Which diet works best for prediabetes? A review of 10 trials finds no single winner

Prediabetes is common and dangerous, yet it is also one of the most reversible metabolic conditions. Prevalence ranges from about 3.0% in Southeast Asia to 15.4% in North America, and the condition independently raises the risk of mortality, cardiovascular disease, and microvascular complications. Awareness is low: only about 15% of people know they have it, and just 18.2% of patients realize weight loss can reverse it. The 2026 American Diabetes Association Standards of Care recommend 7%-10% weight loss, but the last ADA nutrition consensus dates to 2018 and drew only on PubMed.

The authors followed PRISMA 2020 and prospectively registered the protocol (PROSPERO CRD42023458958). They searched PubMed, Embase, the Cochrane Library, and Web of Science through August 23, 2025. Of 1,126 screened records, 105 reached full-text review and ten randomized trials (1,098 participants) were included. Trials lasted 3 weeks to 3 years, and most participants were over 55. Risk of bias was assessed with Cochrane RoB 2. Because interventions, comparators, and outcomes varied so much, the authors skipped meta-analysis and synthesized the 18 diet arms narratively in four groups.

Calorie- and carbohydrate-restricted diets produced the fastest results. One diet cut weight by 5.6 kg and fat by 4.5 kg at 12 weeks, with modest HbA1c declines and early lipid gains that attenuated by 52 weeks. A very low-calorie, low-carbohydrate program achieved roughly 11% weight loss at two months and about 40% reversion to normoglycemia, with partial maintenance despite weight regain. Longer-term carbohydrate restriction sustained fasting glucose reductions and improved triglycerides and cholesterol at 12 months, though some individuals may see unfavorable LDL responses.

Carbohydrate quality mattered even without weight loss. Low-glycemic, fiber-rich diets improved β-cell function, postprandial glucose, and blood pressure. A whole-grain, legume, and vegetable pattern consistently lowered fasting glucose, insulin, HbA1c, HOMA-IR, and inflammatory markers independent of weight change. Mediterranean and traditional Jiangnan diets produced about 7% weight loss over six months with lower fasting glucose and less ectopic fat. An anti-inflammatory diet gave the largest metabolic and hs-CRP improvements, while one low-carbohydrate, high-MUFA diet raised HbA1c and postprandial triglycerides and reduced β-cell function.

Intermittent fasting results differed by regimen. Alternate-day fasting caused the greatest weight loss (up to 14.7 kg) and improved fasting glucose, but had limited or adverse effects on triglycerides. Time-restricted eating gave moderate weight loss with more favorable lipid and glucose effects. Weekly 24-hour fasting improved insulin sensitivity and metabolic syndrome components without marked weight change, and the 5:2 regimen reduced adiposity, fasting glucose, and HbA1c but with moderate adherence challenges and variable lipid and inflammatory responses.

The evidence is thin. Five of ten trials (50%) were at high risk of bias, four (40%) raised some concerns, and only one (10%) was low risk. Deviations from intended interventions were the weakest domain, with 70% of trials showing some concerns. The authors conclude that no single diet is universally optimal. Choice should match the goal, such as rapid weight loss versus durable insulin sensitivity, and the patient's ability to adhere.