Abstract
Purpose
Low-carbohydrate diets (LCD) are a popular dietary strategy for weight reduction. The effects of LCD on long-term outcome vary depending on type of LCD, possibly due to the fact that effects on cardiometabolic risk factors may vary with different types of LCD. Accordingly, we studied these relations.
Methods
We assessed serum concentrations of high-density lipoprotein cholesterol (HDLc), low-density lipoprotein cholesterol (LDLc), high-sensitivity C-reactive protein (CRP), total cholesterol, glycated hemoglobin, and uric acid, and nutrient intakes by standardized methods in men and women ages 40–59 years from four population samples of Japanese in Japan (553 men and 544 women, combined). For people consuming usual, animal-based, and plant-based LCDs, we calculated LCD scores, based on relative level of fat, protein, and carbohydrate, by modifying the methods of Halton et al. Instead of calculating scores based on animal or vegetable fat, we used saturated fatty acids (SFA) or monounsaturated fatty acids (MUFA) + polyunsaturated fatty acids (PUFA).
Results
In multivariate regression analyses with adjustment for site, age, sex, BMI, smoking, alcohol intake, physical activity, and years of education, all three LCD scores were significantly positively related to HDLc (all P < 0.001), but not to LDLc. The plant-based LCD score was significantly inversely related to log CRP (coefficient = −0.010, P = 0.018).
Conclusions
All three LCD scores were significantly positively related to HDLc. The plant-based LCD score was significantly inversely related to CRP. Carbohydrate intake below 50 % of total energy with higher intakes of vegetable protein and MUFA + PUFA, and lower intakes of SFA may be favorable for reducing cardiometabolic risk factors.
Keywords: Low-carbohydrate diet, C-reactive protein, Cardiometabolic risk factors, High-density lipoprotein cholesterol
Introduction
Low-carbohydrate diets (LCD) are a popular dietary strategy for weight reduction [1]. The term LCD is generally applied to diets that restrict carbohydrates to <20 % of caloric intake [1], but can also refer to diets that simply restrict or limit carbohydrates to less than recommended proportions (generally <45 % of total energy coming from carbohydrates). Effectiveness for weight loss and reduction in cardiometabolic risk factors (CMRF) [including obesity (particularly central), hyperglycemia, hypertension, insulin resistance, dyslipoproteinemia] of LCD has been reported in several review and meta-analysis studies [2–5]. However, Sacks et al. [6] reported that reduced-calorie diets result in clinically meaningful weight loss regardless of the macronutrients they emphasize. Although dropout in this study was low (20 %), the dietary goals were only partially achieved as judged from objective biomarkers [6]. LCDs often involve high intake of saturated fatty acids (SFA) or diet patterns with adverse lipid composition, leading to increase in low-density lipoprotein cholesterol (LDLc). It has been suggested that LCDs may be associated with increased risk of cardiovascular disease (CVD) morbidity and mortality [7–11]. Recently, a meta-analysis of observational studies on the association of LCD with total mortality has been carried out by Noto et al. [12]. This study showed LCD to be associated with significantly higher risk of total mortality without significant differences in CVD mortality and incidence. Although differential effects of animal-based and plant-based LCDs on health outcomes have been noted [7, 11], actual data are sparse on the effects on CMRF with different types of LCD.
In observational studies, dichotomizing participants into those who are on and those who are not on an LCD is difficult. Halton et al. [11] devised a system to classify women in the Nurses’ Health Study according to their relative levels of fat, protein, and carbohydrate intake and created a summary score designated the LCD score. One type of LCD score is based on dietary patterns involving varied mixes of both vegetable and animal products; we call this type the usual LCD score. Halton et al. calculated two other types of LCD score based on animal or vegetable fat. With use of INTERLIPID/INTERMAP data collected from 1997 to 1999 [13, 14], we assessed relationships of usual, animal-based, and plant-based LCD score to CMRF including serum concentrations of total cholesterol (TC), high-density lipoprotein cholesterol (HDLc), LDLc, uric acid (UA), and an inflammatory marker, high-sensitivity C-reactive protein (CRP), in population-based samples of Japanese residing in Japan. In contrast to Halton et al., we here used SFA, or monounsaturated fatty acids (MUFA) + polyunsaturated fatty acids (PUFA), given available data on harmful effects of SFA and beneficial effects of MUFA + PUFA on serum lipids and health. We designate these other two LCD scores as animal-based LCD score and plant-based LCD score. In these three LCD scores, usual, animal-based, and plant-based, higher scores indicate lower carbohydrate intake.
Methods
Participants
Detailed methods of the INTERMAP Study have been described [13, 14]. INTERLIPID participants aged 40–59 years were five INTERMAP population samples: four in Japan and one in Hawaii [15–17]; for the present study, the four samples in Japan were used. They were: (1) Japanese factory workers in Toyama, central Japan (149 men and 150 women); (2) Japanese factory workers in Sapporo, northern Japan (149 men and 148 women); (3) Japanese residents in Aito-town, a rural town in Shiga Prefecture, central Japan (129 men and 129 men); (4) Japanese factory workers in Wakayama, central Japan (145 men and 143 women). Among these four research sites, 32 persons (11 men and 21 women) were excluded because volume of their stored serum specimen was not enough to measure CRP and serum lipids; 13 persons (8 men and 5 women) were excluded because their CRP concentrations were more than 10 mg/L, leaving 1097 individuals (553 men and 544 women). Ethics committees of the Shiga University of Medical Science, the Pacific Health Research Institute, and Northwestern University approved the study protocol. Written informed consent was obtained from all participants.
Anthropometric and lifestyle assessment
Participants visited the research centers four times on two pairs of consecutive days on average 3 weeks apart. Height and weight with light clothes were measured at each visit; the four measurements of height and weight were averaged. Two standardized blood pressure (BP) measurements were made on each of the four different days; these eight measurements were averaged. Using a questionnaire, trained observers inquired about physical activity, smoking status, previous medical history of CVDs/diabetes, and use of medication (including anti-hypertensive medication). Body mass index (BMI) was calculated as weight divided by height squared (kg/m2). To evaluate physical activity, questions were posed about number of hours per day spent in heavy activity, moderate activity, light activity, watching TV, other sedentary and no activity (sleeping); the interviewer ensured that the total time added up to 24 h. A physical activity index score was calculated by multiplying the times spent in different activities by corresponding weighting factors that parallel the increased rate of oxygen consumption associated with increasingly more intense physical activity; for this, the procedure in the Framingham Offspring Study [18] was followed.
Dietary assessment
Four in-depth multi-pass 24-h dietary recalls per participant were conducted during the four visits by specially trained and certified dietary interviewers. All participants attended all four study visits; their energy intakes from all 24-h dietary recalls were between 500 and 5000 kcal/day. Validation of the methods has been reported [14]. For each person, means of individual nutrients from the four 24-h dietary recalls were used in the analyses. Data are presented as the contribution to total energy intake (percentage of kcal [% kcal]) from total carbohydrates, total protein, animal protein, plant protein, total fats, SFA, MUFA, PUFA, and alcohol (% kcal). Keys dietary lipid score, predictive of serum TC, was calculated as 1.35 × (2 × SFA – PUFA) + 1.5 × C1/2, where SFA is % kcal from SFA; PUFA, % kcal from PUFA; and C, dietary cholesterol in mg/1000 kcal [19].
Biochemical measurements
For the INTERLIPID Study, non-fasting blood was drawn ad libitum time after last meal on the second day of the first 2-day visit pair [15–17]. We used data on analytes measured in these blood samples, as well as data from INTERMAP. Serum and plasma were obtained by centrifugation within 30 min of blood drawing and immediately refrigerated. Within 24-h, all specimens were frozen and stored locally at −70 °C. Samples were sent to a central laboratory in Japan on dry ice. TC, HDLc, LDLc, triglycerides (TG), and UA were directly measured by enzymatic methods on an auto-analyzer (Hitachi 7107; Hitachi, Tokyo, Japan). TGs were analyzed for inclusion as a covariate in the models, not as an outcome variable in this study. Serum CRP was measured by immunoturbidimetric assay. Percentage of glycated hemoglobin A1c (HbA1c) was measured using the standard method of the Japan Diabetes Society.
Calculation of the LCD score
Usual LCD score was calculated by modifying the methods of Halton et al. [11]. Data are shown as percentage of total energy. Men and women considered together were analyzed into 11 strata for fat, protein, and carbohydrate intake. For fat and protein, participants in the highest stratum received 10 points for that macronutrient, down to participants in the lowest stratum, who received 0 points. For carbohydrate, the order of the strata was reversed. The points for each of the three macronutrients were then summed to create the overall diet score, which ranged from 0 to 30. Type of dietary lipid influences serum lipids; previous studies also showed differential effects of animal-based and plant-based LCD on health outcomes [7, 11]. Accordingly, we also generated two additional LCD scores. One was calculated according to the percentage of energy as carbohydrate, animal protein (from egg, meat, chicken, other poultry, seafood, and dairy products), and SFA. Although some SFA, like myristic and palmitic acids, have been reported as more harmful than the others [20], we did not subdivide SFA in this study. The other LCD score was calculated according to the percentage of energy as carbohydrate, vegetable protein (from soy, other legumes, nuts, grains, and other non-animal sources), and MUFA plus PUFA. The former is called animal-based LCD score; the latter, plant-based LCD score.
Data analyses
SAS version 9.4 for Windows (SAS Institute, Cary, NC) was used. As the distribution of CRP values was positively skewed, logarithmic transformation was used to normalize them. We divided participants into quintiles as well as deciles according to their LCD score. The Mantel–Haenszel Chi-square statistical test for nominal variables and the “contrast” option for analysis of variance for continuous variables were used to assess whether there were significant differences in variables across quintiles of LCD score. Trend P values were obtained. Multiple linear regression analysis with adjustment for confounders was used to examine relationships of deciles of LCD score to CRP (log CRP), HDLc, LDLc, TC, UA, and HbA1c. Model 1 included site (site 1–4; site 4 as reference), age, sex, BMI, and decile of LCD score. Model 2 included model 1 covariates + cigarettes/day, alcohol (% kcal), physical activity index, and years of education (≤9, 10–12, 13–15, ≥16 year; ≤9 year as reference). In sensitivity analyses, TG was entered into model 2 analyses for relation of LCD scores to LDLc, because TG may influence the relation even though LDLc was measured by the direct method. Interaction P values for sex-LCD were not significant; hence, all results are presented for both sexes combined. Finally, for guidance in identifying a cut point for designating a diet as LCD, we scrutinized successive plant-based LCD scores in descending order to find where significant differences occurred in CRP or HDLc using age- and sex-adjusted analysis of variance. All P values were two tailed; P < 0.05 was considered significant.
Results
Descriptive statistics
Data on the criteria for determining the three LCD scores are shown in Table 1. On average, about 54 % of energy intake was from carbohydrate, 16 % from protein, 25 % from fat, and 4.8 % from alcohol. For participants in the lowest carbohydrate stratum, carbohydrate intake ranged from 29.3 to 44.1 (mean 40.2) % kcal; in the highest stratum, from 63.7 to 78.6 (mean 66.6) % kcal.
Table 1.
Category limits (% kcal) for nutrients used in determining the LCD score INTERMAP/INTERLIPID Study, Japan
| Points | Carbohydrate | Total protein | Total fat | Animal protein | SFA | Veg protein | MUFA + PUFA |
|---|---|---|---|---|---|---|---|
| 0 | 63.7–78.6 | 10.0–13.1 | 9.3–18.5 | 2.6–5.9 | 1.8–4.4 | 4.0–5.6 | 5.4–11.3 |
| 1 | 60.6–63.6 | 13.2–14.0 | 18.6–20.3 | 6.0–6.8 | 4.5–5.0 | 5.7–6.0 | 11.4–12.5 |
| 2 | 58.8–60.5 | 14.1–14.6 | 20.4–22.0 | 6.9–7.4 | 5.1–5.5 | 6.1–6.4 | 12.6–13.5 |
| 3 | 57.3–58.7 | 14.7–15.0 | 22.1–23.3 | 7.5–7.9 | 5.6–5.9 | 6.5–6.7 | 13.6–14.3 |
| 4 | 55.5–57.2 | 15.1–15.6 | 23.4–24.4 | 8.0–8.4 | 6.0–6.3 | 6.8–6.9 | 14.4–15.0 |
| 5 | 53.9–55.4 | 15.7–16.0 | 24.5–25.5 | 8.5–8.9 | 6.4–6.7 | 7.0–7.2 | 15.1–15.6 |
| 6 | 52.3–53.8 | 16.1–16.6 | 25.6–26.5 | 9.0–9.5 | 6.8–7.1 | 7.3–7.4 | 15.7–16.4 |
| 7 | 50.1–52.2 | 16.7–17.1 | 26.6–27.7 | 9.6–10.2 | 7.2–7.5 | 7.5–7.7 | 16.5–17.3 |
| 8 | 47.3–50.0 | 17.2–17.9 | 27.8–29.3 | 10.3–10.9 | 7.6–8.0 | 7.8–8.1 | 17.4–18.3 |
| 9 | 44.2–47.2 | 18.0–19.1 | 29.4–31.6 | 11.0–12.1 | 8.1–9.0 | 8.2–8.6 | 18.4–19.7 |
| 10 | 29.3–44.1 | 19.2–26.2 | 31.7–42.6 | 12.2–18.2 | 9.1–12.9 | 8.7–11.5 | 19.8–26.7 |
LCD score was calculated by modifying the methods of Halton et al. [12]. Data are shown as a percentage of energy. Number of people in each cell is either 99 or 100. We divided the study participants into 11 strata of fat, of protein, and of carbohydrate intake, expressed as a percentage of energy. For fat and protein, participants in the highest stratum received 10 points for that macronutrient, down to participants in the lowest stratum, who received 0 points. For carbohydrate, the order of the strata was reversed. The points for each of the three macronutrients were then summed to create the overall diet score, which ranged from 0 to 30 (usual LCD score). We also created two additional LCD scores by modifying usual LCD score calculation. One was calculated according to the percentage of energy as carbohydrate, the percentage of energy as animal protein, and the percentage of energy as SFA (including plant SFA) (animal-based LCD score), and the other according to the percentage of energy as carbohydrate, the percentage of energy as vegetable protein, and the percentage of energy as MUFA plus PUFA (plant-based LCD score)
LCD low-carbohydrate diet, SFA saturated fatty acids, Veg protein vegetable protein, MUFA mono-unsaturated fatty acids, PUFA poly-unsaturated fatty acids
Characteristics of participants by quintile of the three types of LCD score are shown in Table 2. Mean age was lower in higher LCD score groups of the three types (trend Ps = 0.007–0.040). Percentage of men was lower in higher plant-based LCD score groups (trend P = 010), and mean BMI was greater in higher usual LCD score groups (trend P = 0.039). Mean HDLc was greater with higher values of all three LCD score groups (trend Ps < 0.001); increments from quintile 1 to quintile 5 were similar in all three groups (about +11 %). Median CRP values were lower in higher usual and plant-based LCD score groups (trend Ps = 0.044 and 0.003), slightly more so in plant-based LCD score groups (−15.2 % in usual vs −17.7 % in plant-based LCD score groups). Mean cigarettes smoked were lower in higher usual and plant-based LCD score groups (trend Ps = 0.039 and <0.001). Mean physical activity score was lower with higher LCD score for all three LCD score groups (trend Ps <0.001), and percentage of participants with ≥16 years of education was greater with higher LCD score for all three LCD score groups (trend Ps = 0.007 to <0.001). Other variables were not different across the groups.
Table 2.
Characteristics of participants by quintile of usual, animal-based, and plant-based LCD score-INTERMAP/INTERLIPID Study, Japan
| Variable | Usual LCD |
Animal-based LCD |
||||||
|---|---|---|---|---|---|---|---|---|
| Quintile 1 | Quintile 3 | Quintile 5 | Trend P | Quintile 1 | Quintile 3 | Quintile 5 | Trend P | |
| LCD score | 0–8 | 14–17 | 22–30 | 0–8 | 14–17 | 22–30 | ||
| N (1097) | 219 | 242 | 216 | 222 | 229 | 207 | ||
| Age (year) | 50.1 ± 5.7 | 49.4 ± 5.0 | 49.1 ± 5.0 | 0.040 | 50.4 ± 5.6 | 49.8 ± 5.0 | 48.6 ± 4.9 | 0.007 |
| Men (%) | 52.5 % | 46.3 % | 49.5 % | 0.702 | 47.8 % | 45.9 % | 49.3 % | 0.501 |
| BMI (kg/m2) | 23.3 ± 2.7 | 23.5 ± 2.9 | 23.7 ± 3.0 | 0.039 | 23.4 ± 2.6 | 23.3 ± 2.8 | 23.8 ± 3.0 | 0.054 |
| LDLc (mg/dL) | 121.1 ± 30.7 | 124.3 ± 29.9 | 117.7 ± 30.0 | 0.052 | 122.8 ± 31.1 | 123.1 ± 29.8 | 119.6 ± 29.4 | 0.067 |
| HDLc (mg/dL) | 52.5 ± 12.9 | 58.0 ± 14.8 | 58.1 ± 14.9 | <0.001 | 52.4 ± 12.8 | 58.1 ± 14.4 | 58.3 ± 15.2 | <0.001 |
| TC (mg/dL) | 196.3 ± 32.3 | 203.3 ± 31.1 | 200.4 ± 30.5 | 0.164 | 196.7 ± 33.4 | 201.3 ± 32.1 | 202.0 ± 30.4 | 0.110 |
| UA (mg/dL) | 5.0 ± 1.3 | 4.9 ± 1.4 | 5.1 ± 1.4 | 0.224 | 5.0 ± 1.4 | 4.9 ± 1.4 | 5.1 ± 1.3 | 0.120 |
| CRP# (mg/L) | 0.33 | 0.30 | 0.28 | 0.044 | 0.32 | 0.31 | 0.29 | 0.248 |
| (0.19, 0.73) | (0.16, 0.56) | (0.15, 0.57) | (0.16, 0.71) | (0.16, 0.58) | (0.15, 0.64) | |||
| HbA1c (%) | 4.6 ± 0.7 | 4.7 ± 0.6 | 4.7 ± 0.7 | 0.350 | 4.6 ± 0.6 | 4.6 ± 0.6 | 4.7 ± 0.7 | 0.076 |
| Smoking (cigarettes/day) | 7.6 ± 12.0 | 5.3 ± 10.4 | 5.3 ± 10.0 | 0.039 | 6.3 ± 11.4 | 5.1 ± 9.9 | 6.0 ± 11.1 | 0.434 |
| PA score | 33.7 ± 8.6 | 32.0 ± 7.6 | 30.4 ± 5.5 | <0.001 | 33.6 ± 8.2 | 32.1 ± 7.3 | 31.0 ± 7.1 | <0.001 |
| Edu ≥ 16 year | 5.5 % | 8.3 % | 15.3 % | <0.001 | 5.9 % | 3.9 % | 15.9 % | <0.001 |
| Plant-based LCD |
||||||||
| Quintile 1 | Quintile 3 | Quintile 5 | Trend P | |||||
| LCD score | 0–10 | 14–16 | 20–29 | |||||
| N (1097) | 241 | 218 | 239 | |||||
| Age (year) | 49.7 ± 5.6 | 49.1 ± 5.1 | 48.9 ± 5.3 | 0.022 | ||||
| Men (%) | 57.7 % | 52.3 % | 45.6 % | 0.010 | ||||
| BMI (kg/m2) | 23.3 ± 2.9 | 23.5 ± 2.9 | 23.5 ± 2.9 | 0.457 | ||||
| LDLc (mg/dL) | 122.2 ± 30.3 | 122.5 ± 28.8 | 121.0 ± 31.3 | 0.055 | ||||
| HDLc (mg/dL) | 53.5 ± 13.6 | 56.4 ± 14.5 | 59.1 ± 15.0 | <0.001 | ||||
| TC (mg/dL) | 199.0 ± 31.9 | 200.8 ± 30.5 | 201.4 ± 32.8 | 0.649 | ||||
| UA (mg/dL) | 5.1 ± 1.3 | 5.1 ± 1.4 | 4.9 ± 1.4 | 0.090 | ||||
| CRP# (mg/L) | 0.34 | 0.32 | 0.28 | 0.003 | ||||
| (0.18, 0.77) | (0.15, 0.67) | (0.15, 0.54) | ||||||
| HbA1c (%) | 4.6 ± 0.5 | 4.6 ± 0.6 | 4.7 ± 0.7 | 0.651 | ||||
| Smoking (cigarettes/day) | 8.2 ± 12.2 | 6.6 ± 11.0 | 4.7 ± 9.8 | <0.001 | ||||
| PA score | 33.5 ± 8.7 | 32.5 ± 7.3 | 31.1 ± 5.9 | <0.001 | ||||
| Edu ≥ 16 year | 6.6 % | 8.7 % | 12.3 % | 0.007 | ||||
Values by quintile of three types of LCD score are mean ± SD, percent, or #median (25th and 75th percentile) (only quintile 1, 3, 5 values are shown). Trend P values are for the relationship between the variables listed on the left and quintile of the diet score, obtained by the Mantel–Haenzel Chi-square test or the “contrast” option for linear regression analysis, shown
Italic values indicate statistical significance at the 5 % level
LCD low-carbohydrate diet, BMI body mass index (kg/m2), LDLc serum low-density lipoprotein cholesterol concentration (mg/dL), HDLc serum high-density lipoprotein cholesterol concentration (mg/dL), TC total cholesterol concentration (mg/dL), UA uric acid (mg/dL), CRP C-reactive protein (mg/L), HbA1c hemoglobin A1c (%), Smoking smoking (cigarettes/d), PA score the Framingham Study physical activity index score, Edu education
Nutritional variables of participants by quintile of the three types of LCD score are shown in Table 3. Mean energy intake was greater in higher animal-based LCD score groups (trend P = 0.003). Alcohol intakes were greater in higher usual and animal-based LCD score groups (trend Ps < 0.001). Keys dietary lipid scores were greater with higher LCD scores in all three LCD groups (trend Ps < 0.001); the increment was greatest in animal-based LCD score groups (+54 %) and least in plant-based LCD score groups (+12 %). Means of all dietary lipid variables were significantly higher (trend Ps < 0.001), and means of dietary carbohydrate variables were lower with higher LCD score for all three LCD scores groups (trend Ps < 0.001). Although the findings were qualitatively similar, quantitatively there were differences in dietary lipids among the three groups: the increment of SFA was the most in animal-based LCD score groups (+71 % from quintile 1–5) and least in plant-based LCD score groups (+38 %); the increment of dietary cholesterol was the most in animal-based LCD score groups (+126 %) and least in plant-based LCD score groups (+19 %); the increment of PUFA was the least in animal-based LCD score groups (+19 %) and greatest in plant-based LCD score groups (+62 %).
Table 3.
Nutrient characteristics of participants by quintile of the three types of LCD score INTERMAP/INTERLIPID Study, Japan
| Variable | Usual LCD |
Animal-based LCD |
||||||
|---|---|---|---|---|---|---|---|---|
| Quintile 1 | Quintile 3 | Quintile 5 | Trend P | Quintile 1 | Quintile 3 | Quintile 5 | Trend P | |
| LCD score range | 0–8 | 14–17 | 22–30 | 0–8 | 14–17 | 22–30 | ||
| Energy (kcal) | 2012 ± 449 | 2047 ± 435 | 2043 ± 465 | 0.167 | 1969 ± 437 | 2028 ± 401 | 2085 ± 486 | 0.003 |
| Alcohol (% kcal) | 3.0 ± 4.3 | 5.4 ± 7.3 | 4.7 ± 5.8 | <0.001 | 2.5 ± 3.9 | 4.8 ± 6.6 | 5.5 ± 6.4 | <0.001 |
| Keys score | 25.2 ± 5.4 | 30.0 ± 5.2 | 34.8 ± 6.1 | <0.001 | 23.7 ± 4.5 | 29.7 ± 4.1 | 36.7 ± 5.5 | <0.001 |
| Prot (% kcal) | 14.0 ± 1.4 | 15.7 ± 1.9 | 18.2 ± 1.9 | <0.001 | 14.3 ± 1.6 | 16.0 ± 2.0 | 18.0 ± 2.1 | <0.001 |
| Fat (% kcal) | 19.9 ± 3.4 | 25.5 ± 3.7 | 29.9 ± 3.6 | <0.001 | 20.2 ± 3.6 | 25.3 ± 3.7 | 29.6 ± 4.0 | <0.001 |
| SFA (% kcal) | 5.2 ± 1.3 | 6.8 ± 1.6 | 8.03 ± 1.65 | <0.001 | 4.91 ± 1.02 | 6.65 ± 1.21 | 8.38 ± 1.56 | <0.001 |
| MFA (% kcal) | 7.0 ± 1.5 | 9.3 ± 1.8 | 11.0 ± 1.8 | <0.001 | 7.06 ± 1.52 | 9.13 ± 1.76 | 10.84 ± 1.89 | <0.001 |
| PUFA (% kcal) | 5.3 ± 1.2 | 6.5 ± 1.3 | 7.4 ± 1.4 | <0.001 | 5.80 ± 1.42 | 6.60 ± 1.39 | 6.88 ± 1.41 | <0.001 |
| n-3 PUFA (% kcal) | 1.1 ± 0.3 | 1.4 ± 0.3 | 1.6 ± 0.4 | <0.001 | 1.15 ± 0.32 | 1.34 ± 0.33 | 1.54 ± 0.41 | <0.001 |
| n-6 PUFA (% kcal) | 4.3 ± 1.1 | 5.1 ± 1.2 | 5.8 ± 1.3 | <0.001 | 4.63 ± 1.24 | 5.23 ± 1.23 | 5.31 ± 1.29 | <0.001 |
| Trans FA (% kcal) | 0.4 ± 0.3 | 0.5 ± 0.3 | 0.5 ± 0.3 | <0.001 | 0.31 ± 0.21 | 0.47 ± 0.31 | 0.53 ± 0.27 | <0.001 |
| Total carb (% kcal) | 63.1 ± 4.2 | 53.4 ± 5.1 | 47.1 ± 4.8 | <0.001 | 62.9 ± 4.1 | 53.8 ± 4.8 | 46.9 ± 5.1 | <0.001 |
| Sugar (% kcal) | 19.9 ± 5.9 | 19.0 ± 4.6 | 17.3 ± 3.5 | <0.001 | 19.4 ± 5.4 | 19.1 ± 4.5 | 17.7 ± 3.8 | <0.001 |
| Starch (% kcal) | 43.1 ± 6.6 | 34.4 ± 5.5 | 29.8 ± 5.1 | <0.001 | 43.5 ± 6.0 | 34.7 ± 5.2 | 29.2 ± 4.9 | <0.001 |
| Dietary chol (mg/1000 kcal) | 154.5 ± 55.3 | 190.9 ± 56.8 | 242.7 ± 66.6 | <0.001 | 153.1 ± 53.7 | 194.6 ± 59.2 | 246.5 ± 65.9 | <0.001 |
| Plant-based LCD |
||||||||
| Quintile 1 | Quintile 3 | Quintile 5 | Trend P | |||||
| LCD score range | 0–10 | 14–16 | 20–29 | |||||
| Energy (kcal) | 2066 ± 461 | 2114 ± 517 | 2028 ± 425 | 0.612 | ||||
| Alcohol (% kcal) | 4.5 ± 6.3 | 5.9 ± 7.4 | 3.8 ± 4.9 | 0.131 | ||||
| Keys score | 27.9 ± 6.1 | 30.4 ± 6.2 | 31.2 ± 6.5 | <0.001 | ||||
| Prot (% kcal) | 14.8 ± 2.1 | 15.9 ± 2.1 | 16.9 ± 2.2 | <0.001 | ||||
| Fat (% kcal) | 19.9 ± 3.7 | 25.0 ± 3.2 | 29.8 ± 3.6 | <0.001 | ||||
| SFA (% kcal) | 5.5 ± 1.6 | 6.7 ± 1.6 | 7.6 ± 1.6 | <0.001 | ||||
| MFA (% kcal) | 7.0 ± 1.5 | 9.0 ± 1.5 | 10.9 ± 1.8 | <0.001 | ||||
| PUFA (% kcal) | 4.9 ± 1.0 | 6.3 ± 0.9 | 8.0 ± 1.2 | <0.001 | ||||
| n-3 PUFA (% kcal) | 1.1 ± 0.3 | 1.4 ± 0.3 | 1.6 ± 0.4 | <0.001 | ||||
| n-6 PUFA (% kcal) | 3.8 ± 0.9 | 4.9 ± 0.9 | 6.3 ± 1.1 | <0.001 | ||||
| Trans FA (% kcal) | 0.4 ± 0.2 | 0.4 ± 0.3 | 0.5 ± 0.3 | <0.001 | ||||
| Total carb (% kcal) | 60.7 ± 6.4 | 53.1 ± 6.9 | 49.5 ± 4.7 | <0.001 | ||||
| Sugar (% kcal) | 20.4 ± 6.1 | 18.4 ± 4.6 | 17.2 ± 3.5 | <0.001 | ||||
| Starch (% kcal) | 40.3 ± 8.1 | 34.8 ± 7.0 | 32.3 ± 5.0 | <0.001 | ||||
| Dietary chol (mg/1000 kcal) | 177.5 ± 65.1 | 199.5 ± 62.5 | 210.4 ± 69.3 | <0.001 | ||||
Values by quintile of three types of LCD score are mean ± SD (only quintile 1, 3, 5 values are shown). All values are in % kcal, except for dietary cholesterol in mg/1000 kcal. Trend P values are for the relationship between the variables listed on the left and quintile of the diet score, obtained by the “contrast” option for linear regression analysis, shown
Italic values indicate statistical significance at the 5 % level
Energy dietary energy (kcal/d), Alcohol alcohol intake (% kcal), Total prot total protein (% kcal), SFA saturated fatty acids, MFA mono-unsaturated fatty acids, PUFA polyunsaturated fatty acids, Total carb total carbohydrate, Dietary chol dietary cholesterol
Food group characteristics of participants by quintile of the three types of LCD score are shown in Supplement Table 1. Mean intakes of rice and of fruits were significantly lower with higher LCD scores in all three LCD groups (trend Ps < 0.001); mean intakes of meats, eggs, and dairy products were greater with higher LCD scores in all three LCD groups (trend Ps < 0.001). Mean intakes of soy, other beans, and vegetables were greater in higher usual and plant-based LCD score groups (trend Ps < 0.001). Mean sweets intake was lower and mean fish intake was greater with higher usual and animal-based LCD score groups (trend Ps < 0.001).
Relations of LCD scores to CMRF
Relations of the three LCD scores to CMRF are shown in Table 4. All three LCD scores were significantly positively related to HDLc in both models (all Ps < 0.001). The plant-based LCD score alone was significantly inversely related to CRP in both models (coefficient = −0.013 to −0.010, P = 0.003–0.018). Animal-based LCD score was positively related to TC in model 1, but the relation was no longer statistically significant after adjustment for other covariates in model 2. The results of model 2 analyses including TG for relation of LCD scores to LDLc did not differ from the initial results (data not shown). The three LCD scores were not significantly related to the other CMRF.
Table 4.
Relations of three LCD scores to CMRF-INTERLIPID Study, Japan
| Outcome variables | Log CRP | HDLc | LDLc | TC | UA | HbA1c |
|---|---|---|---|---|---|---|
| Usual LCD score | ||||||
| Model 1 | ||||||
| β | −0.008 ± 0.005 | 0.905 ± 0.146 | −0.547 ± 0.310 | 0.511 ± 0.329 | 0.010 ± 0.011 | 0.004 ± 0.006 |
| P | 0.079 | <0.001 | 0.078 | 0.121 | 0.385 | 0.524 |
| Model 2 | ||||||
| β | −0.005 ± 0.005 | 0.731 ± 0.144 | −0.359 ± 0.315 | 0.496 ± 0.336 | 0.002 ± 0.012 | 0.004 ± 0.006 |
| P | 0.240 | <0.001 | 0.255 | 0.141 | 0.864 | 0.499 |
| Animal-based LCD score | ||||||
| Model 1 | ||||||
| β | −0.006 ± 0.005 | 0.940 ± 0.148 | −0.452 ± 0.316 | 0.669 ± 0.335 | 0.009 ± 0.012 | 0.007 ± 0.006 |
| P | 0.232 | <0.001 | 0.153 | 0.046 | 0.456 | 0.248 |
| Model 2 | ||||||
| β | −0.003 ± 0.005 | 0.702 ± 0.148 | −0.206 ± 0.324 | 0.634 ± 0.345 | −0.002 ± 0.012 | 0.007 ± 0.007 |
| P | 0.502 | <0.001 | 0.525 | 0.067 | 0.900 | 0.307 |
| Plant-based LCD score | ||||||
| Model 1 | ||||||
| β | −0.013 ± 0.004 | 0.678 ± 0.142 | −0.474 ± 0.300 | 0.031 ± 0.319 | −0.004 ± 0.011 | −0.002 ± 0.006 |
| P | 0.003 | <0.001 | 0.114 | 0.922 | 0.709 | 0.754 |
| Model 2 | ||||||
| β | −0.010 ± 0.004 | 0.648 ± 0.148 | −0.405 ± 0.303 | 0.066 ± 0.323 | −0.008 ± 0.011 | −0.00002 ± 0.006 |
| P | 0.018 | <0.001 | 0.181 | 0.838 | 0.483 | 0.997 |
Coefficients (β) ± standard errors, and P values from multiple linear regression models used to examine relations of deciles of the three LCD scores to CMRF are shown. Model 1 included site, age, sex, and BMI; Model 2, Model 1 variables + smoking (cigarettes/day), and alcohol intake (% kcal), the Framingham physical activity score, and years of education (≤9, 10–12, 13–15, ≥16 year; ≤9 year as reference)
Italic values indicate statistical significance at the 5 % level
LCD low-carbohydrate diet, CMRF cardiometabolic risk factors, BMI body mass index, LDLc serum low-density lipoprotein cholesterol concentration, HDLc serum high-density lipoprotein cholesterol concentration, TC serum total cholesterol concentration, UA serum uric acid concentration, CRP C-reactive protein, HbA1c hemoglobin A1c
Cutoff carbohydrate intake for designating LCD
Scrutinizing decile cut points of plant-based LCD score in descending order to identify where significant differences in CRP or HDLc occur, we found a significant increment in CRP at decile 6; significant decrement in HDLc at decile 8 (Supplement Table 2). Mean intakes of carbohydrate in decile 6 and decile 8 of plant-based LCD score were 50.4, and 49.5 %, respectively. Thus, <50 % may be suitable for designating a diet as LCD. From Table 1 results, about 300 participants (about 27 %) with carbohydrate intake in strata 8–10 fulfilled this criterion.
Discussion
Main findings here were, in four middle-aged population samples of Japanese in Japan combined, with median carbohydrate intake relatively high (54.6 % of total energy intake), three LCD scores (usual, animal-based, and plant-based) were significantly positively related to non-fasting HDLc. The plant-based LCD score was significantly inversely related to CRP. LCD scores were not related to other CMRF, LDLc, TC, HbA1c, or UA.
Previous studies have shown that LCD was positively related to HDLc [21–23]. No previous studies compared the relation of usual, animal-based, and plant-based LCD to HDLc. Results of our study indicate that all three types of LCD, usual, animal-based, plant-based, related positively to HDLc. The fact that all three types of LCD increase HDLc may be related to the fact that various types of LCD are associated with an increase in fatty acids intake. A meta-analysis by Mensink et al. [24] reported that intake of three classes of fatty acids, SFA, MUFA, and PUFA, increased HDLc relative to carbohydrates.
Although all three LCD scores were significantly positively related to non-fasting HDLc, only the plant-based LCD score significantly inversely related to CRP. Data are available from previous studies on the association of CRP with various dietary patterns. In cross-sectional studies, low concentration of CRP was associated with intake of a Mediterranean diet [25–27], nuts and whole-grain foods [28], and low glycemic index diets [29]. A recent meta-analysis by Schwingshackl et al. [30] of long-term intervention trials with low glycemic index diets showed a decrease in CRP. With regard to MUFA, a recent meta-analysis showed no statistically significant differences in CRP between high-MUFA and low-MUFA diets [31]. Previous meta-analyses of cohort studies reported inconsistent associations of MUFA with coronary heart disease (CHD) [32, 33]. These inconsistencies may be due to the fact that differences in sources of MUFA were not taken into account. Adopting a western diet means that MUFA is predominantly supplied by foods of animal origin, while in south European countries, olive oil is the dominant source of this type of fatty acid [34]. A so-called prudent dietary pattern, characterized by higher intakes of vegetables, legumes, fruits, fish, lean poultry, and whole grains, and lower intakes of red and processed meats, fried foods, refined grains, desserts, and other sweets, had an inverse association with plasma CRP. The Western dietary pattern, characterized by high intakes of red and processed meats, fried foods, refined grains, desserts, and sweets, showed a positive relation with CRP, significant after adjustment for age, BMI, physical activity, smoking status, and alcohol consumption [35–37]. The above evidence is concordant with our finding that the plant-based LCD score was significantly inversely related to CRP, but usual and animal-based LCD scores were not related CRP.
In our cross-sectional study, we observed no relation of LCDs to LDLc, and TC. Data available from longitudinal studies may help to understand our findings: Sacks et al. [6] showed with 65 % carbohydrate—20 % fat diet, LDLc decreased by 6 % from baseline in 2 years; with 35 % carbohydrate—40 % fat diet, it decreased by 1 %. A meta-analysis by Hu et al. [2] showed that weighted mean changes in LDLc and TC with low fat diet were −6.0 and −10.1 mg/dL, respectively; −2.1 and −4.6 mg/dL, respectively, with LCD. These studies showed some reduction in LDLc and TC with LCD, but the effects were significantly less than those with low fat diets. Yancy et al. [21] reported LDLc and TC did not change with LCD, whereas Noakes et al. [22] showed LDLc increased significantly by 7 % on very LCD, without significant change in TC. Gögebakan et al. [23] also showed significant increase in LDLc and TC with low glycemic index diet. Thus, previous studies reported varying changes in LDLc, from decrease to increase, probably depending on compositions of LCD.
As to a cutoff carbohydrate intake for defining LCD, we came to our recommendation that it be set at no more than 50 % kcal, because we found significant increment in CRP at decile 6; significant decrement in HDLc at decile 8 of plant-based LCD score; mean intakes of carbohydrate in decile 6 and decile 8 of plant-based LCD score were 50.4, and 49.5 %, respectively.
Main strengths of the present study are: (1) its population-based samples; (2) standardized collection of high-quality nutrition and laboratory measures; and (3) use of multiple procedures for quality-control. The study was limited by its cross-sectional design. Findings may or may not be generalizable to other populations. Especially due to the cross-sectional nature of this study, its results must be interpreted cautiously in regard to cause–effect relationships. The observed putatively beneficial effect of LCD on HDLc needs to be interpreted with caution since HDLc has been reported to be a weaker predictor of CVD than LDLc and CRP [38]. However, in the Japanese general population, HDLc was found to be inversely associated with all-cause mortality in a cohort study [39]. Other limitations include the following: the effect-size underestimation due to limited reliability in nutrient measurement (regressiondilution bias), despite four standardized state-of-the-art measurements; limited generalizability to persons younger than 40 years and older than 59 years; measurement error in blood chemical variables due to only one blood sample collected; non-fasting blood sample used, possibly producing error in LDLc measurement; insulin not measured, thus interpretation is speculative related to the link from LCD, lower insulin concentration, and higher HDLc concentration. INTERMAP/INTERLIPID field work was from 1997 to 1999; thus, there is the need to consider whether the Japanese habitual diet changed since. Available data indicate that the Japanese diet has not changed much: according to the National Nutritional Survey in Japan, % kcal from protein, fat, and carbohydrate were 16.0, 26.3, and 57.7 % in 1998; 14.6, 26.2, and 59.2 % in 2012 [40, 41].
Conclusions
Usual, animal-based, and plant-based LCD scores were significantly positively related to non-fasting HDLc. The plant-based LCD score was significantly inversely related to CRP. The plant-based LCD score was related to greater PUFA and lesser SFA and dietary cholesterol intake in comparison with the other two LCD scores. These may have contributed to our finding that only the plant-based LCD score was significantly inversely related to CRP. Carbohydrate intake below 50 % of total energy with higher intakes of vegetable protein and MUFA + PUFA, and lower intakes of SFA may be favorable for reducing CMRF.
Supplementary Material
Acknowledgments
The INTERMAP/INTERLIPID Study has been accomplished through the fine work of staff at local, national, and international centers. We thank Dr. Robert D. Abbott, Shiga University of Medical Science for pertinent suggestions on statistical analyses. A partial listing of colleagues is in the acknowledgement of Ref. [15]. This study was supported in part by a grant-in-aid of the Japanese Ministry of Education, Culture, Sports, Science and Technology (Grant-in-Aid for Scientific Research: (A) 090357003, (C)17590563 and (C)19590655 in Japan and the Suntory Company; the Pacific Research Institute is supported by the Robert Perry Fund and the Hawaii Community Foundation. The INTERMAP Hawaii Center was funded by the National Heart, Lung, and Blood Institute, National Institutes of Health (Grant 5-R01-HL54868–03). The INTERMAP Study is supported by the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD, U.S.A. (Grant 2-RO1-HL50490), as well as national and local agencies in the four countries.
Footnotes
Compliance with Ethical Standards
Conflict of interest None.
Electronic supplementary material The online version of this article (doi:10.1007/s00394-015-0969-z) contains supplementary material, which is available to authorized users.
References
- 1.Atkins RC (1998) Dr. Atkins’ new diet revolution, Rev edn. Avon Books, New York [Google Scholar]
- 2.Hu T, Mills KT, Yao L, Demanelis K, Eloustaz M, Yancy WS Jr, Kelly TN, He J, Bazzano LA (2012) Effects of low-carbohydrate diets versus low-fat diets on metabolic risk factors: a meta-analysis of randomized controlled clinical trials. Am J Epidemiol 176(Suppl 7):S44–S54. doi: 10.1093/aje/kws264 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Atallah R, Filion KB, Wakil SM, Genest J, Joseph L, Poirier P, Rinfret S, Schiffrin EL, Eisenberg MJ (2014) Long-term effects of 4 popular diets on weight loss and cardiovascular risk factors: a systematic review of randomized controlled trials. Circ Cardiovasc Qual Outcomes 7:815–827. doi: 10.1161/CIRC0UTC0MES.113.000723 [DOI] [PubMed] [Google Scholar]
- 4.Krieger JW, Sitren HS, Daniels MJ, Langkamp-Henken B (2006) Effects of variation in protein and carbohydrate intake on body mass and composition during energy restriction: a meta-regression 1. Am J Clin Nutr 83:260–274 [DOI] [PubMed] [Google Scholar]
- 5.Clifton PM, Condo D, Keogh JB (2014) Long term weight maintenance after advice to consume low carbohydrate, higher protein diets: a systematic review and meta analysis. Nutr Metab Cardiovasc Dis 24:224–235. doi: 10.1016/j.numecd.2013.11.006 [DOI] [PubMed] [Google Scholar]
- 6.Sacks FM, Bray GA, Carey VJ, Smith SR, Ryan DH, Anton SD, McManus K, Champagne CM, Bishop LM, Laranjo N et al. (2009) Comparison of weight-loss diets with different compositions of fat, protein, and carbohydrates. N Engl J Med 360:859–873. doi: 10.1056/NEJMoa0804748 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Fung TT, van Dam RM, Hankinson SE, Stampfer M, Willett WC, Hu FB (2010) Low-carbohydrate diets and all-cause and cause-specific mortality: two cohort studies. Ann Intern Med 153:289–298. doi: 10.7326/0003-4819-153-5-201009070-00003 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sjögren P, Becker W, Warensjö E, Olsson E, Byberg L, Gus-tafsson IB, Karlström B, Cederholm T (2010) Mediterranean and carbohydrate-restricted diets and mortality among elderly men: a cohort study in Sweden. Am J Clin Nutr 92:967–974. doi: 10.3945/ajcn.2010.29345 [DOI] [PubMed] [Google Scholar]
- 9.Lagiou P, Sandin S, Lof M, Trichopoulos D, Adami HO, Wei-derpass E (2012) Low carbohydrate-high protein diet and incidence of cardiovascular diseases in Swedish women: prospective cohort study. BMJ 344:e4026. doi: 10.1136/bmj.e4026 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Trichopoulou A, Psaltopoulou T, Orfanos P, Hsieh CC, Tricho-poulos D (2007) Low-carbohydrate-high-protein diet and long-term survival in a general population cohort. Eur J Clin Nutr 61:575–581 [DOI] [PubMed] [Google Scholar]
- 11.Halton TL, Willett WC, Liu S, Manson JE, Albert CM, Rexrode K, Hu FB (2006) Low-carbohydrate-diet score and the risk of coronary heart disease in women. N Engl J Med 355:1991–2002 [DOI] [PubMed] [Google Scholar]
- 12.Noto H, Goto A, Tsujimoto T, Noda M (2013) Low-carbohydrate diets and all-cause mortality: a systematic review and meta-analysis of observational studies. PLoS One 8:e55030. doi: 10.1371/journal.pone.0055030 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Stamler J, Elliott P, Dennis B, Dyer AR, Kesteloot H, Liu K, Ueshima H, Zhou BF (2003) INTERMAP: background, aims, design, methods, and descriptive statistics (nondietary). J Hum Hypertens 17:591–608 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Dennis B, Stamler J, Buzzard M, Conway R, Elliott P, Moag-Stahlberg A, Okayama A, Okuda N, Robertson C, Robinson F et al. (2003) INTERMAP: the dietary data—process and quality control. J Hum Hypertens 17:609–622 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Ueshima H, Okayama A, Saitoh S, Nakagawa H, Rodriguez B, Sakata K, Okuda N, Choudhury SR, Curb JD (2003) Differences in cardiovascular disease risk factors between Japanese in Japan and Japanese-Americans in Hawaii: the INTERLIPID Study. J Hum Hypertens 17:631–639 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Okuda N, Ueshima H, Okayama A, Saitoh S, Nakagawa H, Rodriguez BL, Sakata K, Okuda N, Choudhury SR, Curb JD (2005) Relation of long chain n-3 polyunsaturated fatty acid intake to serum high density lipoprotein cholesterol among Japanese men in Japan and Japanese-American men in Hawaii: the INTERLIPID Study. Atherosclerosis 178:371–379 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Nakamura Y, Ueshima H, Okuda N, Higashiyama A, Kita Y, Kadowaki T, Okamura T, Murakami Y, Okayama A, Choudhury SR et al. (2008) Relation of dietary and other lifestyle traits to difference in serum adiponectin concentration of Japanese in Japan and Hawaii: the INTERLIPID Study. Am J Clin Nutr 88:424–430 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Kannel WB, Sorlie P (1979) Some health benefits of physical activity. The Framingham Study. Arch Intern Med 139:857–861 [PubMed] [Google Scholar]
- 19.Keys A (1984) Serum cholesterol response to dietary cholesterol. Am J Clin Nutr 40:351–359 [DOI] [PubMed] [Google Scholar]
- 20.Katan MB, Zock PL, Mensink RP (1995) Dietary oils, serum lipoproteins, and coronary heart disease. Am J Clin Nutr 61(6 Suppl):1368S–1373S [DOI] [PubMed] [Google Scholar]
- 21.Yancy WS Jr, Olsen MK, Guyton JR, Bakst RP, Westman EC (2004) A low-carbohydrate, ketogenic diet versus a low-fat diet to treat obesity and hyperlipidemia: a randomized, controlled trial. Ann Intern Med 140:769–777 [DOI] [PubMed] [Google Scholar]
- 22.Noakes M, Foster PR, Keogh JB, James AP, Mamo JC, Clifton PM (2006) Comparison of isocaloric very low carbohydrate/high saturated fat and high carbohydrate/low saturated fat diets on body composition and cardiovascular risk. Nutr Metab (Lond) 3:7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Gögebakan O, Kohl A, Osterhoff MA, van Baak MA, Jebb SA, Papadaki A, Martinez JA, Handjieva-Darlenska T, Hlavaty P, Weickert MO et al. (2011) Effects of weight loss and long-term weight maintenance with diets varying in protein and glycemic index on cardiovascular risk factors: the diet, obesity, and genes (DiOGenes) study: a randomized, controlled trial. Circulation 124:2829–2838. doi: 10.1161/CIRCULATIONAHA.111.033274 [DOI] [PubMed] [Google Scholar]
- 24.Mensink RP, Zock PL, Kester AD, Katan MB (2003) Effects of dietary fatty acids and carbohydrates on the ratio of serum total to HDL cholesterol and on serum lipids and apolipoproteins: a meta-analysis of 60 controlled trials. Am J Clin Nutr 77:1146–1155 [DOI] [PubMed] [Google Scholar]
- 25.Fung TT, McCullough ML, Newby PK, Manson JE, Meigs JB, Rifai N, Willett WC, Hu FB (2005) Diet-quality scores and plasma concentrations of markers of inflammation and endothelial dysfunction. Am J Clin Nutr 82:163–173 [DOI] [PubMed] [Google Scholar]
- 26.Panagiotakos DB, Dimakopoulou K, Katsouyanni K, Bellander T, Grau M, Koenig W, Lanki T, Pistelli R, Schneider A, Peters A (2009) Mediterranean diet and inflammatory response in myocardial infarction survivors. Int J Epidemiol 38:856–866. doi: 10.1093/ije/dyp142 [DOI] [PubMed] [Google Scholar]
- 27.Pitsavos C, Panagiotakos DB, Tzima N, Lentzas Y, Chrysohoou C, Das UN, Stefanadis C (2007) Diet, exercise, and C-reactive protein levels in people with abdominal obesity: the ATTICA epidemiological study. Angiology 58:225–233 [DOI] [PubMed] [Google Scholar]
- 28.Jiang R, Jacobs DR Jr, Mayer-Davis E, Szklo M, Herrington D, Jenny NS, Kronmal R, Barr RG (2006) Nut and seed consumption and inflammatory markers in the multiethnic study of atherosclerosis. Am J Epidemiol 163:222–231 [DOI] [PubMed] [Google Scholar]
- 29.Liu S, Manson JE, Buring JE, Stampfer MJ, Willett WC, Ridker PM (2002) Relation between a diet with a high glycemic load and plasma concentrations of high-sensitivity C-reactive protein in middle-aged women. Am J Clin Nutr 75:492–498 [DOI] [PubMed] [Google Scholar]
- 30.Schwingshackl L, Hoffmann G (2013) Long-term effects of low glycemic index/load vs. high glycemic index/load diets on parameters of obesity and obesity-associated risks: a systematic review and meta-analysis. Nutr Metab Cardiovasc Dis 23:699–706. doi: 10.1016/j.numecd.2013.04.008 [DOI] [PubMed] [Google Scholar]
- 31.Schwingshackl L, Strasser B, Hoffmann G (2011) Effects of monounsaturated fatty acids on cardiovascular risk factors: a systematic review and meta-analysis. Ann Nutr Metab 59:176–186. doi: 10.1159/000334071 [DOI] [PubMed] [Google Scholar]
- 32.Jakobsen MU, O’Reilly EJ, Heitmann BL, Pereira MA, Bälter K, Fraser GE, Goldbourt U, Hallmans G, Knekt P, Liu S et al. (2009) Major types of dietary fat and risk of coronary heart disease: a pooled analysis of 11 cohort studies. Am J Clin Nutr 89:1425–1432. doi: 10.3945/ajcn.2008.27124 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Mente A, de Koning L, Shannon HS, Anand SS (2009) A systematic review of the evidence supporting a causal link between dietary factors and coronary heart disease. Arch Intern Med 169:659–669. doi: 10.1001/archinternmed.2009.38 [DOI] [PubMed] [Google Scholar]
- 34.Linseisen J, Welch AA, Ocké M, Amiano P, Agnoli C, Ferrari P, Sonestedt E, Chajès V, Bueno-de-Mesquita HB, Kaaks R et al. (2009) Dietary fat intake in the European Prospective Investigation into Cancer and Nutrition: results from the 24-h dietary recalls. Eur J Clin Nutr 63(Suppl 4):S61–S80. doi: 10.1038/ejcn.2009.75 [DOI] [PubMed] [Google Scholar]
- 35.Lopez-Garcia E, Schulze MB, Fung TT, Meigs JB, Rifai N, Manson JE, Hu FB (2004) Major dietary patterns are related to plasma concentrations of markers of inflammation and endothelial dysfunction. Am J Clin Nutr 80:1029–1035 [DOI] [PubMed] [Google Scholar]
- 36.Chrysohoou C, Panagiotakos DB, Pitsavos C, Das UN, Stefanadis C (2004) Adherence to the Mediterranean diet attenuates inflammation and coagulation process in healthy adults: the ATTICA Study. J Am Coll Cardiol 44:152–158 [DOI] [PubMed] [Google Scholar]
- 37.Esposito K, Marfella R, Ciotola M, Di Palo C, Giugliano F, Giugliano G, D’Armiento M, D’Andrea F, Giugliano D (2004) Effect of a Mediterranean-style diet on endothelial dysfunction and markers of vascular inflammation in the metabolic syndrome: a randomized trial. JAMA 292:1440–1446 [DOI] [PubMed] [Google Scholar]
- 38.Briel M, Ferreira-Gonzalez I, You JJ, Karanicolas PJ, Akl EA, Wu P, Blechacz B, Bassler D, Wei X, Sharman A et al. (2009) Association between change in high density lipoprotein cholesterol and cardiovascular disease morbidity and mortality: systematic review and meta-regression analysis. BMJ 338:b92. doi: 10.1136/bmj.b92 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Okamura T, Hayakawa T, Kadowaki T, Kita Y, Okayama A, Ueshima H (2006) The inverse relationship between serum high-density lipoprotein cholesterol level and all-cause mortality in a 9.6-year follow-up study in the Japanese general population. Atherosclerosis 184:143–150 [DOI] [PubMed] [Google Scholar]
- 40.The results of the National Nutritional Survey in Japan (1998) http://www0.nih.go.jp/eiken/chosa/kokumin_eiyou/doc_year/1998/1998_kek01.pdf. (in Japanese)
- 41.The results of the National Nutritional Survey in Japan (2012) (in Japanese). http://www.e-stat.go.jp/SG1/estat/GL08020101.do?_toGL08020101_&tstatCode=000001041744&requestSender=dsearch. (in Japanese)
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.