A low-carbohydrate diet rapidly and dramatically reduces intrahepatic triglyceride content


  • Potential conflict of interest: Nothing to report.

During November of 2005, a 42-year-old white woman taking conjugated estrogen since age 16 was seen for evaluation of right upper quadrant pain of 6 months' duration. She was obese with mixed hyperlipidemia but had no evidence of diabetes mellitus or impaired fasting glucose. Initial workup indicated hepatomegaly (craniocaudal dimension: 26 cm) with an increased serum alanine aminotransferase concentration (55 U/L). Viral and autoimmune serologies were negative, and no heritable or environmental (i.e., ethanol) cause of liver disease was identified. Ultrasound examination of the liver was consistent with diffuse fatty infiltration of the liver. A diagnosis of non-alcoholic fatty liver disease (NAFLD) was confirmed by liver proton magnetic resonance spectroscopy, which demonstrated a liver triglyceride content of 44.6% (normal: ≤5.5%)1 (Fig. 1). Discontinuation of conjugated estrogen was unsuccessful at relieving symptoms, and because of the patient's body mass index (32.2 kg/m2), dietary therapy was initiated to promote weight loss. The diet chosen was the Atkins diet, emphasizing the selection of healthy sources of protein and fat.2

Figure 1.

Serial proton magnetic resonance spectra. Spectra were obtained from a 27-cm3 voxel positioned within the liver to avoid major blood vessels, bile ducts, and the liver edge. Liver triglyceride content is calculated as the ratio of the area under the triglyceride signal (methylene groups: (−CH2−)n) to the area under both the water signal (H2O) and the triglyceride signal. The spectrum depicted in blue was obtained before initiation of therapy and corresponds to a liver triglyceride content of 44.6%. The spectrum depicted in red was obtained 5 weeks after initiation of a low-carbohydrate diet and a moderate exercise program and corresponds to a nearly fourfold reduction in liver triglyceride (11.9%).


NAFLD, non-alcoholic fatty liver disease.

At day 7 of the diet, she was ketotic with a urinary ketone concentration of 150 mg/dL. Liver biopsy was performed at this visit and showed pan-lobular macrovesicular steatosis without evidence of inflammatory activity or fibrosis. At day 14 of the diet, she remained ketotic with a respiratory quotient of 0.72 by indirect calorimetry. Body mass index had fallen to 30.9 kg/m2 (−6.4 kg) with a modest decrement in abdominal pain. Dietary composition over this 2-week period was 54% protein, 41% fat, and 5% carbohydrate with an estimated daily caloric intake of approximately 1,000 kcal. Saturated, monounsaturated, and polyunsaturated fats accounted for 57%, 35%, and 8% of fat intake, respectively.

The relative contribution to hepatic glucose production from all sources (glycogen, glycerol, phosphoenolpyruvate) was determined by nuclear magnetic resonance spectroscopy of monoacetone-glucose after ingestion of deuterium oxide in this patient.3 Glycogenolysis accounted for only 25% of hepatic glucose produced, whereas glycerol gluconeogenesis accounted for 6%. In contrast, glycogenolysis accounts for 53% and glycerol gluconeogenesis for only 3% of newly formed glucose on a Western diet.3 The patient was instructed to continue dieting and initiate a low-impact exercise regimen.

Five weeks after initiating the diet, her body mass index was 29.9 kg/m2 (−9.1 kg), her abdominal discomfort was significantly improved, and her serum alanine aminotransferase level was normal (32 U/L). Her exercise regimen consisted of the use of an elliptical machine and low-weight strength training for 30 minutes 3 times per week. Repeat proton magnetic resonance spectroscopy demonstrated a nearly fourfold reduction in liver triglyceride content to 11.9% (Fig. 1).

This report shows the dramatic effect of a low-carbohydrate diet and moderate exercise on hepatic steatosis. Triglycerides were rapidly mobilized from liver despite significant dietary fat intake. In addition, a dramatic alteration occurred in pre-cursor utilization for hepatic glucose production as compared with a typical Western diet. This change in hepatic glucose metabolism occurred rapidly (within 2 weeks of initiating the diet) and is not observed in individuals on a controlled Western diet with or without calorie restriction (unpublished data); whether these changes in metabolism provide a benefit to liver function/histology greater than those obtained from weight loss alone remains to be determined.

Currently, the low-calorie, low-fat diet is recommended for weight reduction in clinical practice.4 A study of 41 morbidly obese, non-alcoholic subjects has demonstrated that this dietary approach (58% protein, 11% fat, 31% carbohydrate; 388-900 kcal/d) leads to a significant reduction in liver triglyceride content.5 However, the number of subjects with portal inflammation in this study more than doubled at the end of the treatment period (pre-diet: 15%, post-diet: 34%; P = .039); a finding that was unrelated to any identifiable histological/demographic variable or the rapidity of weight loss.5 The macronutrient composition of the diet recommended for weight loss therefore may be clinically relevant in patients with NAFLD. In fact, a recent cross-sectional study of morbidly obese subjects undergoing obesity reduction surgery demonstrated a relationship between dietary macronutrient composition and hepatic inflammation: high carbohydrate or low fat intake was associated with significantly higher odds of inflammatory activity.6 Because of the cross-sectional nature of this study, a cause-and-effect relationship between subject diet composition and liver histology cannot be made; however, the data are intriguing.

Patients with NAFLD are at increased risk for cardiovascular disease and endothelial dysfunction.7 Some concern has been expressed that the increased dietary fat intake as a result of a low-carbohydrate diet may be deleterious to the cardiovascular risk profile in such patients. However, two recent studies have shown that the long-term use of a low-carbohydrate diet (6-14 months) has a favorable effect on serum cholesterol, lipoprotein sub-fractions, and high-sensitivity C-reactive protein despite the increased intake of dietary fat.8, 9 However, the use of a low-carbohydrate diet does result in increased serum levels of chylomicrons and chylomicron remnants, both of which are capable of contributing to vascular plaque formation.8

A theoretical concern exists that increased hepatic lipid oxidation as a result of decreased dietary carbohydrate intake could exacerbate the inflammatory milieu via enhanced free-radical production.10 Patients on a low-carbohydrate diet do indeed exhibit elevated levels of serum and urinary ketone bodies, presumably because of increased availability of acetyl-CoA (derived from lipid oxidation) to hydroxymethylglutaryl coenzyme A synthase, the first enzymatic step in the synthesis of β-hydroxybutyrate and acetoacetate. The current report can provide no information on the long-term histological effects of a low-carbohydrate diet, because repeat liver biopsy was not performed at the end of the 5-week treatment period. However, 1 week of carbohydrate restriction was not associated with significant histological inflammatory activity in this patient.

We propose that the “one-size-fits-all” approach to weight reduction currently recommended should be reexamined. Though weight reduction should be a key component of the approach to overweight patients, the proportion of dietary macronutrients available during that weight reduction may be important in different disease states and different individuals. The low-carbohydrate diet is one alternative to the traditional low-calorie, low-fat diet; however, it is by no means the only alternative. Dietary therapy should be individualized to promote patient compliance and to minimize potential adverse outcomes. A greater understanding of the effect of dietary macronutrient manipulation may be important for achieving these goals.