Non-alcoholic fatty liver disease is associated with low bone mineral density in obese children

Authors

  • P. E. Pardee,

    1. Division of Gastroenterology, Hepatology, and Nutrition, Department of Pediatrics, University of California, San Diego, San Diego, CA, USA.
    2. Department of Medicine, University of California, San Diego, San Diego, La Jolla, CA, USA.
    Search for more papers by this author
  • W. Dunn,

    1. Division of Gastroenterology and Hepatology, Department of Medicine, University of Kansas Medical Center, Kansas City, Kansas, USA.
    Search for more papers by this author
  • J. B. Schwimmer

    1. Division of Gastroenterology, Hepatology, and Nutrition, Department of Pediatrics, University of California, San Diego, San Diego, CA, USA.
    2. Department of Gastroenterology, Rady Children’s Hospital San Diego, San Diego, CA, USA.
    3. Liver Imaging Group, Department of Radiology, University of California, San Diego, San Diego, CA, USA.
    Search for more papers by this author

Dr J. B. Schwimmer, Division of Gastroenterology, Hepatology, and Nutrition, Department of Pediatrics, University of California, San Diego, 3020 Children’s Way, MC 5030 San Diego, CA 92123, USA.
E-mail: jschwimmer@ucsd.edu

Abstract

Aliment Pharmacol Ther 2012; 35: 248–254

Summary

Background  Non-alcoholic fatty liver disease (NAFLD) is the most common cause of chronic liver disease in children. Liver disease can be a cause of low bone mineral density. Whether or not NAFLD influences bone health is not known.

Aim  To evaluate bone mineral density in obese children with and without NAFLD.

Methods  Thirty-eight children with biopsy-proven NAFLD were matched for age, gender, race, ethnicity, height and weight to children without evidence of liver disease from the National Health and Nutrition Examination Survey. Bone mineral density was measured by dual energy X-ray absorptiometry. Age and gender-specific bone mineral density Z-scores were calculated and compared between children with and without NAFLD. After controlling for age, gender, race, ethnicity and total per cent body fat, the relationship between bone mineral density and the severity of histology was analysed in children with NAFLD.

Results  Obese children with NAFLD had significantly (< 0.0001) lower bone mineral density Z-scores (−1.98) than obese children without NAFLD (0.48). Forty-five per cent of children with NAFLD had low-bone mineral density for age, compared to none of the children without NAFLD (P < 0.0001). Among those children with NAFLD, children with NASH had a significantly (P < 0.05) lower bone mineral density Z-score (−2.37) than children with NAFLD who did not have NASH (−1.58).

Conclusions  The NAFLD was associated with poor bone health in obese children. More severe disease was associated with lower bone mineralisation. Further studies are needed to evaluate the underlying mechanisms and consequences of poor bone mineralisation in children with NAFLD.

Introduction

Attention to skeletal health during childhood is important, as low-bone mineral density (BMD) during this period has immediate and long-term consequences. In children, low BMD is associated with an increased risk of fracture.1–3 Fractures in childhood may disrupt normal activities such as school and sports, and can be detrimental to growth. In addition, low BMD in childhood has been shown to persist into early adulthood.1 This is important because peak bone accrual occurs before 20 years of age,3 and is the primary determinant of adult osteoporosis risk.4 Although several causes of low BMD in children have been identified, much remains unclear. In particular, the relationship between bone mineralisation and obesity remains controversial, with numerous studies in conflict as to whether obesity increases or decreases BMD in children.5–11 Since low BMD in childhood has important consequences and obesity is very prevalent in children, it is important to identify factors underlying the discrepancy in prior studies of BMD and obesity. One potential factor related to BMD in obese children is non-alcoholic fatty liver disease (NAFLD).

The NAFLD is the most common cause of chronic liver disease in children, estimated to affect up to one-third of obese children.12 First reported in children in 1983,13 NAFLD was initially considered only as a spectrum of liver disease. However, furthermore research has expanded the phenotype of NAFLD to include other health conditions, independent of obesity. For example, children with NAFLD have a worse cardiometabolic profile than equally obese children without NAFLD.14 Similarly bone health may also be an important issue that differs among obese children based on the presence or absence of NAFLD. An initial study from Turkey reported that obese children with abnormal liver ultrasonography suggestive of hepatic steatosis had lower spine BMD Z-scores than those in obese children with normal liver ultrasonography.15 However, this study did not include children with definitively diagnosed NAFLD by biopsy, and only examined spine BMD Z-score. Thus, we sought to determine whether or not children with biopsy-proven NAFLD have a lower total body BMD than obese children without NAFLD. NAFLD may be associated with poor bone mineralisation, as the pathogenesis is thought to involve systemic inflammation,16 one of the known aetiologies of low BMD in children.17–20 Non-alcoholic steatohepatitis (NASH), a severe, progressive form of NAFLD, is associated with a greater degree of inflammation.21 Thus, paediatric NAFLD, and particularly NASH, may be associated with low BMD. NAFLD is also strongly associated with obesity.12, 22 However, we hypothesise children with NAFLD have lower BMD than children without NAFLD, independent of the degree of obesity.

Materials and methods

Study design

We performed a case-control study of obese children to compare the BMD of children with and without NAFLD. The parent(s) or legal guardian of all subjects provided written informed consent. Written assent was obtained from all children. The protocol was approved by the institutional review boards of the University of California, San Diego and Rady Children’s Hospital San Diego.

Cases with NAFLD

Cases were children aged 10–17 years with biopsy-proven NAFLD seen at the Fatty Liver Clinic at Rady Children’s Hospital in San Diego. The diagnosis of NAFLD was based on liver biopsy with ≥5% of hepatocytes containing macrovesicular fat and exclusion of other causes of chronic liver disease, including hepatitis B (hepatitis B surface antigen), hepatitis C (hepatitis C antibody), alpha-1 antitrypsin deficiency (serum alpha-1 antitrypsin level and histology), autoimmune hepatitis (antinuclear antibody, antismooth muscle antibody and histology), Wilson’s disease (serum ceruloplasmin), drug toxicity, total parenteral nutrition and chronic alcohol intake (clinical history).23 The determination of the presence or absence of steatohepatitis was based upon the features of steatosis, lobular and portal inflammation and ballooning degeneration of hepatocytes. Children were excluded for factors that could have adversely influenced BMD including a history of a fracture within the past year, history of orthopaedic surgery, or history of chronic glucocorticoid use.

Controls without NAFLD

Controls were selected from the National Health and Nutrition Examination Survey (NHANES) 1999–2004.24 Subjects with incomplete DXA data were excluded. To determine the absence of liver disease, controls had to have normal serum alanine aminotransferase (ALT) and be negative for hepatitis B (hepatitis B surface antigen), hepatitis C (hepatitis C antibody) and iron overload (transferrin saturation ≥50%). To minimise the false classification of controls, a stringent definition of normal ALT, based upon recent biology-based thresholds for abnormal serum ALT, was used; thus boys with serum ALT >25 U/dL and girls with serum ALT >22 U/dL were excluded.25 Controls were then matched with the cases for age, gender, race, ethnicity, height and weight.

Clinical data collection

Clinical data were obtained for each participant at a single fasting intake visit conducted at the Clinical and Translational Research Institute (CTRI) at the University of California, San Diego Medical Center (cases), or by trained NHANES personnel at sites throughout the United States (controls). Each participant’s age and gender were recorded. The race and ethnicity of the child was self identified by the guardian present at the intake visit. Height was measured to the nearest tenth of a centimetre on a clinical stadiometer. Weight was measured on a clinical scale to the nearest tenth of a kilogram. Body mass index (BMI) was calculated as weight in kilograms divided by height in metres squared. Phlebotomy was performed after a 12-h overnight fast, and assays for serum ALT and aspartate aminotransferase (AST) were performed using an enzymatic rate method.26

Bone and body composition measurements

All cases and controls underwent whole-body DXA scans with standard positioning.27 A fan-beam densitometer (Hologic, Inc., Bedford, MA, USA) was used for cases (model QDR 4500W, software version 12.4) and controls (model QDR 4500A, software version 12.1). Both software versions include the same paediatric-specific data analysis package. Studies of similar differences in software versions in other models have not shown significant changes in body composition or BMD data.28 Total per cent body fat and BMD were determined from the whole-body scan data. BMD Z-scores were calculated using race and gender-specific LMS curves based on over 1500 children.29 Low BMD for age was defined as a BMD Z-score of ≤−2.0, as recommended by the International Society for Clinical Densitometry guidelines.30

Data analysis

Data are expressed as mean (s.d.) or as number and percentage. Continuous variables were compared using Student’s t-tests. Categorical variables were compared using Chi-squared tests. The BMD Z-scores of the case and control groups were compared using two-way ancova, adjusting for per cent body fat. The prevalence of low BMD for age (BMD Z-score ≤−2.0) in the case and control group was compared using conditional logistic regression and McNemar’s test exact method. After controlling for age, gender, race, ethnicity and total per cent body fat, BMD Z-scores were compared among children with NAFLD, comparing those with and without NASH using linear regression. sas 9.1 (SAS Institute Inc., Cary, NC, USA) was used for statistical analysis.

Results

Study sample

Thirty-eight obese children with biopsy-proven NAFLD were matched to 38 obese children without evidence of liver disease. Consistent with the epidemiology of NAFLD, there were more boys than girls among the groups. The demographic and clinical characteristics for cases and controls in the primary analysis are shown in Table 1. By study design cases and controls were matched for age, gender, race, ethnicity, height and weight. The mean age of the cases and controls was 13 ± 2 years. The matching for height and weight yielded the same mean BMI for both groups. Obese children with NAFLD did have significantly (P < 0.001) higher mean ALT, AST, and total per cent body fat than obese children without NAFLD. Of the 38 children with NAFLD, 20 had NASH.

Table 1.   Characteristics of study population by liver status
CharacteristicNormal liver (N = 38)NAFLD (N = 38)
  1. P < 0.001.

Age, mean (s.d.), years13 (2)13 (2)
Gender, N (%)
 Boys33 (87)33 (87)
 Girls5 (13)5 (13)
Race/Ethnicity, N (%)
 Hispanic34 (90)34 (90)
 White, non-Hispanic4 (11)4 (11)
Weight, mean (s.d.), kg81 (21)81 (21)
Height, mean (s.d.), cm159 (13)159 (14)
Body Mass Index (kg/m2)
 Mean (s.d.)31 (5)31 (4)
ALT, mean (s.d.), U/L20 (3)133 (136)*
AST, mean (s.d.), U/L22 (4)73 (55)*
Body fat, mean (s.d.), (%)39 (6)46 (7)*

Bone mineral density

Obese children with NAFLD had a significantly (P < 0.0001) lower BMD Z-score (−1.98) than obese children without liver disease (0.48), as shown in Figure 1. This relationship remained significant (P < 0.0001) after adjusting for total per cent body fat (mean BMD Z-score cases, −1.47; controls, −0.03). Among children with NAFLD, 45% (17/38) had BMD Z-scores ≤−2.0, compared to none of the controls (P < 0.0001). Of the 17 children with NAFLD and BMD Z-scores <−2.0, 10 (59%) had NASH. As shown in Figure 2, among children with NAFLD, NASH was associated with the lowest BMD. Children with NASH had a significantly (P < 0.05) lower BMD Z-score (−2.37) than children without NASH (−1.58). Notably, those children who had NAFLD but did not have NASH still had a significantly lower BMD Z-score than matched controls of obese children without liver disease.

Figure 1.

 Scatterplot of bone mineral density Z-scores for obese children with and without NAFLD. Children with NAFLD had significantly (P < 0.0001) lower BMD Z-scores than children without NAFLD.

Figure 2.

 Scatterplot of bone mineral density Z-scores for obese children NAFLD subdivided into those with and without NASH. Children with NASH had significantly (P < 0.05) lower BMD Z-scores than children with NAFLD who did not have NASH.

Discussion

We performed a case-control study of obese children with and without NAFLD to determine whether or not an association exists between NAFLD and BMD in children. Low BMD was a frequent finding among obese children with NAFLD, but not among obese children without NAFLD. This finding remained after controlling for total per cent body fat. BMD also differed by the severity of histology. Children with NASH had lower BMD for age compared to children with NAFLD who did not have NASH.

In adults, the association between some forms of chronic liver disease and osteoporosis is well established, with prevalence estimates of 1–21%, reaching as high as 50% in pretransplant patients.31, 32 However, in contrast to the children with NAFLD in this study, in adults, markedly low BMD is rare in the absence of cirrhosis or advanced cholestatic disease.31 In addition, in the adult population much of the relationship between chronic liver disease and osteoporosis is ascribed to classic risk factors, such as advanced age, postmenopausal status, excessive alcohol consumption, hypogonadism and the use of glucocorticoids33– risk factors that are not found in children with NAFLD. Furthermore, unlike many forms of chronic liver disease associated with low BMI, a risk factor for poor bone mineralisation, NAFLD is strongly associated with obesity.22 Thus, although NAFLD in children may be a form of chronic liver disease associated with low BMD, considerations in evaluating this relationship are strikingly different from those in adults. Obesity is one important factor that may complicate the relationship between NAFLD and low BMD in children.

Obesity and bone mineralisation in children has been studied extensively and remains a topic of great interest, as data are conflicting regarding whether obesity in this age group is detrimental or protective to bone. One reason for the controversy may be that obesity is not a unique phenotype. Rather it has become increasingly apparent that obesity is a heterogeneous condition, with significant differences existing even among children with the same degree of overweight. These differences may account for some of the discrepancies seen in the current BMD and obesity data. The distribution of body fat is one such difference among obese children that may affect bone mineral status. Visceral adipose tissue has been associated with low BMD in female adolescents.34 Inversely, subcutaneous adipose tissue has been positively associated with BMD.35 Moreover, differences in visceral and subcutaneous adipose tissue may be related to differences in hepatic steatosis. Recent estimates suggest that one-third of obese children are affected12 and many of these children are undiagnosed.36, 37 Thus, NAFLD represents a common and under-recognised difference among obese children that may contribute to the inconsistencies in the assessment of BMD in this group.

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 bone loss in this potentially vulnerable population. Why children with NAFLD, and particularly NASH, have low BMD compared to equally obese children without NAFLD is unknown, but, as hypothesised in other forms of chronic liver disease,38, 39 may involve the inflammatory response. Systemic inflammation is well known to contribute to low BMD in several disease states.17, 19, 20 The inflammatory cytokines tumour necrosis factor alpha (TNF-α), interleukin-6 and interleukin-1 all increase osteoclast activity by up-regulating receptor activator of nuclear factor kβ ligand (RANKL).40 In addition, TNF-α inhibits osteoblast differentiation and promotes osteoblast apoptosis.41 These cytokines have been implicated in the pathogenesis of NAFLD as well.16 NASH is thought to develop in part from this persistent inflammatory state, via the activation of hepatic stellate and dendritic cells.42, 43 Thus, the presence of inflammation in NAFLD may contribute to the association between NAFLD and low BMD found here. However, the complete pathophysiology is likely to be complex and multifactorial. For future studies to elicit a mechanistic understanding of this association, they will need to have rigorous, accurate phenotyping of study subjects to include assessment of both the presence or absence of NAFLD and to determine disease severity based upon features such as steatohepatitis.

The strengths of this study were the inclusion of children with rigorously defined biopsy-proven NAFLD and the use of a control group that was matched for age, gender, race, ethnicity, height and weight. The large number of participants available in NHANES allowed for precise matching of cases and controls. This study did have limitations. Due to the use of NHANES for control participants, controls were determined to not have NAFLD based upon serum ALT activity, rather than biopsy or imaging. However, we used recently defined, biologically based values for normal ALT, which are more stringent than the ALT values that are in widespread clinical use.25 Thus, the likelihood of falsely classifying a child in the control group as not having liver disease was relatively small. Furthermore, the misclassification of controls would favour the null hypothesis, rather than the large difference in BMD found in this study. In addition, the BMD of cases and controls was measured on different densitometers. However, all measurements were made using the same fan-beam technology from the same manufacturer and model series. Though the use of different machines may have produced small differences in BMD between cases and controls, it is very unlikely that the large difference seen in this study could be explained solely on this basis. Finally, the cross-sectional nature of the study allowed only for association rather than causation. Moreover, these data cannot be used to determine the timing of bone mineralisation in obese children with or without NAFLD.

Conclusions

Poor bone mineralisation was common among children with biopsy-proven NAFLD, but not among obese children without liver disease. Importantly, children with more severe histology had worse bone mineral status than children with more mild abnormalities. These differences persisted after controlling for total per cent body fat. This relationship may be an important aetiology underlying the discrepancy regarding whether obese children have higher or lower BMD than normal weight children, as many obese children have undiagnosed NAFLD. In addition, this finding has important implications for the long-term skeletal health of the child with NAFLD, and particularly the child with NASH. Whether or not fracture rates are higher in this population is not known, although we have observed several clinically significant fractures among children with NAFLD in our clinic. Longitudinal studies are needed to examine whether or not rates of fracture are indeed higher in this population. Whether bone density screening should be recommended for children with NAFLD or NASH, and what interventions should be taken to improve low BMD in this population, await furthermore research.

Acknowledgements

Declaration of personal interests: None. Declaration of funding interests: This study was supported in part by NHLBI T32RR023254 and by UL1RR031980 from the NCRR for the Clinical and Translational Research Institute at UCSD. The funders did not participate in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.

Ancillary