Of the 123 / 195 (63 %) children in NAFLD Database with liver biopsies that underwent central review, 37 were excluded because they were not within 6 months of clinical data. All of the 173 (100%) children in TONIC with liver biopsies who underwent central review were within 6 months of clinical data. One child was excluded for age < 6 years and four were excluded for absence of steatosis on liver biopsy, leaving a total of 254 children with NAFLD eligible for study inclusion. Characteristics of children included in this study are summarized in . Overall, 65 (25.6 %) met criteria for diagnosis of MetS. Among children with MetS, there was a higher proportion of females who were older in age and pubertal (Tanner stage 2 – 4). Race / ethnicity did not differ between children with and without MetS. Measures of obesity were more severe among children with MetS (higher BMI, BMI z-score, and waist and hip circumferences). Children with MetS had a higher prevalence of previously diagnosed and / or treated diabetes, hyperlipidemia, and hypertension. As anticipated, children with MetS had higher mean values of blood pressure, triglycerides, and HOMA-IR with lower mean values of HDL cholesterol and QUICKI. Frequency and distribution of the individual components of MetS are shown in . Among children with NAFLD, 33 (13.5 %) had no features of MetS and only two (0.8 %) had all five features present. Central obesity (67.1 %) and hypertension (44.6 %) were the most prevalent MetS components in this study population. About one quarter of children with NAFLD had dyslipidemia. Impaired fasting glucose was the least prevalent (12.2 %) of the MetS features.
Baseline characteristics of children with nonalcoholic fatty liver disease with and without metabolic syndrome (MetS)
Figure 1 Percentage of children with NAFLD and metabolic syndrome (MetS) features. (a) Percent with 0, 1, 2, 3, 4, or 5 features of MetS. (b) Percent with each MetS feature. Obesity: central obesity, High TG: high triglycerides, Low HDL: low level of high-density (more ...)
Because of the high percentage of children of Hispanic ethnicity in the study population, the prevalence of MetS features was evaluated according to ethnicity to determine if any important differences were observed. Hispanic children (n = 153) had no difference in measures of obesity (prevalence of central obesity 66.5 vs. 68.0 % , P = 0.80) compared with non-Hispanic children (n = 101). The prevalence of hypertriglyceridemia was lower among children of Hispanic ethnicity (22.9 vs. 33.3 % , P = 0.07), whereas the prevalence of low HDL cholesterol was higher (30.1 vs. 19.2%, P = 0.05). No difference was seen in the prevalence of HTN (47.0 % in Hispanic vs. 41.0 % in non-Hispanic children, P = 0.35). Prevalence of IFG did not differ according to ethnicity, but severity of insulin resistance by fasting insulin (43.2 ± 52.7 μU/ml vs. 28.4±19.1μU/ml, P = 0.003), HOMA-IR (9.8±13.0 vs. 6.2±4.2, P = 0.003),and QUICKI (0.30±0.04 vs. 0.31±0.04, P = 0.003) was greater among Hispanic children.
The frequency and severity of histological features of NAFLD was compared among children with and without MetS (). Severity of steatosis, hepatocellular ballooning, presence of advanced fibrosis, NAFLD pattern, and NAS were all significantly associated with a diagnosis of MetS after adjustment for sex, age, ethnicity, and Tanner stage. The risk of MetS was greatest among those with severe steatosis (OR = 2.85 for grade 3 vs. grade 1 steatosis, P = 0.001). The COR was 2.11 (P = 0.008) when comparing each level of steatosis to the ones before it. Higher levels of hepatocellular ballooning were significantly associatedwith MetS (OR = 2.01, P = 0.03). There was no significant difference in the association of MetS with the overall category of fibrosis, but those with advanced fibrosis (stage 3 / 4) had an OR for MetS of 3.21 (P = 0.04) vs. those without fibrosis (stage 0). Borderline zone 1 or definite NASH patterns compared with not NASH were strongly associated with MetS (OR = 4.44, P = 0.005 and OR = 4.07, P = 0.002, respectively). The mean NAS was greater among children with MetS vs. those without (4.8 +/− 1.4 vs. 4.3 +/− 1.4, P = 0.01).
Relationship between presence of metabolic syndrome (MetS) and histological features among children with non-alcoholic fatty liver disease
Steatosis severity, fibrosis severity, NAFLD Activity Score (NAS), hepatocellular ballooning, and NAFLD pattern were further evaluated to see if the number of MetS features or the severity of individual MetS components had any relationship to these histological variables ( and ). In addition to an association with MetS, severity of steatosis was also significantly associated with the severity of certain component features of MetS. Steatosis grade was most strongly associated with severity of HTN (COR = 2.08, P = 0.003). The MetS score or total number of features (overall P = 0.05) and central obesity as reflected by waist hip ratio (COR = 1.48, P = 0.05) were also associated with steatosis severity. Although IFG was not significantly associated with steatosis severity, insulin resistance as measured by QUICKI was (COR 0.928, P = 0.04) ().
Figure 2 Relationship between metabolic syndrome (MetS) risk factors and selected histological features in children with nonalcoholic fatty liver disease (NAFLD). (a) Ordinal logistic regression analyses of steatosis. (b) Ordinal logistic regression analyses of (more ...)
Relationship between metabolic syndrome (MetS) risk factors and NAFLD pattern in children with NAFLD
As noted earlier, overall severity of fibrosis was not significantly associated with MetS (). However, odds of more severe fibrosis were markedly increased if obesity was present as defined by BMI z-score (COR = 3.40, P <0.001) as well as central obesity (COR = 1.78, P = 0.03), increased waist circumference (COR = 1.03, P <0.001) and increased waist hip ratio (OR = 1.99, P <0.001). The only other MetS feature associated with fibrosis stage was lower HDL cholesterol (COR = 0.97, P = 0.04), though insulin resistance, as measured by fasting insulin (COR = 1.03, P = 0.002), HOMAIR (COR = 1.15, P = 0.001), and QUICKI (COR = 0.90, P = 0.002) were associated with this histological feature. When the model included BMI z-score as an additional adjustment factor, an association with central obesity and increased severity of fibrosis remained as defined by the waist hip ratio (COR = 1.74, P = 0.007), as well as insulin resistance measured by fasting insulin, HOMA-IR and QUICKI (COR = 1.03, P = 0.008, COR = 1.14, P = 0.006 and COR = 0.92, P = 0.03, respectively). The association with lower HDL cholesterol no longer remained significant (COR = 0.97, P = 0.08).
Although the presence of MetS was predictive of a higher NAS (COR = 2.12, P = 0.009), associations were also presentwith higher triglyceride levels (COR = 1.003, P = 0.04), lower HDL cholesterol levels (COR = 0.97, P = 0.02), and a diagnosis of hypertension (COR = 1.65, P = 0.04). Insulin resistance as measured by QUICKI (COR = 0.86, P <0.001) was also associated with NAS ().
Hepatocellular ballooning was associated with the presence of central obesity (OR = 2.15, P = 0.01), BMI z-score (OR = 2.17, P = 0.04), and higher triglycerides (OR = 1.0043, P = 0.02). Insulin resistance as measured by QUICKI (OR = 0.90, P = 0.01) was also associated, and for non-Hispanics, IFG was predictive of hepato-cellular ballooning (OR = 14.31, P = 0.01) ().
In addition to the strong association between diagnosis of MetS and borderline zone 1 or definite NASH vs. not NASH patterns, certain component features of MetS were also associated with the NAFLD pattern as shown in . Central obesity was highly associated with borderline zone 1 or definite NASH patterns (OR = 2.71, P = 0.02 and OR = 4.08, P = 0.001 vs. not NASH, respectively). HTN was predictive of borderline zone 1 patternvs.notNASH (OR = 3.02, P = 0.006). Although IFG was not significantly associated with NAFLD pattern, severity of insulin resistance (Fasting insulin, HOMA-IR, and QUICKI) were associated with the definite NASH pattern vs. not NASH (OR = 1.052, P = 0.01, OR = 1.283, P = 0.007, and OR = 0.786, P <0.001, respectively).