A recent clinical trial finds that a ketogenic diet can significantly decrease liver fat and improve metabolic markers in adults with obesity and prediabetes, sparking renewed discussion on its long-term safety and applicability.
A ketogenic diet reduced liver fat sharply in a clinical trial of adults with obesity, prediabetes and fatty liver disease, adding fresh evidence to a long-running debate over whether very low-carbohydrate eating offers advantages beyond weight loss. Researchers at Washington University School of Medicine, writing in Cell Metabolism, found that people on keto cut liver fat by 67% over about five months, even though participants on a Mediterranean diet and a low-fat, plant-forward diet lost roughly the same amount of weight.
The trial enrolled 55 adults and provided all meals, with 42 people completing the study. According to the report, the ketogenic group ate just 4% of calories as carbohydrate and 73% as fat, while the other two diets were much higher in carbohydrate and lower in fat. All three groups improved muscle insulin sensitivity by around 50%, but the keto arm saw the strongest liver response and the greatest shift away from prediabetes.
That matters because the burden of glucose dysregulation remains enormous in the United States. The Centres for Disease Control and Prevention says about 40.1 million Americans have diabetes and 115.2 million adults have prediabetes. In that context, even modestly different dietary effects on liver fat and blood sugar control can be clinically important, especially for people already at elevated metabolic risk.
The study’s results were encouraging in another respect: cholesterol did not worsen in the ketogenic group. Researchers reported no meaningful differences between diets in LDL cholesterol, apolipoprotein B or 24-hour triglycerides. First author Max Petersen said the keto participants “didn’t experience an increase in blood fat or cholesterol”, but the finding should be read cautiously, because the trial was small, tightly controlled and relatively short.
That caution is well founded. The same broad pattern of ketogenic eating has produced very different lipid responses in other research. A feeding trial in Nutrients found that a ketogenic low-carbohydrate, high-fat diet nearly doubled LDL cholesterol and apolipoprotein B in healthy, young, normal-weight women. Specialists say that variation is not unusual: body composition, genetics, the type of fat consumed and whether someone is losing weight can all alter the outcome.
The wider literature also suggests that keto is not a universally benign intervention. A 2026 systematic review in BMC Nutrition found adverse events in 43% of participants across 36 studies, with constipation the most common complaint and gastrointestinal problems the largest category overall. The review also noted that stricter diets tended to bring more frequent and more severe side effects, while drop-out rates averaged 17%.
For that reason, many clinicians frame keto as a tool rather than a default prescription. A review in Annals of Medicine says people with porphyria, certain rare metabolic disorders, advanced liver or kidney disease, or familial hypercholesterolaemia may need to avoid it or use it only with close supervision. Caution is also advised for those taking insulin, sulfonylureas, SGLT2 inhibitors and other medicines that can complicate rapid changes in carbohydrate intake.
Even among people who can tolerate the diet, long-term quality matters. The more carefully designed versions keep non-starchy vegetables, emphasise unsaturated fats and limit refined grains and added sugars, yet they still remove fruit, legumes and whole grains, which can reduce fibre and some micronutrients. In the end, the Washington University trial adds weight to the argument that keto can work well for some patients with fatty liver disease and prediabetes, but it does not settle the broader question of who should use it, for how long or under what level of medical supervision.
Disclaimer: This content is for informational purposes only and is not intended to be a substitute for professional medical judgment, advice, diagnosis, or treatment.
