A randomized clinical trial published in Cell Metabolism found that a ketogenic diet reduced liver fat by roughly 67% in adults with obesity and prediabetes. That is a meaningfully larger reduction than the approximately 45% seen in both the Mediterranean and low-fat comparison groups. The read-through is almost too clean, which is exactly where the complications start.
The study enrolled 42 adults with obesity, prediabetes, and excess liver fat, dividing them into three groups of roughly 14 participants each. The keto diet derived about 4% of calories from carbohydrates and 73% from fat. The Mediterranean diet drew about 50% of calories from carbohydrates and 35% from fat. The low-fat group ran at about 70% carbohydrates and 15% fat. All three groups improved muscle insulin sensitivity by about 50%. The divergence emerged in liver-specific measures. Liver insulin sensitivity improved two to three times as much in the keto group as in the others, and the liver fat reduction was tied to a decrease in the liver's own production of new fat. Weight loss occurred across all three groups at broadly similar rates, which suggests macronutrient composition, separate from the calorie deficit, is doing meaningful metabolic work.
The prediabetes remission numbers sharpen the case. Half the keto group no longer met criteria for prediabetes by the trial's end. The Mediterranean group reached 29%. The low-fat group reached 7%.
The counterargument sits on sample size and duration. Fourteen participants per arm is a thin foundation for population-level conclusions, and the researchers themselves said the study's short length limits how broadly the findings can be applied. Ilana Muhlstein, a registered dietitian nutritionist based in California, said she does not recommend keto long-term: the high saturated fat content and typically lower fiber intake carry risks for heart health. Her read is that keto may be worth a short-term intervention for someone with prediabetes, followed by a transition to Mediterranean for longer-term heart and metabolic maintenance. She noted that cholesterol and blood-fat measurements did not increase in the keto group during the trial, though she treats that observation cautiously given the study's length.
On balance, the Cell Metabolism paper does not settle whether keto belongs in long-term metabolic disease management. What it does resolve is narrower: among people with obesity, prediabetes, and excess liver fat, diet composition appears to drive metabolic outcomes beyond what weight loss alone explains. The line to watch in any follow-on research is whether a larger, longer trial can hold the keto group's 50% prediabetes remission rate.