Low dietary carbohydrates and liver fat: hold the butter and fat-bombs!

(Originally appeared on Medium)

A very interesting new study was published recently on hepatic steatosis (liver fat) and low carbohydrate diets (LCD). Briefly, the researchers tried to understand, using a multi-omics approach, what happened if obese subjects with non-alcoholic fatty liver disease (NAFLD) followed a non-energy restricted diet low in carbohydrates (< 30g per day). This experimental set up is highly relevant given that energy restriction per se reduces liver fat and previous studies using LC diets in NAFLD were energy restricted. Still, at a similar level of caloric restriction and weight loss, a LCD appeared to be more efficient than a non-carbohydrate, calorie restricted diet: liver fat was reduced ~26% on a low calorie diet vs. ~55% on the low carbohydrate diet with a similar level of weight loss on both groups.

Why it makes sense

Major sources of hepatic lipids are:

  • ~59% from plasma free fatty acids (FFA).
  • ~26% from de novo lipogenesis (DNL).
  • ~15% from the diet.

Thus, an effective therapy should address both hepatic lipid re-esterification from plasma FFA and reduction of DNL. DNL is elevated in subjects with NAFLD, for which insulin resistance and hyperinsulinemia is a major contributor.

Most of the time, a LCD reduces insulin levels compared to a iso-energetic, non-carbohydrate restricted diet. In the current study, insulin levels were also reduced despite no calorie restriction. A low carbohydrate diet also reduces postprandial glucose levels.

Regulation of DNL occurs mainly at the level of transcription of enzymes involved in fatty acid synthesis by the activity of sterol regulatory element binding protein 1c (SREBP1c) and carbohydrate response element binding protein (ChREBP), and allosteric regulation of acetyl-CoA carboxylase (ACC). For a detailed description on the activity of these pathways please read this review.

Taken from Wang et al., 2015

A high glucose load will stimulate DNL via these three pathways: direct activation of ChREBP by glucose (its called carbohydrate response element binding protein after all) and SREBP1c by insulin. The expression of ACC is induced by both ChREBP and SREBP1, and activity levels of ACC are controlled by insulin, citrate and glutamate.

It is then not surprising that a eucaloric high carbohydrate (high sugar)-low fat diet increases DNL in both lean and obese subjects:

NI = normoinsulinemic, HI=hyperinsulinemic. Taken from Schwarz et al., 2003.

This was despite no weight or body fat gained during the intervention, and a reduction in fasting insulin levels. This agrees with the predominant role of glucose over insulin (via ChREBP) on regulation of hepatic lipid synthesis, as deletion of ChREBP in the liver reduces by ~65% fatty acid synthesis rates, compared to ~50% when SREBP1c is deleted (although most die in utero). However, no increase is seen when using a whole-food, complex carbohydrate diet, suggesting that sugar (and the corresponding increase in post-prandial glycemia) is a major trigger for DNL. Thus, the proportion of sugar in a high carbohydrate-low fat diet is the main determinant of hepatic lipid synthesis.

What needs to be taken into account

Getting back to the study, while the salient feature of the diet was that it restricted carbohydrates to less than 30 g per day, there were also other relevant major dietary changes:

  • Protein intake doubled from a mean intake of ~90 to 185g. Final values were around 1.76g/kg.
  • Fat intake was increased dramatically from ~99 to 248g.
  • Saturated fat (SFA) intake was almost unchanged (from ~34 to 36g).
  • Monounsaturated fat (MUFA) intake went up from ~35 to 126.5g, while polyunsaturated fat intake also tripled (from ~15 to 49g).
  • EPA (~168 to 400mg) and DHA (~322 to 1267mg) were also dramatically up.
  • Sugar intake was reduced 4.5-fold (from ~90 to 20g).

So it looks like while the diet was very high fat and protein, the sources were mainly rich in PUFA and specially MUFA, the latter which constituted most of the fat intake. It also included some marine food or fish oil, although no use of supplements was stated. A diet that could achieve a similar nutritional profile would be one that is high in protein from fish and meat such as poultry or pork, fat from marine sources, olive oil, olives, avocado and nuts (this is highly probable as subjects were encouraged to eat 200 kcal from macadamia nuts if they started to lose weight) and various vegetables as the micronutrient profile was not affected significantly. All in all, what participants followed was a high fat version of the so called “mediterranean diet”. This type of diet was HIGHLY efficacious in all participants (every subject reduced their liver fat), as shown in the following figure:

Individual change in body weight (panel C) and liver fat (%) (panel E), and mean changes in weight composition (panel D). For Panel C and E subjects are colored the same.

Was it the high fat or the high protein?

From the molecular evidence we currently have, the success of this diet appears to be due to the low carbohydrate (specially sugar) and the high protein in the diet, provided through complex, whole foods.

In addition to the effects due to reduced glucose availability (explained above), protein promotes glucagon secretion. A high protein-low carbohydrate diet increases total glucagon levels in type 2 diabetic subjects. Glucagon inhibits SREBP1c activity in the liver, thereby reducing lipogenesis. This effect appears to be mediated by a protein called Insig-2a, which inactivates SREBP1c. Interestingly, similarly to glucagon, fasting increases the expression and protein levels of Insig-2a.

Edit 28/02: High protein diets seem to decrease liver fat in humans (see also here) and rodents (see also here and here).

Finally, there is some evidence that the nature of the fatty acids in the diet might be important for hepatic steatosis. In the current diet, subjects consumed mainly MUFA and increased their intake of marine omega 3 fatty acids (DHA & EPA) significantly. In mice, DHA & EPA supplementation reduces SREBP1 activity and activates PPAR-alpha in the liver, ameliorating hepatic steatosis. There also appears to be different effects of SFAs versus MUFAs. It looks like some of the effects are due to the different intracellular metabolism of SFA vs. MUFA; esterification of SFA is not as efficient as for MUFA and thus, diacyl glycerol (DAG) and ceramides accumulate.

In vitro, SFA induce endoplasmic reticulum stress. Interestingly, ER stress promotes the proteolytic cleavage (activation) of SREBP1c independently of insulin.


This study further adds evidence towards a low carbohydrate diet as a primary intervention for hepatic steatosis, which in my opinion, is also the most efficient nutritional intervention for several metabolic abnormalities.

However, as shown by the dietary data, previous evidence and molecular knowledge, the higher protein and fatty acid composition also plays an important role on the observed effects. Thus, the results from this study can’t be directly extrapolated to other types of low-carb diets, and definitely not to those in which copious amounts of butter, cream and fat-bombs (and specially under high calorie conditions) are included concomitantly with low protein.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s