Division of Nutrition and Metabolic Diseases Department of Internal Medicine and the Center for Human Nutrition The University of Texas Southwestern Medical Center at Dallas Dallas, Texas 75390-9052
Address all correspondence and requests for reprints to: Abhimanyu Garg, M.D., The University of Texas Southwestern Medical Center at Dallas, 5323 Harry Hines Boulevard, Dallas, Texas 75390-9052. E-mail: . Abhimanyu.garg{at}utsouthwestern.edu
For a long time, fatty liver or hepatic steatosis was considered as a benign manifestation. However, recent data indicate a wide spectrum of clinical and pathological manifestations that subjects with nonalcoholic hepatic steatosis develop, which are termed as nonalcoholic fatty liver disease (NAFLD). Interestingly, the manifestations of NAFLD are similar to those seen in patients with alcoholic liver disease and range from mild hepatic steatosis, steatohepatitis, fibrosis, to cirrhosis (1, 2, 3), and, rarely, to hepatocellular carcinoma (4, 5, 6).
Nonalcoholic hepatic steatosis is usually found in obese subjects (1, 3, 3, 7). Recent studies have documented a strong relationship between hepatic steatosis and insulin resistance (1, 8, 9, 10, 11, 12, 13), which may account for hepatic steatosis in nonobese and even lean subjects (11, 12, 13, 14). The prevalence of other metabolic abnormalities associated with insulin resistance such as abnormal glucose tolerance, hypertriglyceridemia, and low levels of high-density lipoprotein cholesterol is also high in subjects with NAFLD (2, 9, 10).
The underlying mechanisms for the association of insulin resistance and obesity to hepatic steatosis remain unclear. To explain these relationships, it has been hypothesized that excess of "portal" or intraperitoneal fat can increase flux of free fatty acids via portal vein directly to the liver and, thus, may induce hepatic insulin resistance and hepatic steatosis (15). In contrast, we previously reported that hepatic insulin resistance in men was more strongly related to sc abdominal fat mass than to intraperitoneal fat mass (16, 17). The possibilities that intraperitoneal fat mass may preferentially affect hepatic very low-density lipoprotein-triglyceride secretion and hepatic fat accumulation still remain. Thus, the relationships between regional adiposity, insulin resistance, and hepatic steatosis continue to be an area of active investigation.
Fatty liver infiltration can be determined by many different methods. Although the direct measurement of hepatic fat using a biopsy is considered the "gold standard," its use is limited due to risks involved and availability of very small sample of tissue (0.01 to 0.05 cc), which in case of inhomogeneous fat distribution may not provide an accurate estimate. Ultrasound, computerized tomography (CT), magnetic resonance imaging (MRI), and 1H magnetic resonance spectroscopy (1H MRS) are noninvasive and should be used instead. However, ultrasound does not provide reliable quantitative information. Both the CT and MRI techniques are nonspecific and can be affected by various processes such as excess glycogen accumulation, edema, inflammation, etc. The best method for frequent, repetitive, and highly specific estimation of hepatic fat in vivo is localized 1H MRS. Methylene proton signals estimated by spectroscopy are specific for the mobile triglycerides and create a clear resonance peak. The 1H MRS method has been validated against direct determination of triglyceride content of liver biopsies in the animals (18) as well as in humans (19) and has become the method of choice. Additionally, because 1H spectra are collected from a large volume (
10 cc or more) of liver during MRS, it may provide a better estimate of average hepatic fat than liver biopsy.
Most investigators define hepatic steatosis as triglyceride content exceeding 5% of the liver weight (20), however, normal values have not been determined precisely. The 1H MRS revealed no liver fat in 3 normal subjects (21), and so did CT scanning in 20 healthy subjects (22). Therefore, clinically meaningful hepatic steatosis could be present even at less than 5% triglyceride content.
In this issue of JCEM, Seppälä-Lindroos et al. (23) report their observations on hepatic fat (measured with 1H MRS), regional adiposity (intra-abdominal, sc abdominal, and total abdominal fat measured with MRI), and hepatic insulin sensitivity (measured using euglycemic, hyperinsulinemic clamp technique) in 30 healthy men. Subjects with high liver fat, compared with those with low liver fat, had reduced suppression of endogenous glucose production and serum free fatty acids during the hyperinsulinemic phase of the clamp study. The two groups showed no significant differences in intra-abdominal and sc abdominal fat volume. In regression analyses, suppression of hepatic glucose output was significantly related to hepatic fat content independent of body mass index and sc abdominal fat volume, however, it explained only 16% of the variance of hepatic fat content. The authors concluded that fat accumulation in liver is independent of intra-abdominal adiposity and overall obesity and that hepatic fat rather than adipose tissue triglyceride content causes insulin resistance in humans. These conclusions, however, must be interpreted with caution.
First, the statistical methods used in arriving at the conclusions need to be commented on. The conclusions were based mainly on dichotomizing their data using the median split to form low and high liver fat groups. This practice can result in loss of power and interpretive errors, particularly when the variable of interest such as liver fat is a continuous variable (24, 25). Furthermore, variables such as percentage suppression of glucose rate of appearance and alcohol intake, which had a skewed distribution, should have been compared using nonparametric tests or by transforming the data.
Second, although the inclusion criteria reported alcohol consumption of less than 20 g/d, the average alcohol consumption was higher in the group with high liver fat content compared with that with low liver fat content, although the difference was reported to be not statistically significant. The wide SD of alcohol intake in the high liver fat group suggests that a few subjects may be consuming alcohol in excess of 20 g/d. From the purists standpoint, to avoid the confounding effects of alcohol intake, the best documentation of such relationship should be conducted in subjects who do not consume alcohol.
Third, the values of total body fat, intra-abdominal fat, and sc abdominal fat were higher in the group with high liver fat by 10.5%, 4%, and 18%, respectively, although the differences were not statistically significant. The lack of statistical significance among variables may be related to the small number of subjects. Thus, these differences should not be ignored, particularly because they may be pathophysiologically important. Therefore, at least the results of rate of endogenous glucose appearance during the hyperinsulinemic phase of the glucose clamp study should have been adjusted for the confounding effects of differences in these anthropometric variables and in alcohol intake (26). It would also be of interest to adjust for intra-abdominal fat volume and alcohol intake in regression analysis between suppression of hepatic glucose output and hepatic fat content to support the authors conclusions.
Theoretically, intraperitoneal and retroperitoneal fat compartments may play different roles in causation of hepatic steatosis. For example, free fatty acids released from the intraperitoneal fat drain directly to the liver through the portal veins whereas those from the retroperitoneal fat drain to the systemic circulation. Therefore, it would have been interesting to separately quantitate the volume of these two compartments. Because no women participated in the study, these findings may not be applicable to women who, in fact, have higher prevalence of nonalcoholic hepatic steatosis than men (27). The factors contributing to hepatic steatosis in women may also be different than those in men.
Interestingly, hepatic steatosis is not only confined to subjects with generalized or regional obesity but also occurs frequently in other disorders of adipose tissue such as lipodystrophies (28). Lipodystrophies are characterized by selective loss of adipose tissue from different regions of the body. The severity of hepatic steatosis seems to be related to the extent of fat loss. Patients with congenital generalized lipodystrophy (CGL) and acquired generalized lipodystrophy, who have near total absence of adipose tissue, have marked hepatic steatosis, which can lead to cirrhosis (28). Specifically, in patients with CGL, hepatic steatosis has been reported during early childhood (29, 30). In patients with familial partial lipodystrophies, who have loss of sc fat restricted to the extremities, usually mild hepatic steatosis occurs (31), although cirrhosis requiring liver transplantation has been reported in an atypical patient (32). The underlying mechanisms for hepatic steatosis in lipodystrophies are beginning to emerge with elucidation of the molecular basis of inherited lipodystrophies. A simple explanation, though, may be related to the limitation in triglyceride deposition in adipose tissue storage depots, which may divert triglycerides for accumulation in other tissues such as the liver. In obese patients, there may be a similar limitation in triglyceride deposition in adipocytes that already have extremely high lipid stores. Thus, triglycerides may be diverted and deposited in other tissues including liver.
Recently, our group reported mutations in the AGPAT2 gene encoding 1-acylglycerol-3-phosphate O-acyltransferase in patients with CGL linked to chromosome 9q34 (33). AGPAT2 is the key enzyme involved in the biosynthetic pathways of triacylglycerol and glycerophospholipids from glycerol-3-phosphate in eukaryocytes. Of the five isoforms of AGPAT (1, 2, 3, 4, and 5), AGPAT2 expression was at least 2-fold higher than AGPAT1 in the adipose tissue. In the liver, both AGPAT1 and AGPAT2 were expressed equally. The other isoforms, AGPAT3, 4, and 5, were barely expressed. Thus, although mutations in AGPAT2 may cause lipodystrophy by inhibiting triacylglycerol synthesis and storage in adipocytes, the liver may accumulate triacylglycerol through AGPAT1-mediated synthesis. The other gene mutated in CGL, BSCL2 (Berardinelli-Seip congenital lipodystrophy 2), encodes a protein, Seipin, of unknown function (34), and, thus, how BSCL2 mutations cause hepatic steatosis remains unclear.
Mutations in LMNA (lamin A/C) and PPARG (peroxisome proliferator-activated receptor-) genes cause autosomal dominant familial partial lipodystrophies (35, 36, 37, 38). Recently, lamin A, a component of nuclear lamina, was found to interact with sterol regulatory element binding proteins 1 and 2 (39), however, how this interaction can lead to fatty liver in patients with familial partial lipodystrophy of Dunnigan variety remains to be determined. Whether patients with familial partial lipodystrophy due to PPARG mutations develop hepatic steatosis is not clear (38).
There is also a possibility that the relationship of adipose tissue disorders including regional and generalized obesity and lipodystrophies to hepatic steatosis may be due to reduced central or peripheral actions of leptin, an adipocyte-derived hormone. High plasma levels of leptin have been related to liver steatosis and steatohepatitis in the obese and nonobese patients (40, 41). In these subjects, leptin resistance may occur centrally or at the level of liver. Patients with severe generalized lipodystrophies who have reduced blood leptin levels are also susceptible to hepatic steatosis. Hepatic steatosis is also observed in ob/ob and db/db mice that have leptin and leptin receptor mutations, respectively. Similarly, patients with congenital leptin deficiency due to leptin mutations and those with leptin resistance due to leptin receptor mutations should also have marked hepatic steatosis, however, none of the patients described so far have been reported to have liver enlargement or hepatic steatosis (42, 43, 44, 45).
In a recent collaborative trial, we reported that leptin replacement therapy in hypoleptinemic patients with lipodystrophies markedly reduced liver volume and improved liver functions, in addition to lowering serum triglycerides and glucose concentrations (46). Leptin therapy also reduced hepatic triglyceride content drastically (47, 48). It remains to be ascertained, however, whether leptin-induced improvement in hepatic steatosis was mainly due to reduction in energy intake and weight loss or could have been partly related to direct action of leptin on the liver.
Finally, whether hepatic steatosis is a consequence of hepatic or peripheral insulin resistance or whether hepatic steatosis causes hepatic insulin resistance remains unclear. It is more likely that excess free fatty acid flux due to peripheral insulin resistance may induce hepatic steatosis. On the other hand, excess fat deposition in the liver may render hepatocytes less sensitive to insulin action and lead to hepatic insulin resistance. It would also be of interest to identify factors leading to progression of hepatic steatosis to steatohepatitis, cirrhosis, and hepatocellular carcinoma.
Footnotes
This work was supported by NIH Grants R01-DK54387 and M01-RR00633.
Abbreviations: CGL, Congenital generalized lipodystrophy; CT, computerized tomography; 1H MRS, 1H magnetic resonance spectroscopy; MRI, magnetic resonance imaging; NAFLD, nonalcoholic fatty liver disease.
Received May 2, 2002.
Accepted May 5, 2002.
References