2026-09-07 · Keto News Desk · 1255 words · Lab-Verified Data

Keto Cut Liver Fat 67% in a Strict Trial — What the Numbers Say

What Happened

A randomized clinical trial from Washington University School of Medicine in St. Louis, published August 27, 2026 in Cell Metabolism (Petersen, Smith, Farabi et al., doi:10.1016/j.cmet.2026.07.020), tested whether the type of diet matters for metabolic health once the amount of weight lost is held equal. Fifty-five adults with obesity, prediabetes and fatty liver were assigned to one of three diets for about five months; forty-two completed the study. Every meal was provided by the research team and participants met a study dietitian weekly, with reported adherence above 95%. The three diets differed sharply: a very-low-carbohydrate ketogenic diet (4% of energy from carbohydrate, 73% from fat), a Mediterranean diet (50% carbohydrate, 35% fat), and a very-low-fat plant-forward diet (70% carbohydrate, 15% fat). Calories were adjusted individually so all three groups lost about the same — roughly 10% of starting body weight.

That design detail is what makes this trial worth your attention. Most diet studies compare plans that produce different amounts of weight loss, so you cannot tell whether the benefits came from the food or the pounds. Here, the pounds were matched on purpose, which isolates what the diet composition itself does. The work was funded by the National Institutes of Health and the Foundation for Barnes-Jewish Hospital, per the trial's published disclosures.

The Headline Numbers

Where the groups diverged was the liver. Measured by MRI-based fat imaging, liver fat fell 67% in the ketogenic group versus 45% in each of the other two — the same weight loss, roughly half again the improvement. Hepatic insulin sensitivity, which measures how well liver cells suppress glucose production, improved two to three times more in the keto group than in the Mediterranean or low-fat groups (WashU Medicine press release, August 27, 2026).

Blood sugar followed the same pattern. Twenty-four-hour glucose dropped 20% from baseline on the ketogenic diet versus 8% on the other two, and 24-hour insulin fell 74% on keto against 44% for Mediterranean and 27% for the low-fat plan. By the end, about half the ketogenic group no longer met the criteria for prediabetes, compared with 29% of the Mediterranean group and 7% of the plant-forward group.

Two results deserve equal billing. First, muscle insulin sensitivity improved about 50% in all three groups — on that measure, the diet composition did not matter; the weight loss did. Second, despite the ketogenic diet deriving 23% of its energy from saturated fat, the trial found no between-group differences in LDL cholesterol, apolipoprotein B, or 24-hour triglycerides, and no serious adverse events were reported. Those are group averages over five months, not individual guarantees — blood lipid responses to very-low-carb diets vary considerably person to person.

What It Means If You Eat Low-Carb

The useful takeaway is a division of labor: weight loss itself repaired muscle insulin sensitivity everywhere, but the carbohydrate level changed what happened in the liver. If you are eating keto and losing weight, this trial is evidence that the diet's composition is doing specific work beyond the scale — at least for people who started with obesity, elevated blood sugar and liver fat, which is who was studied.

Read the conditions honestly, as the authors themselves caution. Participants received all their food free and met a dietitian weekly — real-world keto means supermarket groceries and self-discipline. Forty-two analysed participants is small, five months is short, and the trial tested a strict therapeutic version of keto (4% carbohydrate) that is far more restrictive than the casual "low-carb-ish" pattern most people follow. People who take diabetes medication, have kidney conditions, are pregnant, or have a history of disordered eating generally need medical supervision before attempting it — we are a data editorial team, not clinicians, so treat that as a pointer to your doctor rather than advice.

A Data Editor's Read: What 4% of Energy From Carbohydrate Looks Like

Numbers like "4% of energy from carbs" are abstract until you put food on a plate, so we ran the arithmetic with our own verified library. A breakfast of two fried eggs — about 92 g of egg in total — runs roughly 0.8 g net carbs (0.83 g per 100 g, USDA SR Legacy, FDC #173423), a side of buttered spinach adds roughly 1.5 g net carbs per 100 g spinach with butter at 0.06 g (FDC #168462, FDC #173410), and half an avocado adds about 1.4 g net carbs for a 75 g half (1.83 g per 100 g, FDC #171705). The whole plate lands near 4 g net carbs — before the coffee with a splash of cream. That is the neighborhood the trial's keto arm lived in: eggs, fish, meat, hard cheeses, leafy greens and fats, with grains, potatoes, rice and fruit essentially absent.

Our verdict library prices those staples exactly this way: compare the plate above with a 100 g serving of potato at 15.39 g net carbs (FDC #170026) or rice at roughly 20 g (our rice verdict has the full card), and you can see why a 4%-carbohydrate menu has no room left for staple starches. If you want the mechanics — why total carbs minus fiber equals the number we track — the beginner's guide to keto and our potato verdict both show the arithmetic on real foods.

Source Box

We reported only figures that appear in the press release or the published abstract; where the trial reported group averages, we wrote group averages. This article is a plain-language summary of published research by a data editorial team — it is not medical advice, and it does not replace a conversation with a qualified professional.

Did the keto group lose more weight than the other diets?

No — and that is the point of the design. Calories were individually adjusted so all three groups lost approximately 10% of starting body weight over 4–5 months. The liver-fat (67% vs 45%) and blood-sugar differences (20% vs 8% in 24-hour glucose) emerged at equal weight loss, which is what lets the researchers attribute them to diet composition rather than to the pounds lost. Source: Petersen et al., Cell Metabolism, 2026, doi:10.1016/j.cmet.2026.07.020; WashU Medicine press release.

Does this prove keto is the best diet for fatty liver?

"Prove" is too strong, and we will not say it. The trial shows that, in 42 adults with obesity, prediabetes and fatty liver, under fully controlled feeding and matched weight loss, the ketogenic diet produced larger improvements in liver measures than two comparison diets over five months. It does not tell us about longer horizons, about people without those conditions, or about unsupervised at-home adherence. Source: WashU Medicine press release, August 27, 2026. Discuss any diet change for a liver condition with your clinician.

Why did cholesterol not rise on the high-saturated-fat keto diet?

The trial's group averages showed no differences in LDL cholesterol, apolipoprotein B or 24-hour triglycerides between the keto group (23% of energy from saturated fat) and the other diets. That is a real reported result, but averages hide individuals: responses to very-low-carbohydrate diets vary substantially, and some people do see LDL rises. Five months in 42 participants cannot rule out uncommon effects. Source: Petersen et al., Cell Metabolism, 2026, doi:10.1016/j.cmet.2026.07.020.

Related reading

Filed under newsresearchweight-loss