2021 Volume 28 Issue 12 Pages 1275-1288
Aim: A prospective cohort study in a Japanese urban general population was performed to investigate whether triglyceride (TG) and its related indices were associated with the risk for the incidence of ischemic cardiovascular disease (CVD) after the adjustment for low-density lipoprotein cholesterol (LDL-C) in Asian community dwellers.
Methods: A 15.1-year prospective cohort study was performed in 6,684 Japanese community dwellers aged 30–79 years without a history of CVD and whose fasting TG levels were ＜400 mg/dL. After adjusting for covariates, including LDL-C, the multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) of the deciles (D) of TG and those of 1-standard deviation (SD) increment of log-transformed TG (1-SD of TG) according to LDL-C level (≥ 140 and ＜140 mg/dL) for ischemic CVD incidence were estimated. The multivariable-adjusted HRs and 95%CIs of the quintiles (Q) of TG, TG/HDL-C, and the cardiometabolic index (CMI) for ischemic CVD were also estimated.
Results: In 101,230 person-years, 464 ischemic CVD cases occurred. For D10 of TG, the HR (95%CI) was 1.56 (1.05–2.32), and for 1-SD of TG, it was 1.30 (1.00–1.70) in participants with LDL-C ＜140 mg/dL and 1.07 (0.77–1.50) in those with LDL-C ≥ 140 mg/dL. For Q 5 of the CMI, the multivariable-adjusted HR was higher than those of TG and TG/HDL-C.
Conclusions: Fasting TG was an independent predictor for ischemic CVD incidence after adjusting for LDL-C in Japanese community dwellers with TG ＜400 mg/dL. Among TG, TG/HDL-C, and the CMI, the CMI could be the most powerful predictor for ischemic CVD.
See editorial vol. 28: 1263-1265
Epidemiological studies in Western countries have reported moderate and significant associations between the levels of triglyceride (TG) and the risks of coronary artery disease (CAD) 1) and stroke 2) . Furthermore, elevated TG levels appear to be associated with a residual risk of atherosclerotic cardiovascular diseases (CVDs), despite the use of low-density lipoprotein cholesterol (LDL-C)-lowering statin therapy 3) .
Previous studies in Asian 4) and Japanese 5) populations also showed an independent association between TG and the risk of ischemic CVD or CAD after the adjustment for total or high-density lipoprotein cholesterol (HDL-C). However, the association after adjusting for LDL-C or according to LDL-C level has not been well investigated in Asian prospective cohort studies. In addition, only a few studies have investigated the risk of fasting TG for CVD or CAD among Japanese general population 5, 6) .
Furthermore, the TG/HDL-C ratio (TG/HDL-C) has been reported to be as powerful a predictor of CAD as isolated high LDL-C 7) . In addition, the cardiometabolic index (CMI), defined as TG/HDL-C multiplied by the waist/height ratio, has been demonstrated to be associated with the degree of atherosclerosis in the Japanese population 8) . However, whether these indices are associated with the risk for ischemic CVD after adjusting for LDL-C has not been well investigated in the Asian population.
A prospective cohort study in a Japanese urban general population was performed to investigate whether TG and its related indices are associated with the risk for the incidence of ischemic CVD after the adjustment for LDL-C in Asian community dwellers.
The Suita study, a cohort study of CVD, was established in 1989 in Suita City, Japan. The design of the cohort study and selection criteria for participants have already been described 9) . Briefly, 12,200 and 3,000 community dwellers aged 30–79 years were randomly selected in 1989 and 1996, respectively, from the municipal population registry of Suita City. Of these individuals, 6,485 and 1,329 individuals attended the baseline examination from 1989 to 1996 and from 1996 to 1998, respectively. In addition, 546 volunteer individuals attended the baseline study in 1992–2006. All individuals described above were the participants of the baseline survey.
In the present study, 688 individuals were excluded because of the presence of a CVD history at baseline or lost to follow-up. In addition, participants aged 80 years or more (n=36), those with missing data at the baseline survey (n=291), those with fasting ＜ 10 hours (n=561), and those with TG ≥ 400 mg/dL (n=100) were excluded. Thus, 6,684 individuals were included in the analysis of the present study.
Written informed consent was obtained from each participant. The study protocol conformed to the ethical guidelines of the Declaration of Helsinki, and the protocol was approved by the Institutional Ethics Committee of the National and Cerebral and Cardiovascular Center (NCVC) (approval number M20-050-3).
Baseline StudyWell-trained nurses obtained information on smoking, alcohol drinking, and the medical history of each participant. If the participant answered yes to current drinker, alcohol consumption was determined by alcohol intake per typical week and calculated as ethanol intake per day. One drink was defined as 11.5 g of ethanol (half a gou, a traditional Japanese unit); this is nearly equal to one “standard” drink in most countries 10) .
Well-trained physicians measured blood pressure (BP) three times in the right arm using a standard mercury sphygmomanometer while the participant was seated after a 5-minute rest. The average of the second and third measurements was used in the analyses. Height in stockings and weight in light clothing were measured, and body mass index (BMI) was calculated as weight (kg) divided by height (m) squared. Trained public health nurses and technicians measured waist circumference (WC) at the umbilical level while the participant was standing.
The blood samples were collected at the NCVC after fasting for at least 10 hours. The samples were immediately centrifuged, and a routine blood examination, which included serum total cholesterol (TC), HDL-C, TG, and glucose levels, was then conducted. LDL-C was calculated using the Friedewald formula. The CMI was calculated as TG/HDL-C multiplied by the waist/height ratio. Metabolic syndrome was defined by a joint interim statement of the International Diabetes Federation Task Force on Epidemiology and Prevention; National Heart, Lung, and Blood Institute; American Heart Association; World Heart Federation; International Atherosclerosis Society; and International Association for the Study of Obesity in 2009 11) .
Follow-Up and EndpointsThe method of follow-up has been described elsewhere 12) . Briefly, the participants were followed until December 31, 2013. The first step in the survey involved checking the health status of all participants by repeat visits to the NCVC every two years and yearly questionnaires conducted by mail or telephone interview. The second step involved reviewing the in-hospital medical records of the participants suspected to have new-onset CAD or stroke or those who died because of the diseases. Reviews were performed by registered hospital physicians or research physicians blinded to the baseline information. Definite acute myocardial infarction (MI) and probable MI were defined according to the criteria of the Monitoring Trends and Determinants in Cardiovascular Disease (MONICA) project 13) . The criteria for diagnosing CAD included sudden cardiac death within 24 hours after the onset of acute symptoms and CAD, followed by coronary bypass or angiography. Stroke was defined according to the National Survey of Stroke criteria 14) , which require rapid onset of a variety of neurological deficits lasting at least 24 hours or until death. Strokes were classified as ischemic stroke, intracerebral hemorrhage, subarachnoid hemorrhage, or undetermined type. A definite stroke was defined by autopsy or diagnostic imaging, such as computed tomography or magnetic resonance imaging. Comprehensive systematic searches of death certificates were also performed to complete the surveillance of acute MI and stroke. The endpoint of the follow-up period for incident ischemic CVD was whichever of the following occurred first: (1) first diagnosis of CAD or ischemic stroke event, (2) death, (3) leaving Suita City, or (4) December 31, 2013.
Statistical AnalysisTo compare baseline risk characteristics according to the quintiles of TG, analysis of variance was used for continuous variables, and the chi-squared test was used for dichotomous variables.
To investigate the associations of serum TG levels with the risk for the incidence of ischemic CVD (CAD and ischemic stroke) and its subtypes, a Cox proportional hazards model was used to estimate the hazard ratios (HRs) and 95% confidence intervals (CIs) of the deciles of TG compared to the sixth decile (D6) for the incidence of ischemic CVD, CAD, and ischemic stroke. The models were adjusted for age, sex, the presence of hypertension (systolic BP ≥ 140 mmHg and/or diastolic BP ≥ 90 mmHg, and/or medication for hypertension) and diabetes mellitus (glucose ≥ 126 mg/dL and/or medication for diabetes mellitus), medication for dyslipidemia, smoking (never, ex, current smoker), alcohol drinking (never; ex; current drinker, ＜4 drinks/day; and current drinker, ≥ 4 drinks/day), BMI, and LDL-C in model 1, and they were additionally adjusted for HDL-C in model 2. In addition, similar analyses were performed using the Cox proportional hazards model adjusted for the Suita risk score for 10-year risk of CVD 15) . In addition, the participants were classified into two groups according to LDL-C level (≥ 140 and ＜140 mg/dL), and the multivariable-adjusted HRs and 95%CIs of 1-standard deviation (SD) increment of log-transformed TG (1-SD of TG) for ischemic CVD, CAD, and ischemic stroke were estimated in the two groups after adjusting for the same covariates.
To investigate the associations of the TG-related indices (TG, TG/HDL-C, and the CMI) with the risk for the incidence of ischemic CVD and its subtypes, a Cox proportional hazards model was used to estimate the HRs and 95% CIs of the quintiles of those indices compared with the third quintile (Q3) for the incidence of ischemic CVD, CAD, and ischemic stroke. The models were adjusted for the covariates in model 1 as mentioned above. Sex-specific analyses were also performed.
All statistical analyses were performed using SPSS, version 26.0 J (IBM, Tokyo, Japan). P＜0.05 (two-tailed) was considered statistically significant.
In 101,230 person-years, 464 ischemic CVD (268 CAD and 196 ischemic stroke) events occurred.
Table 1 shows the baseline characteristics of the participants according to the quintiles of TG level. The percentage of male participants, indices of obesity, TC, percentage of the presence of hypertension, TG/HDL-C, and CMI were significantly higher, and HDL-C was significantly lower in the participants with higher TG levels. The percentages of smoking and alcohol drinking were significantly different among the quintiles.
| Quintiles of serum TG (mg/dL) | p value | |||||
|---|---|---|---|---|---|---|
| Q1 (15-65) | Q2 (66-86) | Q3 (87-112) | Q4 (113-154) | Q5 (155-399) | ||
| N | 1355 | 1358 | 1337 | 1314 | 1320 | |
| Age (years) | 50±13 | 55±13 | 57±13 | 57±12 | 57±12 | ＜0.001 |
| Male (%) | 31.1 | 38.6 | 47.3 | 54.2 | 63.5 | ＜0.001 |
| Waist (cm) | 74.4±8.4 | 77.4±8.9 | 80.3±9.2 | 83.0±8.7 | 85.0±8.0 | ＜0.001 |
| BMI (kg/m2) | 21.0±2.7 | 21.7±2.9 | 22.5±3.0 | 23.2±3.0 | 24.0±2.8 | ＜0.001 |
| Waist/height ratio | 0.47±0.05 | 0.49±0.06 | 0.51±0.06 | 0.52±0.06 | 0.53±0.05 | ＜0.001 |
| Total cholesterol (mg/dL) | 191±32 | 203±34 | 208±34 | 214±35 | 220±38 | ＜0.001 |
| HDL-C (mg/dL) | 63±14 | 59±13 | 55±13 | 51±12 | 45±11 | ＜0.001 |
| LDL-C (mg/dL) | 117±29 | 129±32 | 133±32 | 136±33 | 132±37 | ＜0.001 |
| Non-HDL cholesterol (mg/dL) | 128±30 | 144±32 | 153±32 | 163±33 | 175±37 | ＜0.001 |
| Medication for dyslipidemia (%) | 1.2 | 2.3 | 3.1 | 3.8 | 0.9 | ＜0.001 |
| Hypertension (%) | 16.4 | 26.2 | 33.2 | 36.5 | 40.5 | ＜0.001 |
| Diabetes (%) | 2.4 | 4.0 | 4.6 | 6.0 | 7.7 | ＜0.001 |
| Smoking (%) | ＜0.001 | |||||
| Never | 66.5 | 59.6 | 54.7 | 47.9 | 41.5 | |
| Ex | 12.5 | 15.8 | 15.6 | 19.5 | 20.2 | |
| Current | 21.0 | 24.6 | 29.7 | 32.6 | 38.3 | |
| Alcohol drinking (%) | ＜0.001 | |||||
| Never | 49.6 | 48.5 | 45.9 | 45.1 | 41.4 | |
| Ex | 2.1 | 2.6 | 2.8 | 1.9 | 2.6 | |
| Current (＜4 drinks/day) | 38.7 | 37.4 | 36.9 | 36.3 | 34.6 | |
| Current (≥ 4 drinks/day) | 9.6 | 11.5 | 14.4 | 16.7 | 21.4 | |
| The Suita risk score§ | 6.7 | 13.8 | 17.3 | 19.4 | 21.2 | ＜0.001 |
| Metabolic syndrome (%) | 0.5 | 1.7 | 4.9 | 12.4 | 54.1 | ＜0.001 |
| TG (mg/dL) § | 54 | 76 | 98 | 131 | 200 | ＜0.001 |
| TG/HDL-C ratio§ | 0.86 | 1.29 | 1.82 | 2.65 | 4.77 | ＜0.001 |
| CMI§ | 0.40 | 0.63 | 0.92 | 1.37 | 2.52 | ＜0.001 |
§: median. TG: triglycerides, HDL-C: high-density lipoprotein cholesterol, CMI: cardiometabolic index, BMI: body mass index, LDL-C: low- density lipoprotein cholesterol.
The Suita risk score was calculated by a risk score for predicting 10-year risk of CVD 15) .
Metabolic syndrome was defined by Joint statement from the IDF, AHA/NHLBI, the World Heart Federation, the International Atherosclerosis Society, and the International Association for the Study of Obesity in 2009 11) .
Table 2 shows the age- and sex-adjusted and multivariable-adjusted HRs (95%CI) of the deciles of TG for ischemic CVD, CAD, and ischemic stroke. The multivariable-adjusted HRs of D9 and D10 for ischemic CVD was significantly increased compared with that of D6 (98 ≤ TG ≤ 111 mg/dL) in model 1 (p for trend ＜0.05), although the increase of HRs of D9 and D10 were not statistically significant after the additional adjustment for HDL-C in model 2. The multivariable-adjusted HRs of D10 for CAD was increased with borderline significance, and those of higher deciles for ischemic stroke were increased without statistical significance in model 1. In addition, the HRs of the deciles of TG for ischemic CVD and its subtypes adjusted for the Suita risk score were shown in Supplemental Table 1 . The HRs were almost similar to those in Table 2 ; however, the risk of D9 for ischemic CVD was significantly increased.
| N of participants | No. of events | Crude incidence (/1,000 person- years) | Age and sex-adjusted HR (95%CI) |
Multivariable- adjusted HR (95%CI) (Model 1) |
Multivariable- adjusted HR (95%CI) (Model 2) |
|
|---|---|---|---|---|---|---|
| Ischemic CVD | ||||||
| D1 (15-54) | 699 | 20 | 1.8 | 0.86 (0.50-1.49) | 1.15 (0.66-1.99) | 1.27 (0.73-2.22) |
| D2 (55-65) | 656 | 29 | 2.8 | 1.05 (0.65-1.70) | 1.25 (0.77-2.03) | 1.36 (0.83-2.22) |
| D3 (66-75) | 660 | 37 | 3.7 | 1.18 (0.75-1.85) | 1.25 (0.80-1.98) | 1.34 (0.85-2.12) |
| D4 (76-86) | 698 | 38 | 3.6 | 1.03 (0.66-1.61) | 1.15 (0.73-1.81) | 1.21 (0.77-1.91) |
| D5 (87-97) | 629 | 47 | 5.0 | 1.41 (0.92-2.17) | 1.45 (0.94-2.22) | 1.47 (0.96-2.26) |
| D6 (98-111) | 667 | 38 | 3.8 | 1.00 | 1.00 | 1.00 |
| D7 (112-130) | 686 | 54 | 5.4 | 1.37 (0.90-2.07) | 1.31 (0.86-1.99) | 1.29 (0.85-1.96) |
| D8 (131-153) | 641 | 60 | 6.3 | 1.48 (0.99-2.22) | 1.43 (0.95-2.15) | 1.37 (0.91-2.06) |
| D9 (154-198) | 675 | 70 | 6.9 | 1.68 (1.13-2.49) | 1.52 (1.02-2.27) | 1.44 (0.97-2.15) |
| D10 (199-399) | 673 | 71 | 7.1 | 1.75 (1.18-2.60) | 1.56 (1.05-2.32) | 1.39 (0.93-2.09) |
| Trend p | p＜0.001 | p＜0.05 | p = 0.490 | |||
| CAD | ||||||
| D1 (15-54) | 699 | 11 | 1.0 | 0.86 (0.41-1.79) | 1.17 (0.56-2.46) | 1.34 (0.64-2.83) |
| D2 (55-65) | 656 | 12 | 1.1 | 0.78 (0.38-1.58) | 0.95 (0.47-1.94) | 1.06 (0.51-2.17) |
| D3 (66-75) | 660 | 25 | 2.4 | 1.46 (0.82-2.61) | 1.56 (0.87-2.80) | 1.70 (0.94-3.05) |
| D4 (76-86) | 698 | 20 | 1.8 | 0.97 (0.53-1.80) | 1.09 (0.59-2.02) | 1.17 (0.63-2.16) |
| D5 (87-97) | 629 | 25 | 2.6 | 1.36 (0.76-2.43) | 1.40 (0.79-2.51) | 1.44 (0.81-2.57) |
| D6 (98-111) | 667 | 21 | 2.0 | 1.00 | 1.00 | 1.00 |
| D7 (112-130) | 686 | 36 | 3.5 | 1.66 (0.97-2.85) | 1.59 (0.93-2.73) | 1.58 (0.92-2.71) |
| D8 (131-153) | 641 | 35 | 3.6 | 1.59 (0.93-2.73) | 1.53 (0.89-2.63) | 1.44 (0.84-2.48) |
| D9 (154-198) | 675 | 42 | 4.0 | 1.80 (1.06-3.04) | 1.59 (0.94-2.70) | 1.50 (0.88-2.54) |
| D10 (199-399) | 673 | 41 | 4.0 | 1.82 (1.08-3.08) | 1.70 (0.99-2.89) | 1.48 (0.86-2.55) |
| Trend p | p＜0.01 | p = 0.060 | p = 0.484 | |||
| Ischemic stroke | ||||||
| D1 (15-54) | 699 | 9 | 0.8 | 0.90 (0.40-2.03) | 1.11 (0.49-2.52) | 1.19 (0.52-2.73) |
| D2 (55-65) | 656 | 17 | 1.7 | 1.43 (0.73-2.80) | 1.62 (0.82-3.19) | 1.71 (0.86-3.39) |
| D3 (66-75) | 660 | 12 | 1.2 | 0.86 (0.41-1.81) | 0.92 (0.44-1.92) | 0.96 (0.45-2.02) |
| D4 (76-86) | 698 | 18 | 1.7 | 1.08 (0.56-2.10) | 1.20 (0.62-2.34) | 1.24 (0.64-2.42) |
| D5 (87-97) | 629 | 22 | 2.3 | 1.48 (0.78-2.78) | 1.48 (0.78-2.78) | 1.49 (0.79-2.82) |
| D6 (98-111) | 667 | 17 | 1.7 | 1.00 | 1.00 | 1.00 |
| D7 (112-130) | 686 | 18 | 1.8 | 1.03 (0.53-2.00) | 0.98 (0.51-1.91) | 0.97 (0.50-1.89) |
| D8 (131-153) | 641 | 25 | 2.6 | 1.38 (0.75-2.55) | 1.35 (0.73-2.51) | 1.32 (0.71-2.44) |
| D9 (154-198) | 675 | 28 | 2.7 | 1.51 (0.83-2.76) | 1.43 (0.78-2.62) | 1.38 (0.75-2.54) |
| D10 (199-399) | 673 | 30 | 2.9 | 1.67 (0.92-3.03) | 1.36 (0.74-2.49) | 1.26 (0.68-2.33) |
| Trend p | p＜0.05 | p = 0.491 | p = 0.876 |
TG: triglycerides, LDL-C: low-density lipoprotein cholesterol, HDL-C: high-density lipoprotein cholesterol, HR: hazard ratio, CI: confidence interval, CVD: cardiovascular diseases.
Model 1: age, sex, hypertension, diabetes, medication for dyslipidemia, smoking, alcohol drinking, body mass index, and LDL-C were adjusted. Model 2: the variables in model 1 and HDL-C were adjusted.
| N of participants | No. of events | Crude incidence
(/1,000 person-years) |
Multivariable-
adjustedHR (95%CI) |
|
|---|---|---|---|---|
| Ischemic CVD | ||||
| D1 (15-54) | 699 | 20 | 1.8 | 1.18 (0.68-2.05) |
| D2 (55-65) | 656 | 29 | 2.8 | 1.26 (0.77-2.06) |
| D3 (66-75) | 660 | 37 | 3.7 | 1.32 (0.84-2.09) |
| D4 (76-86) | 698 | 38 | 3.6 | 1.19 (0.75-1.87) |
| D5 (87-97) | 629 | 47 | 5.0 | 1.38 (0.89-2.14) |
| D6 (98-111) | 667 | 38 | 3.8 | 1.00 |
| D7 (112-130) | 686 | 54 | 5.4 | 1.34 (0.88-2.03) |
| D8 (131-153) | 641 | 60 | 6.3 | 1.41 (0.93-2.13) |
| D9 (154-198) | 675 | 70 | 6.9 | 1.51 (1.02-2.25) |
| D10 (199-399) | 673 | 71 | 7.1 | 1.37 (0.92-2.04) |
| Trend p | p = 0.278 | |||
| CAD | ||||
| D1 (15-54) | 699 | 11 | 1.0 | 1.20 (0.57-2.52) |
| D2 (55-65) | 656 | 12 | 1.1 | 0.90 (0.43-1.88) |
| D3 (66-75) | 660 | 25 | 2.4 | 1.68 (0.93-3.03) |
| D4 (76-86) | 698 | 20 | 1.8 | 1.14 (0.61-2.12) |
| D5 (87-97) | 629 | 25 | 2.6 | 1.27 (0.69-2.33) |
| D6 (98-111) | 667 | 21 | 2.0 | 1.00 |
| D7 (112-130) | 686 | 36 | 3.5 | 1.66 (0.96-2.87) |
| D8 (131-153) | 641 | 35 | 3.6 | 1.59 (0.92-2.75) |
| D9 (154-198) | 675 | 42 | 4.0 | 1.68 (0.99-2.86) |
| D10 (199-399) | 673 | 41 | 4.0 | 1.46 (0.85-2.50) |
| Trend p | p = 0.186 | |||
| Ischemic stroke | ||||
| D1 (15-54) | 699 | 9 | 0.8 | 1.18 (0.52-2.66) |
| D2 (55-65) | 656 | 17 | 1.7 | 1.71 (0.87-3.37) |
| D3 (66-75) | 660 | 12 | 1.2 | 0.94 (0.45-1.97) |
| D4 (76-86) | 698 | 18 | 1.7 | 1.20 (0.62-2.34) |
| D5 (87-97) | 629 | 22 | 2.3 | 1.49 (0.79-2.80) |
| D6 (98-111) | 667 | 17 | 1.7 | 1.00 |
| D7 (112-130) | 686 | 18 | 1.8 | 0.97 (0.50-1.88) |
| D8 (131-153) | 641 | 25 | 2.6 | 1.25 (0.67-2.32) |
| D9 (154-198) | 675 | 28 | 2.7 | 1.32 (0.72-2.41) |
| D10 (199-399) | 673 | 30 | 2.9 | 1.27 (0.70-2.31) |
| Trend p | p = 0.858 |
Multivariable-adjusted HR: the scores calculated by the Suita risk score 15) were adjusted.
Table 3 shows the age- and sex-adjusted and multivariable-adjusted HRs (95%CI) of 1-SD of TG for ischemic CVD, CAD, and ischemic stroke according to LDL-C level. The multivariable-adjusted HR for CAD was significantly increased and that for ischemic CVD was increased with borderline significance in the participants with LDL-C ＜140 mg/dL in model 1; however, the increase of HRs was not statistically significant after further adjustment for HDL-C in model 2.
|
N of participants |
No. of events |
Crude incidence (/1,000 person- years) |
Age and sex-adjusted HR (95%CI) |
Multivariable- adjusted HR (95%CI) (Model 1) |
Multivariable- adjusted HR (95%CI) (Model 2) |
|
|---|---|---|---|---|---|---|
| Ischemic CVD | ||||||
| LDL-C ＜140 mg/dL | ||||||
| ln TG | 4307 | 260 | 4.0 | 1.46 (1.15-1.86) | 1.30 (1.00-1.70) | 1.22 (0.91-1.62) |
| LDL-C ≥ 140 mg/dL | ||||||
| ln TG | 2377 | 204 | 5.6 | 1.38 (1.00-1.90) | 1.07 (0.77-1.50) | 0.82 (0.57-1.18) |
| CAD | ||||||
| LDL-C ＜140 mg/dL | ||||||
| ln TG | 4307 | 128 | 1.9 | 1.70 (1.04-2.06) | 1.51 (1.03-2.20) | 1.45 (0.96-2.18) |
| LDL-C ≥ 140 mg/dL | ||||||
| ln TG | 2377 | 140 | 3.7 | 1.42 (0.97-2.08) | 1.09 (0.73-1.63) | 0.76 (0.49-1.18) |
| Ischemic stroke | ||||||
| LDL-C ＜140 mg/dL | ||||||
| ln TG | 4307 | 132 | 2.0 | 1.44 (1.03-2.01) | 1.14 (0.79-1.64) | 1.04 (0.70-1.55) |
| LDL-C ≥ 140 mg/dL | ||||||
| ln TG | 2377 | 64 | 1.7 | 1.32 (0.75-2.34) | 1.07 (0.59-1.95) | 1.00 (0.52-1.91) |
TG: triglycerides, LDL-C: low-density lipoprotein cholesterol, HDL-C: high-density lipoprotein cholesterol, HR: hazard ratio, CI: confidence interval, CVD: cardiovascular diseases.
Model 1: age, sex, hypertension, diabetes, medication for dyslipidemia, smoking, alcohol drinking, body mass index, and LDL-C were adjusted. Model 2: the variables in model 1 and HDL-C were adjusted.
Fig.1 shows the multivariable-adjusted HR of the quintiles of TG, TG/HDL-C, and the CMI for ischemic CVD, CAD, and ischemic stroke compared with that in Q3 after adjusting for covariates, including LDL-C (model 1). For ischemic CVD, the HRs of Q5 in TG and TG/HDL-C and those of Q4 and Q5 in the CMI were significantly increased, and the HR of Q5 was the highest in the CMI among the three indices. For CAD, the HR of Q5 was significantly increased in three indices and that of the CMI was the highest among them. The HRs of the indices for ischemic stroke were not significantly increased; however, the increase of HR of Q5 in the CMI showed borderline significance. The results of sex-specific analysis are presented in Supplemental Tables 2 and 3 . For CVD, the HR of Q5 of TG/HDL-C in women and those of Q5 in the CMI in both sexes were significantly increased. For CAD, the tendencies of the results were almost the same as those for CVD without statistical significance. For ischemic stroke, the HRs of Q2 and Q5 of all indices tended to be increased in both sexes, especially in the CMI of women. In addition, higher CMI was associated with higher risk for ischemic CVD and CAD in the participants without metabolic syndrome with statistical significance of trend p (trend p: ischemic CVD, p＜0.05; CAD, p＜0.01); however, the association between the CMI and the risks of ischemic CVD and its subtypes was not observed in the participants with metabolic syndrome (data not shown).
TG: triglycerides, HDL-C: high-density lipoprotein cholesterol, CMI: cardiometabolic index, HR: hazard ratio.
Quintiles of TG (mg/dL): Q1 15–65, Q2 66–86, Q3 87–112, Q4 113–154, Q5 155–399.
Quintiles of TG/HDL-C: Q1 0.17–1.05, Q2 1.06–1.52, Q3 1.53–2.15, Q4 2.16–3.39, Q5 3.40–16.86.
Quintiles of CMI: Q 1 0.06–0.50, Q2 0.51–0.75, Q3 0.76–1.11, Q4 1.12–1.78, Q5 1.79–8.82.
The HRs age, sex, hypertension, diabetes, medication for dyslipidemia, smoking, alcohol drinking, body mass index, and low-density lipoprotein cholesterol were adjusted.
|
N of participants |
No. of events |
Crude incidence (/1,000 person-years) |
Age-adjusted HR (95%CI) |
Multivariable-adjusted HR (95%CI) |
|
|---|---|---|---|---|---|
| TG (mg/dL) | |||||
| Men | |||||
| Q1 (28-73) | 610 | 45 | 5.1 | 0.90 (0.61-1.34) | 1.08 (0.72-1.62) |
| Q2 (74-98) | 649 | 61 | 6.8 | 1.10 (0.77-1.59) | 1.25 (0.87-1.81) |
| Q3 (99-128) | 627 | 54 | 6.0 | 1.00 | 1.00 |
| Q4 (129-175) | 617 | 69 | 7.8 | 1.31 (0.92-1.87) | 1.28 (0.89-1.83) |
| Q5 (176-399) | 626 | 68 | 7.4 | 1.34 (0.94-1.91) | 1.26 (0.88-1.81) |
| Trend p | p＜0.05 | p = 0.406 | |||
| Women | |||||
| Q1 (15-59) | 703 | 16 | 1.4 | 1.08 (0.58-1.99) | 1.33 (0.71-2.49) |
| Q2 (60-77) | 726 | 24 | 2.0 | 0.98 (0.57-1.67) | 1.07 (0.62-1.83) |
| Q3 (78-98) | 707 | 30 | 2.7 | 1.00 | 1.00 |
| Q4 (99-134) | 714 | 37 | 3.4 | 1.10 (0.68-1.79) | 1.10 (0.68-1.78) |
| Q5 (135-397) | 705 | 60 | 5.5 | 1.57 (1.01-2.44) | 1.37 (0.88-2.15) |
| Trend p | p＜0.05 | p = 0.266 | |||
| TG/HDL-C ratio | |||||
| Men | |||||
| Q1 (0.34-1.29) | 625 | 40 | 4.4 | 0.73 (0.48-1.09) | 0.84 (0.56-1.27) |
| Q2 (1.30-1.91) | 632 | 56 | 6.4 | 1.10 (0.76-1.59) | 1.14 (0.79-1.66) |
| Q3 (1.92-2.72) | 623 | 56 | 6.2 | 1.00 | 1.00 |
| Q4 (2.73-4.13) | 623 | 69 | 7.6 | 1.24 (0.87-1.76) | 1.09 (0.77-1.56) |
| Q5 (4.14-16.86) | 626 | 76 | 8.4 | 1.44 (1.02-2.04) | 1.37 (0.96-1.94) |
| Trend p | p＜0.01 | p＜0.05 | |||
| Women | |||||
| Q1 (0.17-0.93) | 716 | 16 | 1.4 | 1.10 (0.58-2.08) | 1.31 (0.68-2.50) |
| Q2 (0.94-1.27) | 712 | 25 | 2.2 | 1.45 (0.82-2.56) | 1.50 (0.85-2.64) |
| Q3 (1.28-1.77) | 714 | 23 | 2.0 | 1.00 | 1.00 |
| Q4 (1.78-2.65) | 706 | 36 | 3.2 | 1.35 (0.80-2.27) | 1.42 (0.84-2.40) |
| Q5 (2.66-13.90) | 707 | 67 | 6.2 | 2.30 (1.43-3.70) | 2.08 (1.28-3.38) |
| Trend p | p＜0.001 | p＜0.01 | |||
| CMI | |||||
| Men | |||||
| Q1 (0.16-0.62) | 642 | 42 | 4.6 | 0.79 (0.53-1.18) | 0.94 (0.62-1.43) |
| Q2 (0.63-0.93) | 610 | 53 | 6.0 | 1.06 (0.72-1.54) | 1.11 (0.76-1.62) |
| Q3 (0.94-1.37) | 632 | 54 | 6.0 | 1.00 | 1.00 |
| Q4 (1.38-2.14) | 621 | 68 | 7.7 | 1.25 (0.88-1.79) | 1.13 (0.79-1.63) |
| Q5 (2.15-8.82) | 624 | 80 | 8.9 | 1.52 (1.08-2.15) | 1.43 (1.00-2.05) |
| Trend p | p＜0.001 | p＜0.05 | |||
| Women | |||||
| Q1 (0.06-0.43) | 719 | 16 | 1.4 | 1.25 (0.66-2.36) | 1.52 (0.79-2.92) |
| Q2 (0.44-0.62) | 703 | 25 | 2.2 | 1.34 (0.76-2.34) | 1.43 (0.81-2.52) |
| Q3 (0.63-0.91) | 728 | 24 | 2.0 | 1.00 | 1.00 |
| Q4 (0.92-1.42) | 695 | 33 | 3.1 | 1.19 (0.71-2.02) | 1.20 (0.70-2.04) |
| Q5 (1.43-8.82) | 710 | 69 | 6.4 | 2.20 (1.38-3.51) | 1.89 (1.16-3.08) |
| Trend p | p＜0.001 | p＜0.05 |
TG: triglycerides, HDL-C: high-density lipoprotein cholesterol, CMI: cardiometabolic index, HR: hazard ratio, CI: confidence interval, CVD: cardiovascular diseases.
Multivariable-adjusted HR: age, sex, hypertension, diabetes, medication for dyslipidemia, smoking, alcohol drinking, body mass index, and low-density lipoprotein cholesterol were adjusted.
|
N of participants |
No. of events |
Crude incidence (/1,000 person-years) |
Age-adjusted HR (95%CI) |
Multivariable-adjusted HR (95%CI) |
|
|---|---|---|---|---|---|
| CAD | |||||
| TG (mg/dL) | |||||
| Men | |||||
| Q1 (28-73) | 610 | 27 | 3.0 | 0.86 (0.52-1.43) | 1.06 (0.63-1.78) |
| Q2 (74-98) | 649 | 32 | 3.4 | 0.91 (0.56-1.47) | 1.04 (0.64-1.70) |
| Q3 (99-128) | 627 | 34 | 3.7 | 1.00 | 1.00 |
| Q4 (129-175) | 617 | 43 | 4.7 | 1.28 (0.82-2.01) | 1.23 (0.79-1.94) |
| Q5 (176-399) | 626 | 38 | 4.0 | 1.17 (0.74-1.86) | 1.14 (0.71-1.82) |
| Trend p | p = 0.118 | p = 0.589 | |||
| Women | |||||
| Q1 (15-59) | 703 | 9 | 0.8 | 0.97 (0.43-2.19) | 1.33 (0.58-3.02) |
| Q2 (60-77) | 726 | 10 | 0.8 | 0.67 (0.31-1.45) | 0.78 (0.36-1.71) |
| Q3 (78-98) | 707 | 18 | 1.6 | 1.00 | 1.00 |
| Q4 (99-134) | 714 | 23 | 2.0 | 1.12 (0.61-2.08) | 1.15 (0.62-2.13) |
| Q5 (135-397) | 705 | 34 | 3.0 | 1.48 (0.83-2.62) | 1.27 (0.71-2.28) |
| Trend p | p＜0.05 | p = 0.370 | |||
| TG/HDL-C ratio | |||||
| Men | |||||
| Q1 (0.34-1.29) | 625 | 23 | 2.5 | 0.73 (0.43-1.25) | 0.87 (0.50-1.49) |
| Q2 (1.30-1.91) | 632 | 30 | 3.3 | 0.98 (0.60-1.61) | 1.03 (0.62-1.69) |
| Q3 (1.92-2.72) | 623 | 33 | 3.5 | 1.00 | 1.00 |
| Q4 (2.73-4.13) | 623 | 46 | 4.9 | 1.41 (0.90-2.20) | 1.21 (0.77-1.91) |
| Q5 (4.14-16.86) | 626 | 42 | 4.5 | 1.34 (0.85-2.12) | 1.32 (0.83-2.10) |
| Trend p | p＜0.05 | p = 0.103 | |||
| Women | |||||
| Q1 (0.17-0.93) | 716 | 8 | 0.7 | 0.88 (0.37-2.11) | 1.12 (0.47-2.71) |
| Q2 (0.94-1.27) | 712 | 11 | 1.0 | 1.01 (0.46-2.23) | 1.06 (0.48-2.35) |
| Q3 (1.28-1.77) | 714 | 14 | 1.2 | 1.00 | 1.00 |
| Q4 (1.78-2.65) | 706 | 23 | 2.0 | 1.40 (0.72-2.73) | 1.44 (0.74-2.82) |
| Q5 (2.66-13.90) | 707 | 38 | 3.4 | 2.12 (1.15-3.93) | 1.85 (0.98-3.49) |
| Trend p | p＜0.01 | p＜0.05 | |||
| CMI | |||||
| Men | |||||
| Q1 (0.16-0.62) | 642 | 24 | 2.6 | 0.78 (0.46-1.33) | 0.94 (0.54-1.62) |
| Q2 (0.63-0.93) | 610 | 28 | 3.1 | 0.95 (0.57-1.57) | 0.98 (0.59-1.63) |
| Q3 (0.94-1.37) | 632 | 32 | 3.4 | 1.00 | 1.00 |
| Q4 (1.38-2.14) | 621 | 45 | 4.9 | 1.41 (0.90-2.22) | 1.24 (0.78-1.96) |
| Q5 (2.15-8.82) | 624 | 45 | 4.9 | 1.45 (0.92-2.28) | 1.41 (0.88-2.25) |
| Trend p | p＜0.01 | p = 0.066 | |||
| Women | |||||
| Q1 (0.06-0.43) | 719 | 8 | 0.7 | 0.80 (0.35-1.86) | 1.03 (0.43-2.42) |
| Q2 (0.44-0.62) | 703 | 9 | 0.8 | 0.61 (0.28-1.36) | 0.65 (0.29-1.45) |
| Q3 (0.63-0.91) | 728 | 18 | 1.5 | 1.00 | 1.00 |
| Q4 (0.92-1.42) | 695 | 19 | 1.7 | 0.91 (0.47-1.73) | 0.89 (0.46-1.71) |
| Q5 (1.43-8.82) | 710 | 40 | 3.6 | 1.68 (0.96-2.94) | 1.44 (0.80-2.61) |
| Trend p | p＜0.01 | p＜0.05 | |||
| Ischemic stroke | |||||
| TG (mg/dL) | |||||
| Men | |||||
| Q1 (28-73) | 610 | 18 | 2.0 | 0.99 (0.52-1.86) | 1.14 (0.59-2.18) |
| Q2 (74-98) | 649 | 29 | 3.1 | 1.40 (0.79-2.47) | 1.53 (0.86-2.71) |
| Q3 (99-128) | 627 | 20 | 2.2 | 1.00 | 1.00 |
| Q4 (129-175) | 617 | 26 | 2.9 | 1.35 (0.75-2.41) | 1.34 (0.75-2.40) |
| Q5 (176-399) | 626 | 30 | 3.2 | 1.59 (0.90-2.80) | 1.43 (0.80-2.53) |
| Trend p | p = 0.135 | p = 0.571 | |||
| Women | |||||
| Q1 (15-59) | 703 | 7 | 0.6 | 1.23 (0.48-3.15) | 1.35 (0.52-3.52) |
| Q2 (60-77) | 726 | 14 | 1.2 | 1.44 (0.67-3.12) | 1.48 (0.68-3.22) |
| Q3 (78-98) | 707 | 12 | 1.1 | 1.00 | 1.00 |
| Q4 (99-134) | 714 | 14 | 1.3 | 1.04 (0.48-2.26) | 1.00 (0.46-2.17) |
| Q5 (135-397) | 705 | 26 | 2.3 | 1.66 (0.84-3.30) | 1.49 (0.74-3.00) |
| Trend p | p = 0.284 | p = 0.590 | |||
| TG/HDL-C ratio | |||||
| Men | |||||
| Q1 (0.34-1.29) | 625 | 17 | 1.9 | 0.75 (0.40-1.41) | 0.84 (0.44-1.59) |
| Q2 (1.30-1.91) | 632 | 26 | 2.9 | 1.20 (0.69-2.11) | 1.23 (0.70-2.16) |
| Q3 (1.92-2.72) | 623 | 23 | 2.5 | 1.00 | 1.00 |
| Q4 (2.73-4.13) | 623 | 23 | 2.5 | 0.99 (0.56-1.76) | 0.92 (0.51-1.65) |
| Q5 (4.14-16.86) | 626 | 34 | 3.7 | 1.56 (0.92-2.65) | 1.42 (0.83-2.44) |
| Trend p | p＜0.05 | p = 0.152 | |||
| Women | |||||
| Q1 (0.17-0.93) | 716 | 8 | 0.7 | 1.45 (0.56-3.76) | 1.57 (0.60-4.13) |
| Q2 (0.94-1.27) | 712 | 14 | 1.2 | 2.13 (0.92-4.92) | 2.14 (0.92-4.97) |
| Q3 (1.28-1.77) | 714 | 9 | 0.8 | 1.00 | 1.00 |
| Q4 (1.78-2.65) | 706 | 13 | 1.2 | 1.22 (0.52-2.85) | 1.28 (0.54-3.01) |
| Q5 (2.66-13.90) | 707 | 29 | 2.6 | 2.48 (1.17-5.26) | 2.31 (1.07-4.98) |
| Trend p | p = 0.057 | p = 0.145 | |||
| CMI | |||||
| Men | |||||
| Q1 (0.16-0.62) | 642 | 18 | 1.9 | 0.84 (0.45-1.56) | 0.99 (0.52-1.90) |
| Q2 (0.63-0.93) | 610 | 25 | 2.8 | 1.22 (0.69-2.16) | 1.28 (0.72-2.29) |
| Q3 (0.94-1.37) | 632 | 22 | 2.4 | 1.00 | 1.00 |
| Q4 (1.38-2.14) | 621 | 23 | 2.5 | 1.04 (0.58-1.87) | 0.99 (0.55-1.79) |
| Q5 (2.15-8.82) | 624 | 35 | 3.8 | 1.63 (0.96-2.78) | 1.48 (0.86-2.57) |
| Trend p | p＜0.05 | p = 0.215 | |||
| Women | |||||
| Q1 (0.06-0.43) | 719 | 8 | 0.7 | 2.62 (0.90-7.60) | 2.98 (1.01-8.82) |
| Q2 (0.44-0.62) | 703 | 16 | 1.4 | 3.55 (1.39-9.08) | 3.89 (1.51-10.03) |
| Q3 (0.63-0.91) | 728 | 6 | 0.5 | 1.00 | 1.00 |
| Q4 (0.92-1.42) | 695 | 14 | 1.3 | 1.99 (0.77-5.20) | 2.00 (0.76-5.23) |
| Q5 (1.43-8.82) | 710 | 29 | 2.6 | 3.64 (1.51-8.78) | 3.11 (1.26-7.69) |
| Trend p | p = 0.098 | p = 0.365 |
TG: triglycerides, HDL-C: high-density lipoprotein cholesterol, CMI: cardiometabolic index, HR: hazard ratio, CI: confidence interval, CAD: coronary artery disease.
Multivariable-adjusted HR: age, sex, hypertension, diabetes, medication for dyslipidemia, smoking, alcohol drinking, body mass index, and low- density lipoprotein cholesterol were adjusted.
In the present study, higher serum TG levels were significantly associated with the increased risk for the incidence of ischemic CVD after the adjustment for CVD risk factors, including LDL-C, in Japanese general population with TG ＜400 mg/dL, although the association was attenuated after the additional adjustment for HDL-C. In addition, the association between TG and the risk for ischemic CVD and CAD was more evident in those with LDL-C ＜140 mg/dL. Furthermore, the risk of the top quintile of the CMI for ischemic CVD was higher than those of TG and TG/HDL-C.
In previous large meta-analyses of cohort studies in Western populations, significant associations between TG levels and the risk for CAD and stroke have been reported 1, 2) ; however, another meta-analysis showed that TG was not an independent predictor for CAD and ischemic stroke after adjusting for HDL-C and non-HDL-C 16) . In the Asian-Pacific population, a meta-analysis showed that an elevated TG level was an independent risk factor for CAD after adjusting for the TC/HDL-C ratio 4) . In the Japanese population, Iso et al. showed a significantly increased risk of a high TG level for ischemic CVD after the adjustment for TC in community dwellers 5) , and Satoh et al. showed that the risk of a high TG level for CAD was significantly increased after the adjustment for TC and HDL-C in middle-aged male workers 6) . However, the risks of the TG level for the incidence of CVD after adjusting for LDL-C have not been well investigated in the Asian general population fasting ≥ 10 hours. Because most of the blood samples were collected after fasting ≥ 10 hours and because there were only a few participants with TG ≥ 400 mg/dL at baseline in the Suita study, the model could be adjusted for LDL-C calculated using Friedewald formula in the present study. Thus, to the best of our knowledge, the present study is the first to show the risk of TG level for incident ischemic CVD and its subtypes after adjusting for LDL-C calculated using Friedewald formula in the Japanese general population. In addition, the present study is also the first to show the risk of the CMI for ischemic CVD in a prospective cohort study.
According to a previous study in Japan, the risk of patients with fasting TG ＞100 mg/dL for CVD events was significantly elevated compared with those with TG ＜100 mg/dL, after adjusting for confounders, including LDL-C and HDL-C 17) . However, because the participants of the previous study were patients with CAD, comparing the results in the previous and present study is difficult. Furthermore, only a few previous studies have investigated the risk of fasting TG for ischemic CVD or CAD among Asian general population 4- 6) . Thus, the cutoff value of fasting TG for the risk of ischemic CVD should be further investigated among Asian general population in future.
In the present study, the risk of TG for ischemic CVD and CAD was more evident in the participants with LDL-C ＜140 mg/dL than those with LDL-C ≥ 140 mg/dL. In the previous study in Western population 16) , the risk of non-HDL-C according to HDL-C level was investigated; however, studies investigating the risk of TG according to LDL-C level in the Asian general population are few. The present study showed the importance of controlling TG levels for CAD prevention, even in individuals with LDL-C ＜140 mg/dL.
Among several TG-related indices, TG/HDL-C and the CMI were selected in the present study. Although the lipid accumulation product (LAP) is also an index that includes TG in its formula 18) , the LAP is not always useful for the Japanese population because WC is lower than 65 cm in some men or 58 cm in some women. In addition, as we have previously reported, the CMI was more strongly associated with both the cardio-ankle vascular index in a Japanese general population and the degree of atherosclerotic progression in patients with PAD than the LAP 7, 19) . In addition, according to previous studies, HDL-C was independent of the TG concentration 15) , and the waist/height ratio is a predictive factor for CVD 20) . Thus, TG/HDL-C and the CMI could be candidate predictors in identifying individuals at high risk for ischemic CVD. The present study suggests that the CMI could be the most significant predictor for future ischemic CVD risk among the three TG-related indices; it also provides a message for the strategy of lipid control, “to prevent ischemic CVD, in addition to controlling LDL-C, keep lower WC, lower TG, and higher HDL-C,” which is a concept of the metabolic syndrome.
In the sex-specific analyses of the present study, the association between higher quintiles of TG-related indices and ischemic CVD risk was more evident in women than in men. Iso et al. showed a significant increase of ischemic CVD risk for the highest quartile of fasting TG in men, but not in women, due to the small number of fasting women 5) . In the ARIC study, the age-adjusted relative ratio of the top quintile of TG for CAD was greater in women than in men 21) . In other previous studies that investigated the risk of TG/HDL-C for incident CVD, sex-specific analyses were not performed 22) , or only the risk assessment of TG/HDL-C ＞3.5 for men and TG/HDL-C ＞2.5 for women was performed 23) . In addition, few studies have investigated the risk of the CMI for the incidence of CVD. Although the reason for the higher HRs of the higher quintiles of TG-related indices in women than in men is not clear in the present study, a much lower incidence in the reference quintile of women than of men might be one reason for the difference. In addition, higher risks for ischemic CVD and stroke among women with lower quintiles of the indices were found in the present study, which few previous studies have reported. Because the risk of lower TG-related indices might not yet have been fully investigated, especially in sex-specific analyses, larger epidemiological studies, including sex-specific analyses, are needed for the further assessment of the risks of TG-related indices.
The mechanism of the involvement of TGs in the onset of ischemic CVD is considered as follows. According to Ohmura 24) , “the core TGs comprising chylomicrons (CMs) and very low-density lipoproteins (VLDLs) are rapidly catabolized in the circulating blood by lipoprotein lipase (LPL) on the blood vessel walls, and CM and VLDL remnants are produced. VLDL and VLDL remnants penetrate the arterial intima and are taken up by macrophages directly without undergoing oxidative modification. Moreover, LPL on the endothelial surface or within the arterial intima degrades TGs in VLDL and VLDL remnants, with liberation of free fatty acids and monoacylglycerols, both of which cause tissue toxicity and precipitate in local inflammation. Thus, serum TGs are associated with low-grade inflammation and accelerated foam cell formation via macrophage uptake directly at the arterial wall, followed by the development of atherosclerosis.” In addition, high TG levels are associated with increased small-density LDL particles, which could be more atherogenic than larger LDL particles because of their increased susceptibility to oxidation 25) .
Previous studies have reported that non-fasting TG is a superior predictor for risk of CVD events over fasting TG 26) , because the elevated levels of non-fasting TG is considered to reflect elevated remnant lipoprotein levels 26, 27) . In addition, according to the previous study, the obvious superiority of non-fasting TG over fasting TG was observed in TG measured in 2 to ＜4 hours after meal 26) . Because TG levels progressively decrease with longer periods of fasting in a person and because a predictive ability of non-fasting TG for ischemic CVD events has not been well investigated according to fasting hours in Asian population, further investigation about the usefulness of non-fasting TG is needed among Asian population in the future.
The present study has several limitations. First, single TG measurement at the baseline survey might have underestimated the association between TG and ischemic CVD because of regression dilution bias. The longitudinal trend for CVD risk factors and information about medications after the baseline survey were also not investigated, although hypertriglyceridemia is associated with the future onset of metabolic syndrome components, such as hypertension and diabetes 28, 29) . Second, because the blood samples were collected after an overnight fast in the Suita study, the risk of non-fasting TG was not assessed. Third, serum lipoprotein (a), which showed to be a strong risk factor for CAD in a previous study, was not measured 21) . Fourth, ischemic stroke in the present study included cerebral infarction other than atherothrombotic cerebral infarction.
Higher serum TG level was significantly associated with the increased risk for the incidence of ischemic CVD after the adjustment for CVD risk factors, including LDL-C, in Japanese general population with TG ＜400 mg/dL. Because the association between TG and the risk for ischemic CVD and CAD was more evident in those with LDL-C ＜140 mg/dL, TG should be well controlled in individuals with higher TG level whose LDL-C is not evidently high. Furthermore, because the CMI could be the most powerful predictive factor for future ischemic CVD among TG, TG/HDL-C, and the CMI, TG control combined with HDL-C elevation and WC reduction, in addition to LDL-C control, could be effective for the prevention of ischemic CVD.
The authors would like to thank the members of the Suita Medical Foundation and the Suita City Health Center, the paramedical staff of the NCVC, and Satuki-Junyukai, the society of the participants of the Suita study.
None.
This study was supported by the Intramural Research Fund of the NCVC (27-4-3), by a Grant-in-Aid for Scientific Research C 19K10676 from the Japan Society for the Promotion of Science, and by a grant–in–aid from the Ministry of Health, Labor and Welfare, Health and Labor Sciences research grants, Japan (Research on Health Services: H17–Kenkou–007; Comprehensive Research on Cardiovascular Disease and Life–Related Disease: H18–Junkankitou [Seishuu]–Ippan–012; H19–Junkankitou [Seishuu]–Ippan–012; H20–Junkankitou [Seishuu]–Ippan–013; H23–Junkankitou [Seishuu]–Ippan–005; H26-Junkankitou [Seisaku]-Ippan-001 and H29–Junkankitou–Ippan–003 and 20FA1002). These sponsors had no involvements in study design, in analyses and interpretation of data, in the writing of the manuscript, and in the decision to submit the article for publication.
