Cause of NAFLD: physical inactivity
Although physical training predominantly enhances muscle insulin sensitivity, cross-sectional epidemiologic data and studies on effects of physical training suggest that exercise may decrease liver fat even in the face of unchanged body weight . In a cross-sectional study in 72,359 healthy Korean adults, subjects who were exercising regularly had a 28–53% lower risk of NAFLD across almost all BMI deciles . In a small (n=12) but carefully performed mechanistic study, exercising 30–60min five times a week for 16 weeks modestly decreased liver fat by 10% in the absence of changes in body weight . The effect of exercise seems small in comparison to the effects of a hypocaloric ketogenic diet; when the same group followed a low carbohydrate diet, liver fat was shown to decrease by 30% within 48 hours . Exercise may reduce liver fat by channeling dietary carbohydrates to muscle glycogen away from DNL .
Causes of NAFLD: familial and genetic factors
Family history, ethnicity
NAFLD has been reported to run in families with heritability estimates of around 40% . This familial clustering is observed even after adjusting for age, gender, and obesity . The heritability of S-ALT due to nonalcoholic causes based on a study in twins was 55% . Ethnicity influences the prevalence of NAFLD . In the Dallas Heart Study, 45% of Hispanics, 33% of Whites, and 24% of Blacks had NAFLD . These differences were only in part attributable to differences in BMI.As discussed later, genetic variation in PNPLA3 may contribute to ethnic variation in the prevalence of NAFLD.
Age and gender
In the third NHANES, the prevalence of NAFLD increased linearly from subjects less than 30 years to those aged 50–60 years and was higher in men than in women.
The PNPLA3 I148M variant
A common genetic variant, an allele in PNPLA3 (rs738409[G], encoding I148M) has been found to be a major factor in determining liver fat content and susceptibility to NASH, independent of obesity.This variant was discovered in 2008 in a genome-wide association scan in Hispanic, African American, and European American individuals. It showed genetic variation in PNPLA3 to confer susceptibility to NAFLD. An allele in PNPLA3 (rs738409[G], encoding I148M) was associated with increased liver fat and also hepatic inflammation and fibrosis . This finding has subsequently been reproduced. In a meta-analysis comprised of 16 studies published in 2011, homozygous carriers of the variant as compared to noncarriers had a 73% higher lipid fat content, a 3.2-fold greater risk of high necro-inflammatory scores and a 3.2-fold greater risk of
developing fibrosis  (Figure 19.2). The association between the PNPLA3 variant allele and steatosis has been observed in seven out of eight GWAS studies and in multiple other studies including Chinese, African Americans, participants of the third NHANES, children with NAFLD, and morbidly obese patients . The association between the PNPLA3 rs738409[G] allele and severity of histologic liver disease has also been confirmed . As summarized in a meta-analysis, recent studies have also shown that the PNPLA3 rs738409[G] allele is associated with an increased risk of HCC . In the population-based Dallas Heart Study, the prevalence of the PNPLA3 rs738409[G] allele displayed ethnic variation, being highest in Hispanics and lowest in African Americans . The PNPLA I148M variant is indeed common. In subjects with NAFLD diagnosed by US or 1H-MRS, the prevalence of subjects homozygous vs. heterozygous for the rs738409[G] allele has averaged 12 vs. 39% in Europeans , 17 vs. 50% in Chinese , and 19 vs. 51% in Taiwanese children . In subjects without NAFLD, the prevalences were 8 vs. 31%, 11 vs. 47%, and 12 vs. 48%, respectively.
In humans, PNPLA3 is expressed predominantly in the liver. Initial in vitro studies using purified human PNPLA3 showed that the wild-type enzyme hydrolyzes triglycerides and that the I148M substitution abolishes this activity . These data suggest that the I148M substitution is a loss-of-function mutation impairing triglyceride hydrolysis. However, PNPLA3 also stimulates lipogenesis in vitro and the PNPLA3I148M variant was suggested to increase this activity .Thus, there are still uncertainties as to how the gene variant influences the biology of human NAFLD.
In almost all studies, the NAFLD associated with PNPLA3148MM/148MI was not accompanied by either insulin resistance, hyperglycemia, hypertriglyceridemia, or a low HDL cholesterol concentration, or with inflammation in adipose tissue . On the other hand, it is important to recognize that the same person may have NAFLD due to obesity/insulin resistance and NAFLD due to the PNPLA3 I148M gene variant. Genotyping for PNPLA3 at rs738049 may become part of screening strategies in the clinic for patients with steatosis as this may help in predicting risk of NASH and HCC.
Consequences of NAFLD
Hyperglycemia and hyperinsulinemia
After an overnight fast, the main action of insulin is to restrain endogenous hepatic glucose production, while the rate of glucose uptake is largely insulin-independent . In subjects with NAFLD [15,21] the ability of insulin to inhibit glucose production is impaired resulting in mild hyperglycemia. The latter stimulates insulin secretion leading to hyperinsulinemia. Fasting hyperglycemia and hyperinsulinemia are thus consequences of hepatic insulin resistance and correlate positively with liver fat, even independent of BMI (Figure 19.4).
Nondiabetic hyperglycemia and hyperinsulinemia are well established predictors of type 2 diabetes. Consistent with such data and the close association between liver fat and these predictors, NAFLD diagnosed by US has been shown to predict type 2 diabetes in multiple prospective studies (see  for review). The association between NAFLD and type 2 diabetes was independent of obesity in all studies, which adjusted for obesity. NAFLD predisposes to type 2 diabetes by increasing the amount of insulin needed to keep hepatic glucose production in the nondiabetic range. However, in subjects predisposed to develop type 2 diabetes, β cells are unable to sustain insulin secretion leading to relative β cells failure and overt type 2 diabetes (Figure 19.2) . Some studies have suggested that NAFLD predicts type 2 diabetes even independent of all components of the MetS and thus that NAFLD is an even better marker of risk than the MetS.