Table 2.
Studies investigating the interaction between different diets and pretreatment insulinemia status or markersa
| Reference | Study design and country | Participants | Duration and number of participants | Prescribed diets | Groups | Results |
|---|---|---|---|---|---|---|
| Ballesteros-Pomar et al. (7) | Prespecified RCT; Spain | BMI = 28–35 kg/m2; no participants with diabetes | 16 weeks; n = 36 | −1,000 kcal/day of energy needs Low-carbohydrate diet: C 40%: P 30%: F 30% High-carbohydrate diet: C 55%: P 15%: F 30% |
Insulin resistant (n = 21): FI ≥ 15 mU/L and/or HOMA-IR ≥ 3.8 and peak insulin after OGTT >100 mU/L Insulin sensitive (n = 15): remaining participants |
Insulin-resistant participants lost 2.6 kg more (p = 0.37) on the low-carbohydrate diet; insulin-sensitive participants lost 0.7 kg more (p = 0.74) on the low-carbohydrate diet |
| De Luis et al. (18) | Unrandomized, uncontrolled trial; Spain | BMI = 34.6 ± 5.3 kg/m2; no participants with diabetes | 3 months; n = 76 | Prescribed consumption: 1,200 kcal/day, with C 51%: P 18%: F 31%; reported consumption: ≈1,500 kcal/day | Groups based on tertiles of HOMA-IR | No difference between tertiles of HOMA-IR (p > 0.05) |
| Pittas et al. (57) | Prespecified RCT; United States | BMI = 25–29.9 kg/m2; FPG < 100 mg/dL | 24 weeks; n = 32 | −30% of energy needs High-glycemic-load diet: C 60%: P 20%: F 20% with 15 g fiber and 1,000 kcal, GI = 86 and glycemic load = 116 g/1,000 kcal Low-glycemic-load diet: C 40%: P 30%: F 30% with 15 g fiber/1,000 kcal, GI = 53 and glycemic load = 45 g/1,000 kcal |
Groups separated by the median insulin-30 value into low [<473 pmol/L (66 mU/L)] and high [>473 pmol/L (66 mU/L)]; also used HOMA-IR | Participants with high insulin-30 values lost ≈4 kg more (p = 0.047) on the low-glycemic-load diet than on the high-glycemic-load diet; participants with low insulin-30 values lost ≈2 kg more (p = 0.31) on the high-glycemic-load diet than on the low-glycemic-load diet (≈Δ6 kg; no p value reported); no diet by HOMA-IR interaction was found (no p value reported) |
| Chaput et al. (13) (Quebec Family Study) | Observational; Canada | BMI = 24.5 ± 4.6 kg/m2 (low-fat tertile); BMI = 27.1 ± 6.7 kg/m2 (high-fat tertile); no participants with diabetes | 6 years; n = 184 | Lowest tertile of self-reported fat intake (<30% total energy intake) versus highest tertile (>36% total energy intake) | Groups separated by tertiles of insulin-30 values into lowest (<300 pmol/L), top (300–525 pmol/L), and highest (>525 pmol/L) | Participants with high insulin-30 values in the lowest dietary fat tertile gained 1.8 kg more weight (p = 0.034) when compared with the highest dietary fat tertile; no difference was observed among those with low or medium insulin-30 values (p > 0.05) |
| Ebbeling et al. 2007 (20) | Prespecified RCT; United States | BMI ≥ 30 kg/m2; no participants with diabetes (FPG < 126 mg/dL) | 18 months; n = 56 | Ad libitum low-glycemic-load diet emphasizing low-GI sources for carbohydrates (40% carbohydrates and 35% fat) compared with a low-fat diet (55% carbohydrates and 20% fat) | Groups separated by median insulin-30 values into low (≤57.5 μIU/mL) and high (>57.5 μIU/mL) | Participants with high insulin-30 values lost 4.6 kg more p = 0.004) on the low-glycemic-load diet than on the low-fat diet; no difference in diet response was observed among those with insulin-30 values below the median p = 0.90) |
| Gardner et al. (29) | Prespecified RCT; United States | BMI = 28–40 kg/m2 | 12 months; n = 609 | ~500-kcal deficit relative to baseline Reported consumption after 12 months Low-fat diet (C 48%: P 21%: F 29%) versus low-carbohydrate diet (C 30%: P 23%: F 45%) |
Insulin-30 value used as a continuous variable (no groups) | Insulin-30 values did not modify the effect of diet on weight loss p for interaction = 0.474) |
| Gardner et al. (28) | Prespecified RCT; United States | BMI = 28–40 kg/m2; no participants with diabetes | 6 months; n = 49 | Ad libitum low-fat versus low-carbohydrate diet using a limbo-titrate-quality approach | Insulin area under the curve from OGTT: insulin resistant (below median value); insulin sensitive (above median value); also insulin area under the curve was used as a continuous measure as well as insulin-30 values, insulin-120 values, and glucose area under the curve0–30 by insulin area under the curve0–30 | Insulin-sensitive participants lost 1.8 kg more on the low-fat diet than on the low-carbohydrate diet, whereas the insulin-resistant individuals lost 2.2 kg more on the low-carbohydrate diet than on the low-fat diet (Δ4.0 kg), but no weight loss was statistically significant; no meaningful differences were found when using insulin-30 values or other indexes (p > 0.05) |
| McClain et al. (50) (A TO Z study) | Secondary analysis of an RCT; United States | BMI = 27–40 kg/m2; women; no participants with diabetes | 1 year (2 months plus 10 months); n = 81 | Low-carbohydrate Atkins diet (goal to consume ≤50 g carbohydrates/day) versus low-fat Ornish diet (goal to consume ≤10% fat as total energy intake) | Groups separated by tertiles of FI values and middle tertile omitted: low (≤6.9 μIU/mL) and high (>10.6 μIU/mL) | Participants with high FI values lost a statistically insignificant 4 kg more on the low-carbohydrate diet versus the low-fat diet (p > 0.05); the difference among those with low FI values was 0.6 kg (p > 0.05) |
| Cornier et al. (15) | Prespecified RCT; United States | BMI = 30–35 kg/m2; women without diabetes | 16 weeks; n = 21 | −400 kcal/day High-carbohydrate–low-fat diet (C 60%: P 20%: F 20%) Low-carbohydrate–high-fat diet (C 40%: P 20%: F 40%) | Groups based on FI values were low (≤ 10 μU/mL) and high (>15 μU/mL) | Participants with low FI values lost 5.2 kg more on the low-fat diet (p < 0.01), and participants with high FI values lost 3.7 kg more on the low-carbohydrate diet (p < 0.05); 8.8-kg interaction; no p value |
| Rock et al. (62) | Prespecified RCT; United States | BMI = 33.5 (range, 27–40) kg/m2; nondiabetic women | 1 year; n = 214 | −500–1,000 kcal/day on one of three diets Low fat–high carbohydrate (C 65%: P 15%: F 20%); low carbohydrate–high fat (C 45%: P 20%: F 35%); low carbohydrate–high fat with 18% energy intake from walnuts (C 45%: P 20%: F 35%) |
Groups based on HOMA-IR values were insulin sensitive (HOMA-IR ≤ 3) and insulin resistant (HOMA-IR > 3) | Participants with low HOMA-IR lost 3.2 to 3.8 kg less weight (p = 0.04 to 0.06) on the low-carbohydrate diet compared with the low-fat and walnut-rich diets, respectively; no difference between diets among participants with high HOMA-IR (p > 0.05) |
| Hron et al. (37) | Run-in phase of an RCT; United States | BMI = 34.4 ± 4.9 kg/m2 (BMI ≥ 27); no participants with diabetes | 12 weeks; n = 21 | −40% calorie-restricted diet for 12 weeks containing C 45%: P 25%: F 30% with 27.1 g fiber/day | Linear associations using insulin-30 values, insulin secretion, hepatic insulin sensitivity and HOMA-IR | Participants with lower insulin-30 values lost more fat mass (p = 0.04), corresponding to a 1.7-kg difference between the 10th and 90th percentiles of insulin-30 values; no other measures of insulinemia were associated with weight change |
| Hjorth et al. (34) (SHOPUS) | Secondary analysis of an RCT; Denmark | 11% normal; 43% overweight; 46% obese | 26 weeks; n = 176 | Ad libitum New Nordic Diet: C 47% + 44 g fiber/10 MJ: P 18%: F 30% Average Danish Diet: C 46% + 28 g fiber/10 MJ: P 17%: F 34% |
Groups based on FI values with median among prediabetic individuals used as the split | Participants with low and high FI values lost 4.1 kg more (p < 0.001) and 1.6 kg more (p = 0.02), respectively, on the New Nordic Diet than on the Average Danish Diet (Δ2.5 kg; p = 0.006) |
| Hjorth et al. (34) (NUGENOB) | Secondary analysis of an RCT; pan-European | 5% overweight; 95% obese | 10 weeks; n = 743 | −600 kcal/day on either a low-fat–high-carbohydrate diet (C 58%: P 17%: F 25%) or a low-carbohydrate–high-fat diet (C 42%: P 17%: F 41%) | Groups based on FI values with median among prediabetic individuals used as the split | Participants with low and high FI values lost 0.4 kg more (p = 0.046) and 0.1 kg more (p = 0.84), respectively, on the low-fat–high-carbohydrate diet than on the high-fat–low-carbohydrate diet (Δ0.1 kg; p = 0.33) |
| Hjorth et al. (34) (DiOGenes) | Secondary analysis of an RCT; pan-European | 19% overweight; 81% obese | 26 weeks; n = 266 | Ad libitum High-glycemic-load diet: C 51% (GI = 61): P 17%: F 30% Low-glycemic-load diet: C 46% (GI = 56): P 21%: F 30% |
Groups based on FI values; median among prediabetic individuals used as the split | Participants with low and high FI values regained 2.3 kg more (p < 0.001) and 0.9 kg more (p = 0.02), respectively, on the high-glycemic-load diet than on the low-glycemic-load diet (Δ1.4 kg; p = 0.14) |
| Hjorth et al. (33) (MUFObes) | Secondary analysis of an RCT; Denmark | BMI = 31 (range, 29.3–33.0) kg/m2 | 6 months; n = 104 | Monounsaturated fatty acid diet: C 43% (42 g fiber/10 MJ): P 15%: F 38% Nordic Nutrition Recommended Diet: C 58% (40 g fiber/10 MJ): P 16%: F 24% Average Danish Diet: C 50% (29 g fiber/10 MJ): P 16%: F 32% |
Groups based on FI values; median among high-glucose group used as the split | Participants with high FI values regained 2.5 kg, 1.5 kg, and 4.2 kg on the monounsaturated fatty acid diet, Nordic Nutrition Recommended Diet, and Average Danish Diet, respectively (p ≥ 0.061 between diets), whereas weight regain was 2.5 kg, 2.1 kg, and 2.4 kg among participants with low FI values, on the same respective diets (p ≥ 0.63 between diets); no differences were found between diets for participants with low and high FI values (all p ≥ 0.16) |
| Hjorth et al. (36) (CHO) | Secondary analysis of an RCT; United States | Obese with BMI = 36.1 ± 3.5 kg/m2 | 24 months; n = 307 | Low-carbohydrate diet (20 g/day for 3 months with low-GI vegetables and unrestricted fat and protein; after 3 months carbohydrates increased by 5 g/day per week until a stable desired weight was achieved) Low-fat diet with limited energy intake (1,200 to 1,800 kcal/day; C 55%: P 15%: F 30%) |
Groups based on FI values; median value used as the split | Participants with low and high FI values lost 0.4 kg more (p = 0.79) and 3.4 kg more (p = 0.012), respectively, on the low-fat diet than on the low-carbohydrate diet (Δ3.7 kg; p = 0.052) |
| Shyam et al. (66) and Ghani et al. (30) | Secondary analysis of an RCT; Malaysia | BMI = 26.4 ± 4.6 kg/m2; women with a history of gestational diabetes | 6 months; n = 77 | −500 kcal/day (women with BMI < 23 or breastfeeding a child aged <6 months were prescribed energy requirements); prescribed intake capped at 1,800 kcal/day Conventional healthy dietary recommendations (diet low in fat and refined sugars and high in fiber) versus conventional healthy dietary recommendation plus low GI (low-glycemic-load diet) |
Groups based on FI values were low (<2 μlU/mL; n = 44) and high (≥2 μlU/mL; n = 33) | Participants with low and high FI values lost, respectively, a statistically insignificant 1.4 kg more (p = 0.41) and 3.3 kg more (p = 0.09) on the low-GI diet (Δ1.9-kg interaction not tested) |
| Kong et al. (41) | Unrandomized trial; France | Overweight and obese individuals | 12 weeks; n = 50 | 6-week hypocaloric diet (women consumed 1,200 kcal/day and men consumed 1,500 kcal/day) rich in fiber and with 35% protein, 25% fat and 40% low-GI carbohydrates; followed by a 6-week weight maintenance diet (individually prescribed by dietitian) | FI value, HOMA-IR value, and a number of other indexes | The more insulin-sensitive participants (based on FI and HOMA-IR values) lost the most weight p ≤ 0.02) |
| McLaughlin et al. (51) | Unrandomized trial; United States | BMI = 28.0–35.0 kg/m2; nondiabetic women | 60 days; n = 31 | −1,000 kcal/day (minimum 1,200 kcal/day) Commercial liquid nutrition formula plus two high-fiber muffins and sodium supplement daily |
Insulin-mediated glucose disposal estimated by insulin and glucose concentrations during the last 30 min of a 180-min infusion of somatostatin, insulin, and glucose; total integrated insulin response across two meals | Weight loss did not vary as a function of insulin resistance or total integrated insulin response (p ≥ 0.83) |
Abbreviations: BMI, body mass index; FI, fasting insulin; FPG, fasting plasma glucose; GI, glycemic index; HOMA-IR, homeostatic model assessment-insulin resistance; OGTT, oral glucose tolerance test; RCT, randomized controlled trial.
a
BMI values are summarized as mean ± standard deviation, median (interquartile range), or as percentage in different weight categories. C:P:F refers to the energy percentages of carbohydrate, protein, and fat in the diet.