|Home | About | Journals | Submit | Contact Us | Français|
The purpose of this study was to evaluate the prevalence of and factors related to liver disease among children in Hawai‘i with metabolic syndrome. The medical charts of children diagnosed with metabolic syndrome by an outpatient endocrinologist between January 2000 and December 2010 were reviewed. Liver disease prevalence was estimated based on serum alanine aminotransferase (ALT) levels, which were then assessed for associations with demographic (age, gender, ethnicity), anthropometric (body mass index), biochemical (fasting blood glucose, hemoglobin A1c, triglycerides, and total, LDL- and HDL-cholesterol), and clinical (blood pressure) characteristics of subjects. Serum ALT was available for 167 of the 195 subjects. The proportion of subjects with liver disease (105/167 [63%]) was greater than many traditional features of metabolic syndrome including hypertriglyceridemia (73/177 [41%]), hypertension (37/194 [19%]) and hyperglycemia (37/170 [22%]). Serum ALT values were positively associated with age (P=.030), and liver disease was more common among boys than girls (62/91 [68%] vs 43/76 [57%]), although this difference was not statistically significant (P=.123). There was a significant difference in liver disease across ethnicities (P=.029), and appeared to be more common in children with Pacific Islander surnames (14/16 [88%]), and less common in children with Hispanic surnames (7/20 [35%]). Diastolic blood pressure was the only obesity-related disease parameter associated with serum ALT after adjusting for age and gender (P=.018). In conclusion, liver disease was common among children diagnosed with metabolic syndrome in Hawai‘i. Age, gender, and ethnicity may be important determinants of liver disease risk, and should be investigated further.
Childhood obesity is a major health concern in Hawai‘i. According to the Hawai‘i Youth Risk Behavior Survey, over one-quarter of adolescents are overweight or obese.1 The term metabolic syndrome has been adopted to describe the clinical and biochemical derangements related to excess body fat. The definition of metabolic syndrome in children has yet to be clearly determined, but usually includes measures of central adiposity, insulin resistance, dyslipidemia, and hypertension.2 Nonalcoholic fatty liver disease (NAFLD) is another consequence of obesity that is closely linked to metabolic syndrome, and has been proposed as a defining feature of the condition.3–5
Most children with NAFLD suffer from non-specific symptoms such as fatigue, which contribute to lower physical and psychosocial health, and reduced quality of life.6 The prevalence and clinical spectrum of pediatric NAFLD varies considerably between populations, but affects roughly 40% of obese adolescents in the United States, and presents with hepatic inflammation and/or fibrosis in the majority cases.7–10 Despite rapidly becoming the leading cause of liver disease in adolescents, screening for elevated hepatic transaminases in serum (transaminasemia) indicative of NAFLD, namely alanine aminotransferase (ALT), is still rarely carried out during general pediatric visits of at-risk subjects.11
Presently, no studies have been published looking at the epidemiology of NAFLD among children in Hawai‘i. In view of the varied risk of pediatric NAFLD and the unique ethnic makeup of Hawai‘i, it is important to evaluate its present burden.12–14 The purpose of this study is to evaluate the prevalence of and factors related to elevated serum ALT values indicative of liver disease in children diagnosed with metabolic syndrome in Hawai‘i.
The study sample included consecutive patients referred to the outpatient pediatric endocrinologist at Kapi‘olani Medical Center for Women and Children (KMCWC) between January 2000 and December 2010 who were described as having “metabolic syndrome” during the initial consultation at age 1–19 years old. Demographic, anthropometric, biochemical, and clinical data were collected from the patient's medical charts. The University of Hawai‘i and Hawai‘i Pacific Health Institution Review Boards approved this study.
Height and weight were converted into body mass index (BMI) standard deviation score (SDS) for age and gender according to the 2000 CDC growth charts using the lambda, mu, and sigma (LMS) technique.15 Systolic and diastolic blood pressures were converted into SDSs for age, gender and height as described previously.16 The prevalence of metabolic syndrome features was evaluated based on the criteria from Graham, et al (2009), except for central obesity, which was approximated using the body mass index (BMI), a measure of excess body weight, because waist circumference was not available (Table 3).12 Subject ethnicity was determined using the surname list method developed for the Multiethnic Cohort Study (MEC), and grouped according to the National Institutes of Health (NIH) standards for race and ethnicity as Asian (Chinese, Filipino, Japanese), Hispanic, Pacific Islander (Hawaiian, Samoan), or White (surname not found).17–18 Although both White and African American ethnicities were not available in the MEC surname list, it is expected that the majority of these subjects would have been White based on the ethnic makeup of Hawai‘i.19 Surnames that corresponded to more than one ethnicity were assigned to the dominant ethnicity in Hawai‘i.19
The Upper Limit of Normal (ULN) for serum alanine aminotransferase (ALT) varies considerably between facilities, largely due to differences in reference populations used.20 As per a telephone communication with the laboratory manager Teresa Walsh (December 18, 2012), Clinical Laboratories of Hawai‘i, which provides laboratory services to KMCWC, use an ULN for serum ALT of 51 U/L for boys and 31 U/L for girls. In this study, subjects were classified as liver disease cases or controls based on the 95th percentile for healthy boys (<25.8 U/L) and girls (<22.1 U/L) in the United States.21 Compared to the ULN for ALT used in many children's hospitals, this cutoff was found to provide higher sensitivity for chronic liver disease with only a slight loss of specificity.21
Subject characteristics were summarized according to liver disease status as frequencies (percentage) for categorical variables, and means ± standard deviation or medians (interquartile range) for parametric and nonparametric continuous variables, respectively (Table 1). Continuous study variables were classified as parametric (systolic and diastolic blood pressure SDS) or nonparametric (age, BMI SDS, fasting blood glucose, hemoglobin A1c, total-, LDL- and HDL-cholesterol, serum triglycerides, alanine aminotransferase) based on Shapiro-Wilk test and visual assessment of frequency distribution graphs. Liver disease cases were compared to controls using chi-square analysis for categorical variables, independent two-sample t-test for parametric continuous variables, Wilcoxon rank sum test for nonparametric continuous variables, and logistic regression analysis for all variables, adjusting for age and gender (Table 1). The relationship between serum ALT and continuous variables was also analyzed using Spearman's rank correlation coefficient and linear regression analysis, adjusting for age and gender (Table 2). Due to the differences in the ULN for serum ALT, direct comparison of serum ALT between boys and girls was not conducted.21 Prevalence of liver disease in relation to other features of metabolic syndrome was determined using the ULN of serum ALT from both Clinical Laboratories of Hawai‘i (boys 51 U/L, girls 31 U/L), and the 95th percentiles for healthy children in the United States (boys <25.8 U/L, girls <22.1 U/L) (Table 3). Finally, subjects with and without serum ALT measurements available were compared across study parameters for sensitivity analysis. Statistical tests were carried out using SAS version 9.2, and graphs were created using Microsoft Excel for Mac 2011 v. 14.2.5.
A total of 195 children (12.1±3.6 years old, 103/195 [53%] boys, Table 1) were referred to the pediatric endocrinologist, and described as having metabolic syndrome during the initial consultation during this period. Data was available at the first appointment for over 80% of subjects for all study variables, including ALT (167/195 [86%], Table 1). The group that was missing ALT measurements was not significantly different (α<.05) in any of the variables measured (data not shown).
Despite slightly higher thresholds, boys were more likely to have elevated ALT values indicative of liver disease (62/91 [68%]) than were girls (43/76 [57%]), although this difference was not statistically significant (P=.123, Table 1). Similarly, the presence of liver disease was not associated with age (P=.369, Table 1), although there was a slight positive correlation between serum ALT and age (P=.030, Table 2). Liver disease was associated with ethnicity (P=.012, Table 1). Compared to White children (33/52 [63%]), Pacific Islander children tended to be more likely (14/16 [88%], P=.069), and Hispanic children were less likely (7/20 [35%], P=.029) to have elevated ALTs suggestive of liver disease (Table 1). Multivariable analyses of study parameters adjusting for age and gender produced similar results (Tables 1 and and22).
The majority of patients was obese (BMI 95th percentile for age and gender; 191/194 [98.5%]), and had multiple comorbidities consistent with the diagnosis of metabolic syndrome (Table 3). Apart from obesity, the most common features of metabolic syndrome were low HDL-cholesterol (40 mg/dL boys, 50 mg/dL girls; 122/177 [69%]) and hypertriglyceridemia (150 mg/dL; 73/177 [41%]) (Table 3). Hypertension (systolic or diastolic blood pressure 95th percentile for age, gender and height) and fasting hyperglycemia (110 mg/dL) were less common, presenting in 37/194 (19%) and 37/170 (22%) of patients, respectively (Table 3). The majority of subjects had elevated serum ALT values suggestive of liver disease (boys 25.8 U/L, girls 22.1 U/L) was 105/167 (63%) (Table 3). Even at the more conservative ULN for serum ALT used locally by Clinical Laboratories of Hawai‘i (boys 51 U/L, girls 31 U/L), an estimated 49/167 (29%) of subjects had liver disease, making it a relatively common complication in this population (Table 3).
As expected, subjects with raised serum ALT values had higher BMIs (2.54 [2.27–2.75] SDS vs 2.35 [2.13–2.61] SDS, P=.036), and were worse for all disease factors measured, although this was only statistically significant for fasting blood glucose (96 [87–111] mg/dL vs 89 [83–97] mg/dL, P=.007), and diastolic blood pressure (0.47±0.91 SDS vs 0.13±0.93 SDS, P=.025) (Table 1). After adjusting for age and gender, fasting blood glucose was no longer associated with the ALT group (P=.216) (Table 1). When examined as a continuous variable, serum ALT was correlated with fasting blood glucose (P=.005), total cholesterol (P=.045), serum triglycerides (P=.044), and systolic and diastolic blood pressure (P=.047 and P=.001), but not BMI (P=.265) (Table 2). Only the association between serum ALT and diastolic blood pressure remained statistically significant when analyzed using linear regression controlling for age and gender (P=.018) (Table 2).
The findings of this retrospective chart review confirm that elevated ALT is common in pediatric patients diagnosed with metabolic syndrome in Hawai‘i. The thresholds for transaminasemia in this study were based on the 95th percentile of ALT in healthy adolescents from the National Health and Nutrition Examination Survey 1999–2006, which is lower than the ULN of ALT that are generally used in US children's hospitals.21 The recommendation to adopt these lower cutoffs comes from the Screening ALT for Elevation in Today's Youth (SAFETY) study, which reported vast improvements in sensitivity for pediatric NAFLD in boys (32% to 80%) and girls (36% to 92%), while still maintaining moderate specificity of 79% and 85%, respectively.21 The ULN used by Clinical Laboratories of Hawai‘i is 51 U/L for boys and 31 U/L for girls, which still yields a prevalence of 29% for transaminasemia, making it a more common co-morbidity than either hypertension or fasting hyperglycemia. The relatively normal fasting glucose and hemoglobin A1c values are surprising given the high BMIs of children in this sample, although insulin resistance may have been disguised by insulin hypersecretion.22 Unfortunately, other glycemic measurements such as serum insulin concentrations and oral glucose tolerance tests were not available to test this hypothesis.
There was a weak positive correlation between serum ALT and diastolic blood pressure, but not other obesity-related disease parameters. This study did not have the statistical power to assess these relationships, which may be partly related to the homogeneity of the sample. Supporting this, nearly every subject was obese. Additionally, fasting blood glucose may not be an appropriate indicator for assessing insulin resistance as it has been demonstrated previously to fail to detect significant relationships with ALT when present.23 The association between ALT and blood pressure may be related to elevated angiotensin II, which has been proposed to promote oxidative stress, inflammation, and fibrosis in the liver.24
The observation that serum ALT levels were correlated with age is consistent with other studies, which report increasing onset of NAFLD through the second decade of life.13,25 Pediatric NAFLD has been found to be more common among boys and Hispanic adolescents, and least likely to present in Black adolescents.4,12–13,23,25–26 In this sample, boys were not more likely to have liver disease, and children with Pacific Islander surnames tended to have higher ALT values. Hispanic ethnicity appeared to be protective in this sample, which was an unexpected finding. Importantly, ethnicity based on subject surname is likely to result in some misclassification, particularly in Hawai‘i where almost one-quarter of the population is ethnically mixed.14 The distinction between Hispanic and Filipino ethnicity based on surnames can be difficult given the early occupation of the Philippines by the Spanish.27 Moreover, among Hispanics in Hawai‘i, 29% are of Mexican ancestry and 36% are of Puerto Rican ancestry compared to 63% Mexican and 9% Puerto Rican of Hispanics nationally.19 The differences in the distribution of NAFLD among adolescents by age, gender and ethnicity may be related to developmental changes with respect to sex hormones, visceral fat deposition, insulin sensitivity, and/or hepatic antioxidant defenses, although this has not been clearly elucidated.28–30
There are several limitations of this study that are noteworthy. Most patients did not have the testing necessary to rule out other causes of liver disease. However, given the high BMIs and age of the patients, NAFLD likely contributed to the majority of the observed prevalence of transaminasemia. Additionally, patients who were missing data may have been healthier, and therefore were not tested for co-morbidities, contributing to overestimation of the prevalence of these conditions in our sample. However, liver disease screening was not associated with any of the variables measured in sensitivity analysis. Finally, this is a pilot study based on a limited sampling frame and size. While it is able to provide preliminary data on liver disease on patients with metabolic syndrome, it did not have adequate power to evaluate relationships between most variables, and may not be representative of pediatric centers in other parts of Hawai‘i.
Screening for NAFLD in this population occurred much more frequently than what has been reported in other pediatric hospitals, indicating a relatively good awareness of the condition locally in this clinical setting.3 The prevalence of liver disease in this sample lend support to this practice. Future studies are needed to further evaluate the risk of NAFLD by ethnicity, and to evaluate the follow up patients that are found to have elevated ALT levels.
The research was partly supported by Award U54MD007584 from the National Center for Research Resources (NCRR), National Institutes of Health (NIH). The authors report no conflict of interest.