Table 4.
Results of the 33 prospective studies included in the meta-analysis.
| Reference, Year | Comparison | Model | RR 95% CI | Adjustment for covariates |
|---|---|---|---|---|
| Lindsay et al.57 | Per SD increment | Multivariable | 0.59 0.38, 0.91 | Age, waist circumference, fasting and 2-h glucose, fasting insulin. Matched for BMI, age and sex |
| Daimon et al.58 | Highest tertile (median, 13.9 μg/mL) vs lowest (4.7 μg/mL) | Multivariable | 0.11 0.01, 0.96 | Age, sex, waist-hip ratio, 2-h glucose, TNF-1 |
| Snehalatha et al.59 | Per 1 μg/mL increment | Multivariable | 0.87 0.79, 0.95 | HbA1c |
| Choi et al.60 | Highest tertile (median, 23.8 μg/mL) vs lowest (7.9 μg/mL) | Multivariable | 0.31 0.14, 0.71 | BMI |
| Kanaya et al.61 | per 1 log μg/mL | Multivariable | 1.04 0.69, 1.56 | Age, sex, race, BMI, visceral fat, hypertension, leptin, PAI-1, fasting glucose and insulin, HDL-C, TG |
| Koenig et al.38 | Highest tertile (median, 10.6 μg/mL) vs lowest (3.8 μg/mL) | Multivariable | 0.55 0.35, 0.89 | Age, BMI, smoking, alcohol intake, physical activity, hypertension, history of myocardial infarction |
| Model 2 | 0.81 0.50, 1.33 | Additional adjustment for HDL-C | ||
| Nakashima et al.62 | Highest tertile (median, 17.4 μg/mL) vs lowest (5.4 μg/mL) | Multivariable | 0.56 0.32, 0.99 | Age, sex, BMI, waist-hip ratio, HOMA-IR, glucose tolerance classification |
| Snijder et al.28 | Highest quartile (median, 28.4 μg/mL in men, 24.8 μg/mL in women) vs lowest (8.1 μg/mL in men, 8.5 μg/mL in women) | Multivariable | 0.43 0.19, 0.94 | Age, waist-hip ratio, smoking, performance of sports, leptin |
| Model 2 | 0.72 0.36, 1.42 | Additional adjustment for fasting and 2-h glucose | ||
| Model 3 | 0.75 0.38, 1.51 | Additional adjustment for TG | ||
| Wannamethee et al.32 | Highest tertile (median, 13.7 μg/mL) vs lowest (3.6 μg/mL) | Multivariable | 0.40 0.23, 0.70 | Age, BMI, social class, physical activity, smoking, alcohol intake, history of coronary heart disease or stroke, statin use, blood pressure, treatment for hypertension |
| Model 2 | 0.59 0.33, 1.04 | Additional adjustment for HOMA-IR | ||
| Model 3 | 0.67 0.38, 1.20 | Additional adjustment for HDL-C and CRP | ||
| Heidermann et al.27 | Highest quintile (median, 28.4 μg/mL) vs lowest (8.1 μg/mL) | Multivariable | 0.17 0.12, 0.25 | Age, BMI, ethnicity, physical activity, smoking, family history of diabetes, hormone therapy, alcohol intake, dietary factors. Matched for age at blood draw ( ± 1 year), date of blood draw ( ± 3 months), fasting status, race |
| Model 2 | 0.16 0.10, 0.27 | Additional adjustment for hyperlipidemia, hypertension and CRP | ||
| Model 3 | 0.26 0.13, 0.51 | Additional adjustment for fasting insulin | ||
| Ley et al.63 | Per SD increment | Multivariable | 0.68 0.51, 0.90 | age, sex, waist circumference, TG, HDL-C, hypertension, IGT |
| Mather et al.31 | Per SD increment | Multivariate | 0.77 0.66, 0.89 | age, sex, race/ethnicity |
| Model 2 | 0.84 0.71, 0.98 | Additional adjustment for change in weight, change in adiponectin and baseline and change in insulinogenic index, 1/fasting insulin | ||
| Tábak et al.64 | Per 1 μg/mL increment | Multivariable | 0.87 0.77, 0.97 | Age, sex, BMI, physical activity, family history of diabetes, employment grade. Matched for matched on sex, age (5-year groups), and BMI (5 kg/m2 groups) |
| Model 2 | 0.89 0.79, 0.99 | Age, sex, BMI, TC, TG, blood pressure | ||
| Model 3 | 0.89 0.79, 1.00 | Additional adjustment for CRP | ||
| Model 4 | 0.94 0.82, 1.07 | Additional adjustment for fasting glucose | ||
| Salomaa et al.6 | Per SD increment | Multivariable | 0.67 0.61, 0.80 | Sex, non-HDL-C, HDL-C, TG, BMI, systolic blood pressure, current smoking, blood glucose, history of cardiovascular disease event, use of antihypertensive medication |
| Per SD increment | Multivariable | 0.70 0.60, 0.90 | Sex, non-HDL-C, HDL-C, TG, BMI, systolic blood pressure, current smoking, blood glucose, history of cardiovascular disease event, use of antihypertensive medication | |
| Thorand et al.7 | Highest tertile (median 13.3 μg/mL in men, 18.1 μg/mL in women) vs. lowest tertile (6.7 μg/mL in men, 9.9 μg/mL in women) | Multivariable | 0.28 0.20, 0.39 | Age, sex, survey, BMI, smoking, alcohol consumption, physical activity |
| Model 2 | 0.38 0.27, 0.53 | Additional adjustment for systolic blood pressure, TC/HDL-C, parental history of diabetes mellitus, CRP, interleukin-6, soluble ICAM-1 and soluble E-selection, and leptin | ||
| Zhu et al.8 | Highest quartile (weighted median: 10.61 μg/mL) vs. lowest (weighted median: 3.48 μg/mL) | Multivariable | 0.40 0.25, 0.64 | Age, sex, ethnicity, center, hypertension, and parental history of diabetes, BMI, waist-to-hip ratio |
| Model 2 | 0.46 0.29, 0.74 | Additional adjustment for inflammation score | ||
| Model 3 | 0.52 0.32, 0.85 | Additional adjustment for fasting insulin | ||
| Model 4 | 0.82 0.48, 1.42 | Additional adjustment for fasting glucose | ||
| Fagerberg et al.9 | Highest tertile (18.28–40.78 μg/mL) vs lowest tertile (3.68–11.36 μg/mL) | Multivariable | 0.22 0.07, 0.69 | HOMA-IR, AIR, smoking, IFG, IGT |
| Hanley et al.10 | per SD increment | Multivariable | 0.67 0.46, 0.97 | Age, sex, ethnicity, smoking, BMI |
| Model 2 | 0.64 0.43, 0.94 | Age, sex, ethnicity, smoking, HDL-C | ||
| Model 3 | 0.69 0.49, 0.99 | Age, sex, ethnicity, smoking, HOMA-IR | ||
| Model 4 | 0.81 0.56, 1.16 | Age, sex, ethnicity, smoking, S1 | ||
| Model 5 | 0.75 0.53, 1.06 | Age, sex, ethnicity, smoking, IFG | ||
| Hivert et al.11 | per SD increment | Multivariable | 0.57 0.40, 0.81 | Age, sex, BMI |
| Model 2 | 0.58 0.41, 0.84 | Additional adjustment for HOMA-IR | ||
| per SD increment | Multivariable | 0.53 0.39, 0.74 | Age, sex, BMI | |
| Model 2 | 0.69 0.51, 0.96 | Additional adjustment for HOMA-IR, resistin, TNF-alpha | ||
| Montonen et al.12 | Highest quintile (median 9.7 μg/mL in men, 14.4 μg/mL in women) vs. lowest quintile (3.07 μg/mL in men, 4.74 μg/mL in women) | Multivariable | 0.18 0.12, 0.28 | Age, sex, education, sport activity, cycling, occupational activity, smoking, alcohol intake, consumptions of red meat, whole grain bread and coffee, BMI, waist-circumference |
| Model 2 | 0.26 0.16, 0.40 | Additional adjustment for GGT, HDL-C, hs-CRP | ||
| Model 3 | 0.28 0.17, 0.44 | Additional adjustment for HbA1c | ||
| Kizer et al.13 | Highest quartile (median 23.6 μg/mL) vs. lowest (median 7.2 μg/mL) | Multivariable | 0.41 0.28, 0.61 | Age, sex, race, income, smoking, alcohol, eGFR, prevalent congestive heart failure, prevalent atrial fibrillation, prevalent CHD, beta-blocker use, health status, BMI |
| Model 2 | 0.79 0.50, 1.23 | Additional adjustment for systolic blood pressure, HDL-C, TG, CRP, HOMA-IR | ||
| Li et al.14 | Highest quintile (median 13.9 μg/mL) vs. lowest quintile (median 4.3 μg/mL) | Multivariable | 0.72 0.42, 1.25 | Age, sex, smoking, physical activity, alcohol consumption, family history of diabetes, BMI |
| Model 2 | 0.85 0.48, 1.49 | Additional adjustment for CRP, fasting blood glucose, insulin | ||
| Lilja et al.15 | Highest quartiles (≥12.1 μg/mL in men, ≥18.4 μg/mL in women) vs. lowest quartile (≤6.2 μg/mL in men, ≤9.2 μg/mL in women) | Multivariable | 0.40 0.30, 0.54 | BMI |
| Model 2 | 0.49 0.35, 0.70 | Additional adjustment for TC, hypertension, smoking, physical activity, university education, first-degree diabetes heredity, fasting and postload glucose | ||
| Model 3 | 0.55 0.38, 0.78 | Additional adjustment for HOMA-IR | ||
| Marques-Vidal et al.16 | Highest quartile (mean 17.3 μg/mL) vs. lowest quartile (mean 3.7 μg/mL) | Multivariable | 0.41 0.26, 0.65 | Age, gender, BMI |
| Model 2 | 0.64 0.40, 1.03 | Additional adjustment for diabetes risk score (age, family history of type 2 diabetes, height, waist circumference, resting heart rate, presence of hypertension, HDL-C, TG, fasting glucose and serum uric acid) | ||
| Kim et al.17 | Highest tertil (≥6.23 μg/mL in men, ≥9.47 μg/mL in women) vs. lowest tertile (<3.90 μg/mL in men, <6.01 μg/mL in women) | Multivariable | 0.67 0.46, 0.96 | Age, sex, BMI, waist circumferences |
| Model 2 | 0.67 0.45, 1.00 | Additional adjustment for fasting serum glucose | ||
| Rubio-Martin et al.18 | Highest tertile (>13.2 μg/mL) vs. lowest tertile (<6.6 μg/mL) | Multivariable | 0.24 0.07, 0.82 | Age, sex, obesity, CRP |
| Sans et al.19 | Per 1 log increase | Multivariable | 0.22 0.08, 0.61 | Age, BMI, leptin |
| Model 2 | 0.25 0.09, 0.70 | Age, BMI, insulin, years of school | ||
| Model 3 | 0.24 0.08, 0.72 | Age, BMI, leptin, years of school, DBP, HDL-C, TG | ||
| Model 4 | 0.47 0.16, 1.40 | Age, BMI, leptin, fasting glucose, years of school, DBP, HDL-C, TG | ||
| Julia et al.20 | Highest tertile vs. lowest tertile | Multivariable | 0.56 0.27, 1.18 | Age, sex, supplementation group, family history of diabetes and BMI. Matched for sex, age, BMI and initial supplementation group. |
| Model 2 | 0.71 0.33, 1.53 | Additional adjustment for baseline glycaemia, TC and TG | ||
| Lindberg et al.21 | Highest quartile (>10.35 μg/mL) vs. lowest quartile (≤5.13 μg/mL) | Multivariable | 0.16 0.04, 0.66 | Age and sex |
| Model 2 | 0.17 0.04, 0.75 | Additional adjustment for hypertension, hypercholesterolemia, current smoking, previous MI, BMI, blood glucose | ||
| Model 3 | 0.15 0.02, 0.90 | Additional adjustment for TC, HDL-C, LDL-C, TG | ||
| Yamamoto et al.22 | Highest quartile (≥9.6 μg/mL) vs. lowest (<5.2 μg/mL) | Multivariable | 0.40 0.25, 0.64 | Age, sex, family history, smoking, alcohol drinking, physical activity, BMI |
| Model 2 | 0.53 0.33, 0.86 | Additional adjustment for HOMA-IR | ||
| Model 3 | 0.56 0.35, 0.91 | Age, sex, family history, smoking, alcohol drinking, physical activity, BMI, HbA1c | ||
| Model 4 | 0.69 0.42, 1.13 | Additional adjustment for HOMA-IR | ||
| Neville et al.23 | Highest tertile (>6.66 μg/mL) vs. lowest (<3.77 μg/mL) | Multivariable | 0.29 0.17, 0.52 | Age, BMI, waist/hip ratio, alcohol status, smoking status, measures of socioeconomic status (includes material conditions and deprivation score), physical activity |
| Model 2 | 0.37 0.20, 0.67 | Additional adjustment for TC, HDL-C, TG, systolic blood pressure, on drug treatment for hypertension, CRP | ||
| Model 3 | 0.54 0.29, 0.99 | Additional adjustment for HOMA-IR |
Abbreviations: OR, odds ratio; CI, confidence interval; TNF-1, the tumour necrosis factor 1; PAI-1, plasminogen activator inhibitor 1; TC, total cholesterol; LDL-C, LDL cholesterol; HDL-C, HDL cholesterol; TG, triglycerides; TG/HDL-C, the ratio of TG to HDL-C; CRP, C-reactive protein; hs-CRP, high-sensitivity CRP; IGT, impaired glucose tolerance; IFG, impaired fasting glucose; GGT, gamma-glutamyltransferase; ICAM-1, intercellular adhesion molecule 1; S1, insulin sensitivity index; eGFR, the epidermal growth factor receptor; CHD, coronary heart disease; DBP, diastolic blood pressure; MI, myocardial infarction.