive skew), but their logarithms were approximately normally distributed. Mean differences in anthropometric, laboratory and body composition variables between subjects were assessed by using the t test. Linear regression analysis was used to identify variables associated with BMD. Then, a stepwise multiple linear regression analysis (including all variables significantly associated with BMD) was used to determine the independent variables associated with BMD. A P value of less than 0.05 was considered to be statistically significant.

Forty four obese children with MRI-diagnosed NAFLD were matched to 44 obese children without evidence of liver disease. By study design cases and controls were matched for age, gender, pubertal stage and BMI-SDS. The mean age of cases and controls was 12.5 (SD 1.8) years. Both cases and controls included 20 girls and 24 boys, and five prepubertal children. The mean BMI-SDS of cases and controls was 2.19 (SD 0.16) and 2.17 (SD 0.16), respectively. The clinical and laboratory characteristics for cases and controls are shown in Table 1. There were no differences between children with and without NAFLD with respect to lean and fat mass. Compared to the non-NAFLD group, children with NAFLD had significantly higher ALT, AST, insulin concentrations, HOMA-IR values, and HSCRP levels, but lower adiponectin concentrations. There were no significant differences between the two groups with respect to glucose as well as leptin.

Liver biopsy was obtained in 35 of the 44 children with MRI-diagnosed NAFLD, with parental refusal in 9 cases. The 35 children did not differ from those having only liver MRI with respect to age, gender, body composition, metabolic parameters, and bone measures.

Among patients with biopsy-proven NAFLD, 20 (57%) had definite NASH, while 15 (43%) no NASH. No statistically significant differences in body composition as well as in laboratory parameters such as glucose, insulin, leptin, adiponectin levels, and HOMA-IR values were found between children with NASH and those with simple steatosis. AST [mean, 41 U/L (95%CI: 34-48) vs 26 U/L (95%CI: 22-29); P < 0.001)], ALT [mean, 58 U/L (95%CI: 41-75) vs 30 U/L (95%CI: 20-45); P < 0.001)] as well as HFF [mean, 24.8% (95%CI: 19.5-30.2) vs 15.7% (95%CI: 5.6-28.8); P < 0.001)] were significantly higher in patients with NASH compared to children without NASH. HSCRP was also higher [mean, 4055 μg/L (95%CI: 2690-5419) vs 2870 μg/L (95%CI: 1794-3936); P = 0.07], although did not reach statistically significance.

Obese children with NAFLD had a significantly lower LS BMD Z-score than those without NAFLD [mean, 0.55 (95%CI: 0.23-0.86) vs 1.29 (95%CI: 0.95-1.63); P < 0.01] (Figure 1A). WB BMD Z-score was also decreased in obese children with NAFLD compared to obese children with no NAFLD, though borderline significance was observed [1.55 (95%CI: 1.23-1.87) vs 1.95 (95%CI: 1.67-2.10); P = 0.06] (Figure 1B). Among children with biopsy-proven NAFLD, those with NASH had a significantly lower LS BMD Z-score than children without NASH [mean, 0.27 (95%CI: -0.17-0.71) vs 0.75 (95%CI: 0.13-1.39); P < 0.05] (Figure 1C). Moreover, children with NASH had a significantly lower WB BMD Z-score than children without NASH [1.38 (95%CI: 0.89-1.17) vs 1.93 (95%CI: 1.32-2.36); P < 0.05] (Figure 1D).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Bone measures. A: Lumbar spine bone mineral density Z-score (LS BMD Z-score) for obese children with and without nonalcoholic fatty liver disease (NAFLD). Box-plots give the median value (bold), 25^th and 75^th percentiles (lower and upper limits of the box), and lower and upper adjacent values (whiskers); B: Whole body bone mineral density Z-score (WB BMD Z-score) for obese children with and without NAFLD. Box-plots give the median value (bold), 25^th and 75^th percentiles (lower and upper limits of the box), and lower and upper adjacent values (whiskers); C: LS BMD Z-score for obese children with biopsy-proven NAFLD subdivided into those with and without nonalcoholic steatohepatitis (NASH). Box-plots give the median value (bold), 25^th and 75^th percentiles (lower and upper limits of the box), and lower and upper adjacent values (whiskers); D: WB BMD Z-score for obese children with biopsy-proven NAFLD subdivided into those with and without NASH. Box-plots give the median value (bold), 25^th and 75^th percentiles (lower and upper limits of the box), and lower and upper adjacent values (whiskers).

In univariate analysis, LS BMD Z-score correlated negatively with NAFLD (standardized β coefficient, -0.202; P < 0.01) and HSCRP (standardized β coefficient, -0.212; P <0.05). In contrast, leptin was positively associated with lumbar BMD (standardized β coefficient, -0.204; P < 0.05). No correlation was found between LS BMD Z-score and insulin as well as HOMA-IR. Likewise, neither BMI-SDS nor lean mass nor fat mass were correlated with LS BMD Z-score. After including in the model all the significant variables as well as age, gender, pubertal status, NAFLD (standardized β coefficient, -0.230; P < 0.01) and HSCRP (standardized β coefficient, -0.195; P < 0.05) remained significantly and independently associated with LS BMD Z-score (Table 2).

WB BMD Z-score was negatively associated with NAFLD (standardized β coefficient, -0.207; P < 0.05), fat mass (standardized β coefficient, -0.222; P < 0.05), and HSCRP (standardized β coefficient, -0.216; P < 0.05). No correlation was found between WB BMD Z-score and insulin as well as HOMA-IR. Likewise, neither BMI-SDS nor lean mass were correlated with WB BMD Z-score. After including in the model all the significant variables as well as age, gender, pubertal status, NAFLD (standardized β coefficient, -0.218; P < 0.05) and fat mass (standardized β coefficient, -0.225; P < 0.05) remained significantly and independently associated with WB BMD Z-score (Table 2).

In parallel with epidemic obesity, nonalcoholic fatty liver disease (NAFLD) has emerged as the leading cause of chronic liver disease in both pediatric and adult patients worldwide. Liver disease can be cause of low bone mineral density (BMD). However, the mechanisms explaining this relationship are still not completely understood.

A better understanding of the factors that may influence bone mineral status in NAFLD may open a new frontier to fight two highly prevalent conditions like NAFLD and osteoporosis.

Recent years have witnessed an increased awareness of the clinical and epidemiological association between NAFLD and bone health, both in terms of reduced BMD and an increased risk of osteoporosis. Given the high prevalence of NAFLD and the adverse consequences of low BMD in childhood, understanding the mechanisms underlying the relationship between NAFLD and low BMD is important to prevent poor bone mineralization in this potentially vulnerable population. In this study, authors showed that obese children with NAFLD have decreased BMD compared to obese children without liver involvement independently of adiposity, and that children with more severe histology have worse mineral status than children with more mild abnormalities. They also found a significant independent association of high sensitivity C-reactive protein with BMD scores, supporting the role of an inflammatory state which may accelerate loss of bone mass in patients with NAFLD.

The presence of systemic inflammation may have important implications for the long-term skeletal health of children with NAFLD, and particularly those with nonalcoholic steatohepatitis (NASH).

NAFLD comprises a disease spectrum ranging from simple fatty liver to NASH, with varying degrees of inflammation and fibrosis, progressing to end-stage liver disease with cirrhosis and hepatocellular carcinoma. Bone density (or BMD) is a medical term normally referring to the amount of mineral matter per square centimeter of bones. Bone density (or BMD) is used in clinical medicine as an indirect indicator of osteoporosis and fracture risk.

In this paper, authors compared lumbar spine (LS) and whole body (WB) BMD measured by dual energy X-ray absorptiometry scans between 44 pediatric patients with magnetic resonance imaging diagnosed NAFLD and controls matched 1:1 for age, gender, and pubertal stage and body mass. They found that LS-BMD Z score was lower in NAFLD than in controls; Thirty three NAFLD patients were biopsied; LS and WB BMD Z score were lower in NASH than in non-NASH children. At multivariate analysis LS-BMD was independently associated with NASH and C-reactive protein levels. They conclude that NAFLD is associated with low BMD in obese children, and systemic low grade inflammation may play a role in such a relationship.