DIABETICMedicine DOI: 10.1111/dme.12430
Research: Metabolism Smoking and lipid-related indices in patients with diabetes mellitus I. Wakabayashi Department of Environmental and Preventive Medicine, Hyogo College of Medicine, Nishinomiya, Hyogo, Japan Accepted 4 March 2014
Abstract Aims Lipid-related indices, including the ratio of LDL cholesterol to HDL cholesterol, the ratio of triglycerides to HDL cholesterol and lipid accumulation product, are known to be good discriminators for cardiovascular disease. The aim of this study was to clarify the relationships between smoking and the lipid indices in patients with diabetes. Methods Subjects were those who had been diagnosed as having diabetes mellitus at annual health check-ups at their places of work (n = 2563). The subjects were divided into three groups of non-smokers, light smokers (≤ 20 cigarettes/ day) and heavy smokers (> 20 cigarettes/day). The relationships between smoking and the lipid indices were investigated.
Both in all subjects and in the subjects without a habit of alcohol drinking, the LDL cholesterol:HDL cholesterol ratio and the log-transformed triglyerides:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking, and the log-transformed lipid accumulation product was significantly higher in heavy smokers than in non-smokers. In the non-alcohol drinking subjects, the odds ratios of heavy smokers vs. non-smokers for high LDL cholesterol:HDL cholesterol ratio [2.32 (95% CI 1.40–3.84)], for high triglycerides:HDL cholesterol ratio [1.69 (95% CI 1.06–2.69)] and for high lipid accumulation product [1.65 (95% CI 1.02–2.67)] were significantly higher than a reference level of 1.00. The associations between smoking and the lipid indices were weaker in alcohol drinkers than in non-drinkers.
Results
Conclusions In patients with diabetes, the levels of lipid-related indices were higher in smokers than in non-smokers, and cardiometabolic disorders, reflected by high lipid indices, are thought to be involved in the proneness of smokers to develop atherosclerotic cardiovascular disease.
Diabet. Med. 31, 868–878 (2014)
Introduction Cigarette smoking, a major health hazard, is a well-known risk factor for atherosclerotic cerebro- and cardiovascular diseases and has been demonstrated to increase inflammation, thrombosis and oxidation of LDL by previous clinical and experimental studies [1]. In addition, smoking influences blood lipid profile: smokers have significantly higher triglyceride levels and lower HDL cholesterol levels [2]. Thus, alteration in the lipid profile is also thought to promote atherosclerosis in smokers. Smoking cessation has been shown to improve HDL cholesterol, total HDL and large HDL particle concentrations, despite weight gain, but not to improve LDL cholesterol and total LDL particle concentrations [3]. Smokers have been shown to have higher HDL
Correspondence to: Ichiro Wakabayashi. E-mail: [email protected]
868
phospholipids, lower hepatic lipase activity, lower phospholipid transfer protein activity, higher cholesteryl ester transfer protein mass, and lower endogenous cholesteryl ester transfer protein activity [4]. Thus, smoking induces changes in many steps of the reverse cholesterol transport system, which might be involved in the atherogenicity of smoking. Patients with diabetes mellitus are prone to suffer from cardiovascular disease, which is a leading cause of death for the patients [5]. Higher triglyceride and lower HDL cholesterol levels and an increase in atherogenic small dense LDL particles have been shown in patients with diabetes compared with persons without diabetes [6,7]. Therefore, cigarette smoking is expected to further accelerate atherosclerotic progression in patients with diabetes, in part, through exacerbation of the lipid profile. The ratio of LDL cholesterol to HDL cholesterol is a classical atherogenic index for cardiovascular disease [8]. In
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
What’s new? • In addition to the ratio of LDL cholesterol to HDL cholesterol that is a classical atherogenic index, the lipid accumulation index and the ratio of triglycerides to HDL cholesterol have recently been proposed to be good discriminators for cardiovascular disease. • Smoking is a major risk factor for cardiovascular disease, and the harmful influence of smoking on cardiovascular health is, in part, explained by its effects on lipid profile. • This study demonstrated that smoking was positively associated with the above lipid-related indices in patients with diabetes independent of age, regular exercise and drug therapy for diabetes. addition, there are more recently proposed lipid-related indices, such as the ratio of triglycerides to HDL cholesterol and the lipid accumulation product, determined by using waist circumference and levels of triglycerides. The triglycerides:HDL cholesterol ratio has been demonstrated to be better for predicting the risk of myocardial infarction than the LDL cholesterol:HDL cholesterol ratio [9,10] and has been proposed to reflect small dense LDL particles [11]. The lipid accumulation product has been shown to be a better discriminator for diabetes than BMI [12,13] and to be associated with risk of cardiovascular disease [12,14]. However, it remains to be determined whether and how these lipid indices are influenced by smoking in patients with diabetes. The purpose of this study was therefore to clarify the relationships between smoking and lipid-related indices in patients with diabetes. Alcohol drinking is known to potently influence the levels of the blood lipids, including HDL cholesterol, LDL cholesterol and triglycerides [15–17], and has been reported to modify lipid indices as well [18– 20], and smokers are generally more inclined to drink than non-smokers [21]. Accordingly, considering the possibility of confounding by alcohol drinking, the relationships between smoking and lipid indices were investigated both in all subjects and in subjects of different drinker categories.
Subjects and methods Subjects
A cross-sectional study was performed using a local population-based database. The subjects in the original database of the health check-up were male workers aged from 35 to 70 years (n = 37 693) who had received periodic health examinations at their places of work in Yamagata Prefecture in Japan. All of the subjects were of Japanese origin. Subjects who were receiving treatment for any illness were requested to state the names of the diseases in a questionnaire at the
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
health check-up. History of anti-diabetic drug therapy, including insulin injection, was asked in the questionnaire. Subjects with diabetes (n = 2563) were extracted from the database according to the definition of diabetes given below. Subjects with diabetes were defined as those showing high HbA1c levels (≥ 47.5 mmol/mol [6.5%]), according to the recent criteria for diagnosis of diabetes by the American Diabetes Association [22], and/or having a current history of drug therapy for diabetes. Those receiving treatment for dyslipidaemia [n = 458 (15.2%)] were excluded from the study. This study was approved by the Ethics Committee of Yamagata University School of Medicine. Histories of alcohol consumption, cigarette smoking, regular exercise (almost every day, exercising for 30 min or longer per day) and illness were also surveyed by questionnaires. The subjects were divided into three groups by average cigarette consumption (non-smokers; light smokers, 20 or fewer cigarettes per day; heavy smokers, more than 20 cigarettes per day). Average alcohol consumption of each subject per week was reported on the questionnaires. Frequency of habitual alcohol drinking was asked in the questionnaire as ‘How frequently do you drink alcohol?’. Frequency of weekly alcohol drinking was categorized as ‘every day’ (regular drinkers), ‘sometimes’ (occasional drinkers) and ‘never’ (non-drinkers).
Measurements
Height and body weight were measured with the subjects wearing light clothes at the health check-up. BMI was calculated as weight in kilograms divided by the square of height in metres. Waist circumference was measured at the navel level according to the recommendation of the definition of the Japanese Committee for the Diagnostic Criteria of Metabolic Syndrome [23]. Fasting blood was sampled from each subject in the morning, and the serum triglyceride, HDL cholesterol and LDL cholesterol levels were measured by enzymatic methods using commercial kits: Pureauto S TG-N, Cholestest N-HDL and Cholestest LDL (Sekisui Medical Co. Ltd, Tokyo, Japan), respectively. The coefficients of variation for the reproducibility of each measurement were ≤ 3% for triglycerides, ≤ 5% for HDL cholesterol and ≤ 5% for LDL cholesterol. Measurement of LDL cholesterol by the method used in the present study has been shown to be satisfactory for subjects with hypertriglyceridaemia as well as for non-hypertriglyceridaemic subjects [24]. Cut-off values that are often used for high triglycerides and low HDL cholesterol are 150 mg/dl (1.69 mmol/l) and 40 mg/dl (1.03 mmol/l), respectively, according to the clinical definition of the metabolic syndrome for men by the National Cholesterol Education Program (NCEP) Adult Treatment Panel III (ATP III) [25]. The cut-off value for high LDL cholesterol that is generally used in Japan is 140 mg/dl (3.62 mmol/l) according to the Japan Atherosclerosis Society Guidelines for Prevention of Atherosclerotic Cardiovascular Diseases 2012.
869
DIABETICMedicine
Smoking and lipid-related indices in patients with diabetes Wakabayashi
As there have been no confirmed cut-off values for high ratio of LDL cholesterol (mg/dl) to HDL cholesterol (mg/dl) and high ratio of triglycerides (mg/dl) to HDL cholesterol (mg/ dl), these cut-off values were calculated to be 3.5 and 3.75, respectively, by using the above cut-off values for low HDL cholesterol, high LDL cholesterol and high triglycerides. The lipid accumulation product was determined by using triglyceride level and waist circumference as follows [12]: lipid accumulation product = triglycerides (mmol/l) 9 [waist circumference (cm) – 65]. The cut-off value of lipid accumulation product used was 37.2 [26]. Because results for total cholesterol were not included in the database used, non-HDL cholesterol was therefore estimated by calculation of total cholesterol using Friedewald’s formula [27] as non-HDL cholesterol (mg/dl) = LDL cholesterol (mg/dl) + 0.2 9 triglycerides (mg/dl) when triglycerides level is less than 400 mg/dl (4.52 mmol/l). Subjects showing triglyceride levels of 400 mg/dl or higher [n = 195 (7.6%)] were excluded in the analysis of non-HDL cholesterol. The cut-off value for high non-HDL cholesterol used in this study was 170 mg/dl (4.40 mmol/l) according to the Japan Atherosclerosis Society guidelines, and the cut-of value for high ratio of non-HDL cholesterol (mg/dl) to HDL cholesterol (mg/dl) (non-HDL cholesterol:HDL cholesterol ratio) was calculated to be 4.25 by using the cut-off values for low HDL cholesterol and high non-HDL cholesterol. HbA1c was measured by the National Glycohemoglobin Standardization Program (NGSP)-approved technique using the latex cohesion method with a commercial kit (Determiner HbA1c; Kyowa Medex, Tokyo, Japan). The coefficient of variation for reproducibility of HbA1c measurement was ≤ 5%. As the standards of HbA1c used for measurement are different in the NGSP method and the Japan Diabetes Society (JDS) method, the HbA1c values were calibrated by using a formula proposed by the Japan Diabetes Society [28]: HbA1c (NGSP) (%) = 1.02 9 HbA1c (JDS) (%) + 0.25(%). The HbA1c values are presented using both International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) units (mmol/mol) and Diabetes Control and Complications Trial (DCCT) units (%).
Statistical analysis
Statistical analyses were performed using a computer software program (SPSS version 16.0 J for Windows; SPSS Inc., Chicago, IL, USA). The percentages of drinkers, subjects exercising regularly and subjects receiving drug therapy for diabetes were compared between each pair of groups using the v2-test for independence. In univariate analysis, the means of each variable were compared among the groups by using analysis of variance (ANOVA) followed by Scheffe’s F-test as a post-hoc test. In multivariate analysis, the mean levels of each variable were compared by using analysis of covariance (ANCOVA) followed by Student’s t-test after Bonferroni correction. As triglyceride, triglycerides:HDL cholesterol ratio and lipid accumulation product levels did
870
not show normal distributions, these parameters were compared among the groups non-parametrically by using the Kruskal–Wallis test, followed by the Steel–Dwass test as a post-hoc test in univariate analysis (ANOVA), or were used after log-transformation in multivariate analysis (ANCOVA). In logistic regression analysis, the adjusted odds ratios for high LDL cholesterol:HDL cholesterol ratio, high triglycerides:HDL cholesterol ratio, high lipid accumulation product, high non-HDL cholesterol or high non-HDL cholesterol:HDL cholesterol ratio were calculated. Age, alcohol drinking, regular exercise and drug therapy for diabetes were used as other explanatory variables or covariates in multivariate analyses. A three-level description (non, occasional and regular drinkers) was used for alcohol drinking as an explanatory variable. In analyses of variables other than waist circumference and lipid accumulation product, BMI was also added to the explanatory variables and covariates. Probability (P) values less than 0.05 were defined as significant.
Results Profiles of the subject groups
Table 1 shows profiles of the subject groups divided by status of smoking. Age tended to be younger with an increase in the amount of smoking. BMI and waist circumference were slightly but significantly higher and larger, respectively, in heavy smokers than in non-smokers. The proportion of subjects doing exercise regularly and that of subjects receiving drug treatment for diabetes tended to be lower with an increase in the amount of smoking. The levels of HbA1c and triglycerides tended to be higher with an increase in the amount of smoking. HDL cholesterol level tended to be lower with an increase in the amount of smoking, while there was no significant difference in the level of LDL cholesterol among non-, light and heavy smokers. LDL cholesterol:HDL cholesterol ratio and triglycerides:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking. The lipid accumulation product was significantly higher in heavy smokers than in non-smokers, while there was no significant difference in the lipid accumulation product between non-smokers and light smokers.
Comparison of mean levels of variables determining lipid-related indices among non-smokers, light smokers and heavy smokers in each drinker group and in all subjects
Figure 1 shows the mean levels of each variable determining the lipid indices after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of variables other than waist circumference. In all of the drinker groups, as well as in the overall subjects, HDL cholesterol level tended to be lower with an increase in the amount of smoking (Fig. 1a), while
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 1 Characteristics of the groups of non-smokers, light smokers and heavy smokers in subjects with diabetes
Number Age (years) BMI (kg/m2) Waist circumference (cm) Alcohol drinkers (%) Occasional drinkers (%) Regular drinkers (%) Regular exercise (%) Therapy for diabetes (%) HbA1c [mmol/mol (%)] Triglycerides (mg/dl) (mmol/l) HDL cholesterol (mg/dl) (mmol/l) LDL cholesterol (mg/dl) (mmol/l) LDL cholesterol:HDL cholesterol ratio Triglycerides:HDL cholesterol ratio Lipid accumulation product
Non-smokers
Light smokers
Heavy smokers
Overall subjects
1231 54.7 6.8 25.58 3.92 88.4 9.8 73.5 36.5 37.0 14.9 61.5 58.2 16.7 (7.48 1.52) 138 (87, 223) 1.56 (0.98, 2.52) 53.6 14.0 1.39 0.36 117.7 31.2 3.04 0.81 2.34 0.86 2.62 (1.56, 4.76) 36.8 (18.5, 64.2)
850 52.8 7.3† 25.30 4.11 87.9 10.2 75.9 35.2 40.7 11.4* 54.2† 60.4 17.5 (7.67 1.60)* 146 (95, 223) 1.65 (1.07, 2.52) 51.3 13.9† 1.33 0.36† 120.3 32.6 3.11 0.84 2.52 0.96† 2.99 (1.70, 5.24)* 37.4 (19.5, 65.3)
482 51.5 7.4† 26.11 4.30* 90.6 10.6† 76.0 26.8 49.2 9.1† 48.3† 62.1 17.2 (7.83 1.58)† 163 (102, 262)† 1.84 (1.15, 2.96)† 49.6 13.6† 1.28 0.35† 121.3 34.9 3.14 0.90 2.64 1.07† 3.44 (1.92, 6.19)† 46.8 (24.2, 81.3)†
2563 53.4 7.2 25.59 4.07 88.7 10.1 74.7 34.2 40.5 12.6 56.6 59.6 17.1 (7.61 1.57) 146 (92, 228) 1.65 (1.04, 2.57) 52.1 14.0 1.35 0.36 119.2 32.4 3.08 0.84 2.46 0.94 2.88 (1.64, 5.15) 38.4 (19.9, 67.1)
Shown are number of subjects in each group, percentage of subjects with alcohol drinking, regular exercise or drug therapy for diabetes, mean with standard deviation of each variable, and median with 25 and 75 percentile values of each variable. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
there were no significant differences in LDL cholesterol level among non-smokers, light smokers and heavy smokers (Fig. 1b). Significant differences in HDL cholesterol from non-smokers were found in the heavy smokers of the non-drinker group, regular drinker group and overall subjects and in light smokers of the non-drinker group and overall subjects (Fig. 1a). In the non-drinker group and overall subjects, log-transformed triglycerides tended to be higher with an increase in the amount of smoking, although a significant difference was found only between non-smokers and heavy smokers in the non-drinker group (Fig. 1c). There was no significant difference in log-transformed triglycerides among non-, light and heavy smokers of the occasional and regular drinker groups (Fig. 1c). In the regular drinker group and overall subjects, waist circumference was significantly smaller in light smokers than in non-smokers, while in the occasional drinker group, waist circumference was significantly larger in heavy smokers than in non-smokers (Fig. 1d).
to be higher with an increase in the amount of smoking (Fig. 2a). In the non-drinker group and overall subjects, there were significant differences in LDL cholesterol:HDL cholesterol ratio between light smokers and non-smokers and between heavy smokers and non-smokers (Fig. 2a). In the non-drinker group and overall subjects, log-transformed triglycerides:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking, while no significant difference was found in smokers and non-smokers of the occasional and regular drinker groups (Fig. 2b). In the non-drinker group and overall subjects, the log-transformed lipid accumulation product level was significantly higher in heavy smokers than in non-smokers, while this difference was not found in the occasional and regular drinker groups (Fig. 2c). In the regular drinker group, the log-transformed lipid accumulation product level was significantly lower in light smokers than in non-smokers (Fig. 2c).
Adjusted odds ratios for high LDL cholesterol:HDL cholesterol Comparison of mean levels of lipid-related indices among non-smokers, light smokers and heavy smokers in each
ratio, triglycerides:HDL cholesterol ratio or lipid accumulation product of light or heavy smokers vs. non-smokers in each
drinker group and in overall subjects
drinker group and in overall subjects
Figure 2 shows the mean levels of each lipid index after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of variables other than the lipid accumulation product. In all of the drinker groups as well as in overall subjects, the LDL cholesterol:HDL cholesterol ratio tended
Table 2 shows the odds ratios for a high lipid-related index of light or heavy smokers vs. non-smokers after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the odds ratios for high LDL cholesterol:HDL cholesterol ratio and high triglycerides:HDL cholesterol ratio. In all of the drinker
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
871
Smoking and lipid-related indices in patients with diabetes Wakabayashi
DIABETICMedicine
(a)
(b)
Non-smokers Light smokers Heavy smokers
60
135
1.5
3.4
† † 1.3
50
†
1.2
†
45
LDL cholesterol (mg/dl)
1.4
3.2
120 3
115 110
2.8 105
1.1 40
125
2.6
100
Non
Occasional
Regular
Overall
Non
Occasional
Regular
Overall
Drinker category
Drinker category (c)
(d) 2.3
94
*
† 92
Waist circumference (cm)
2.25
Log [triglycerides (mg/dl)]
LDL cholesterol (mmol/l)
†
55
HDL cholesterol (mmol/l)
HDL cholesterol (mg/dl)
130
2.2
2.15
2.1
2.05
2
90
*
88
† 86 84 82 80
Non
Occasional
Regular
Overall
Drinker category
Non
Occasional
Regular
Overall
Drinker category
FIGURE 1 Comparison of mean levels of variables determining lipid-related indices [(a) HDL cholesterol, (b) LDL cholesterol, (c) log-transformed triglycerides, (d) waist circumference] among non-smokers, light smokers and heavy smokers in overall subjects and subjects in each drinker group. The means of each variable were calculated after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of variables other than waist circumference. Alcohol drinking was also used as a covariate for analysis of overall subjects. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
groups, as well as in overall subjects, the odds ratios for high LDL cholesterol:HDL cholesterol ratio in heavy smokers vs. non-smokers were significantly higher than a reference level of 1.00. The odds ratio for high LDL cholesterol:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking. In the non-drinker group, the odds ratios of heavy smokers vs. non-smokers for high triglycerides:HDL cholesterol ratio and high lipid accumulation product were also significantly higher than the reference level. In the regular drinker group, the odds ratio of light smokers vs. non-smokers for the high lipid accumulation product was significantly lower than the reference level.
872
Comparison of mean levels of non-HDL cholesterol and non-HDL cholesterol:HDL cholesterol ratio in non-smokers, light smokers and heavy smokers and adjusted odds ratios for high non-HDL cholesterol or high non-HDL cholesterol: HDL cholesterol ratio of light or heavy smokers vs. non-smokers in each drinker group and in overall subjects
The odds ratio vs. non-smokers for high non-HDL cholesterol was significantly higher than a reference level of 1.00 only in heavy smokers of the non-drinker group (Table 3). Similarly, the mean non-HDL cholesterol level of the non-drinker group was significantly higher in heavy smokers
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
LDL cholesterol:HDL cholesterol ratio
†
(b) 0.7
Non-smokers Light smokers Heavy smokers
†
† 3
†
†
2.5
2
1.5
Non
Occasional
Regular
Log(triglycerides:HDL cholesterol ratio)
(a) 3.5
0.65 0.6 0.55
†
0.5 0.45 0.4 0.35 0.3
Overall
*
Non
Drinker category
Occasional
Regular
Overall
Drinker category
(c) 1.8
Log(lipid accumulation product)
† 1.7
* 1.6
*
1.5
1.4
1.3
Non
Occasional
Regular
Overall
Drinker category
FIGURE 2 Comparison of mean levels of lipid-related indices [(a) LDL cholesterol:HDL cholesterol ratio, (b) log-transformed triglycerides:HDL cholesterol ratio, (c) log-transformed lipid accumulation product] among non-smokers, light smokers and heavy smokers in overall subjects and subjects in each drinker group. The means of each variable were calculated after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of LDL cholesterol:HDL cholesterol ratio and log-transformed triglycerides:HDL cholesterol ratio. Alcohol drinking was also used as a covariate for analysis of overall subjects. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
than in non-smokers, while no significant difference was found between light or heavy smokers and non-smokers in the occasional and regular drinker groups (Fig. 3a). In all of the drinker categories, the odds ratios of light and heavy smokers vs. non-smokers for high non-HDL cholesterol: HDL cholesterol ratio were significantly higher than the reference level (Table 3). In non-drinkers and overall subjects, mean non-HDL cholesterol:HDL cholesterol ratio was significantly higher in light and heavy smokers than in
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
non-smokers. A similar trend was found in occasional and regular drinkers, although there was no significant difference in the non-HDL cholesterol:HDL cholesterol ratio of light or heavy smokers and that of non-smokers (Fig. 3b).
Discussion This study, for the first time, demonstrates that smoking is associated with lipid-related indices, including the LDL
873
Smoking and lipid-related indices in patients with diabetes Wakabayashi
DIABETICMedicine
Table 2 Odds ratios for high LDL cholesterol:HDL cholesterol ratio, high triglycerides:HDL cholesterol ratio or high lipid accumulation product of the light and heavy smoker groups vs. the non-smoker group in overall subjects and subjects in each drinker group Non-smokers Odds ratios for high LDL cholesterol:HDL cholesterol ratio Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall subjects 1.00 Odds ratios for high triglycerides/HDL cholesterol ratio Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall subjects 1.00 Odds ratios for high lipid accumulation product Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall subjects 1.00
Light smokers
Heavy smokers
1.48 1.94 1.30 1.61
(0.95–2.31) (1.26–2.97)† (0.70–2.39) (1.23–2.12)†
2.32 2.27 1.96 2.10
(1.40–3.84)† (1.35–3.84)† (1.02–3.76)* (1.53–2.87)†
1.33 1.04 1.04 1.09
(0.91–1.93) (0.77–1.43) (0.76–1.42) (0.90–1.32)
1.69 0.92 1.32 1.27
(1.06–2.69)* (0.60–1.40) (0.93–1.87) (1.01–1.60)*
1.08 1.00 0.72 0.89
(0.74–1.56) (0.73–1.35) (0.54–0.97)* (0.74–1.07)
1.65 1.13 1.02 1.19
(1.02–2.67)* (0.74–1.72) (0.73–1.42) (0.95–1.49)
Adjusted odds ratios with their 95% confidence intervals indicated in the parentheses are shown. Adjusted odds ratios for high LDL cholesterol:/HDL cholesterol ratio, high triglycerides:HDL cholesterol ratio or high lipid accumulation product of each smoker group vs. non-smokers were calculated using age, regular exercise and therapy for diabetes as other explanatory variables. BMI was also used as an explanatory variable for calculation of adjusted odds ratios for high LDL cholesterol:HDL cholesterol ratio and high triglycerides:HDL cholesterol ratio. Alcohol drinking was also used as an explanatory variable for calculation of odds ratios in overall subjects. Symbols denote significantly higher or lower odds ratios compared with a reference level of 1.00 (*P < 0.05; †P < 0.01).
(b) 170
Non-smokers Light smokers Heavy smokers
4.3
165 4.2 160 4.1 155
4 3.9
150
3.8 145 3.7 140
Non
Occasional
Regular
Overall
Drinker category
Non-HDL cholesterol (mmol/l)
Non-HDL cholesterol (mg/dl)
*
Non-HDL cholesterol:HDL cholesterol ratio
(a)
4.5
† 4
† 3.5
† †
3
2.5
2 Non
Occasional
Regular
Overall
Drinker category
FIGURE 3 Comparison of mean levels of non-HDL cholesterol (a) and non-HDL cholesterol:HDL cholesterol ratio (b) among non-smokers, light smokers and heavy smokers in overall subjects and subjects in each drinker group. The means of each variable were calculated after adjustment for age, BMI, history of regular exercise and history of drug therapy for diabetes. Alcohol drinking was also used as a covariate for analysis of overall subjects. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
cholesterol:HDL cholesterol ratio, the triglycerides:HDL cholesterol ratio and the lipid accumulation product, in patients with diabetes. Both in covariance analysis (Fig. 2) and in logistic regression analysis (Table 2), the three lipid-related indices in non-drinkers were positively associated with heavy drinking. In covariance analysis, the LDL cholesterol:HDL cholesterol ratio and the triglycerides:HDL
874
cholesterol ratio were also significantly higher in light smokers than in non-smokers and tended to be higher with an increase in the amount of smoking in non-drinkers (Fig. 2). Thus, there was a dose–response relationship between smoking and risk of high lipid indices. In addition to a proneness to dyslipidaemia induced by diabetes [6,7], the results of the present study indicated that exacerbation of
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 3 Odds ratios vs. non-smokers for high non-HDL cholesterol or high non-HDL cholesterol:HDL cholesterol ratio in overall subjects and subjects in each drinker group Non-smokers Odds ratios for high non-HDL cholesterol Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall 1.00 Odds ratios for high non-HDL cholesterol:HDL cholesterol ratio Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall 1.00
Light smokers
Heavy smokers
1.39 1.06 1.06 1.14
(0.93–2.08) (0.75–1.49) (0.74–1.53) (0.92–1.40)
2.11 0.96 0.96 1.19
(1.30–3.44)† (0.61–1.52) (0.63–1.47) (0.92–1.53)
2.11 1.70 1.80 1.85
(1.36–3.26)† (1.14–2.55)* (1.03–3.12)* (1.43–2.40)†
3.47 1.86 2.26 2.35
(2.08–5.81)† (1.12–3.09)* (1.23–4.15)† (1.73–3.19)†
Adjusted odds ratios with their 95% confidence intervals indicated in the parentheses are shown. Adjusted odds ratios for high non-HDL cholesterol and high non-HDL cholesterol:HDL cholesterol ratio were calculated using age, BMI, regular exercise and therapy for diabetes as other explanatory variables. Alcohol drinking was also used as an explanatory variable for calculation of odds ratios in overall subjects. Symbols denote significantly higher odds ratios compared with a reference level of 1.00 (*P < 0.05; †P < 0.01).
the blood lipid profile by smoking is involved in acceleration of atherosclerotic progression, resulting in an increase in cardiovascular risk in smokers with diabetes. The above-mentioned findings on the relationships between smoking and the three lipid-related indices were not changed when those receiving anti-dyslipidaemic drug therapy were included in subjects for analysis (data not shown). The three lipid-related indices used in this study consist of two variables out of HDL cholesterol, LDL cholesterol, triglycerides and waist circumference. Thus, the influence of smoking on the lipid indices depends on the relationship of smoking with each of the above variables. HDL cholesterol levels tended to be lower with an increase in the amount of smoking, while LDL cholesterol levels were not different among non-smokers, light smokers and heavy smokers. Thus, the inverse association between HDL cholesterol and smoking resulted in a positive association between smoking and the LDL cholesterol:HDL cholesterol ratio. In non-drinkers and overall subjects, log-transformed triglyceride levels tended to be higher with an increase in the amount of smoking (Fig. 1c). Thus, in addition to the inverse association between HDL cholesterol and smoking, the positive association between smoking and triglycerides may contribute to the positive association between smoking and the triglycerides:HDL cholesterol ratio. In non-drinkers and overall subjects, waist circumference was not significantly different in the smoker groups and the non-smoker group, except for a slightly smaller waist circumference in light smokers than in non-smokers of overall subjects (Fig. 1d). Nonetheless, there was a positive association between heavy smoking and lipid accumulation product (Fig. 2c), which may be attributable to the positive association of smoking with triglycerides. In summary, the HDL-lowering and triglycerides-elevating effects of smoking are mainly involved in the modification of levels of the lipid-related indices in smokers with diabetes. Because the lipid indices have been shown to be considerably modified by alcohol drinking [15–17], the relationª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
ships between smoking and lipid indices were investigated in subjects of the different alcohol drinking categories. The odds ratio of heavy smokers vs. non-smokers for high LDL cholesterol:HDL cholesterol ratio was significantly higher in all of the drinker groups and in the overall subjects. The odds ratio and mean level of LDL cholesterol:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking in all of the drinker groups and in overall subjects. The differences in the LDL cholesterol:HDL cholesterol ratio between light smokers and non-smokers and between heavy smokers and non-smokers were greater in the non-drinker group than in the occasional and regular drinker groups. There were positive associations of heavy smoking with the triglycerides:HDL cholesterol ratio and lipid accumulation product in non-drinkers, but not in occasional and regular drinkers. Thus, the positive associations between smoking and the lipid-related indices are weaker in drinkers than in non-drinkers. This finding is plausible because HDL cholesterol is known to be higher in drinkers than in non-drinkers [15,16] and the HDL cholesterol-lowering effect of smoking is thought to be reduced in drinkers. In fact, the differences in HDL cholesterol levels between light smokers and non-smokers and between heavy smokers and non-smokers were smaller in the occasional and regular drinker groups than in the non-drinker group (Fig. 1a). In addition, the level of log-transformed triglycerides of the non-drinker group was significantly higher in heavy smokers than in non-smokers, whereas this difference was not found in the drinker groups (Fig. 1c). Therefore, in the non-drinker group, higher triglycerides in heavy smokers than in non-smokers contribute to a higher triglycerides:HDL cholesterol ratio and higher lipid accumulation product in heavy smokers than in non-smokers. Although no significant association was found between smoking and LDL cholesterol, there was a positive association between heavy smoking and non-HDL cholesterol, which was found in non-drinkers but not in drinkers. In
875
DIABETICMedicine
Smoking and lipid-related indices in patients with diabetes Wakabayashi
addition, there was a positive association between heavy smoking and non-HDL cholesterol:HDL cholesterol ratio in drinkers and non-drinkers, and the association was stronger in non-drinkers than in drinkers. Therefore, smoking is positively associated with non-HDL cholesterol and the non-HDL cholesterol:HDL cholesterol ratio, which are useful markers to predict adverse cardiovascular outcomes [29–31], and these associations are also weakened by alcohol drinking. The confounding effects of alcohol are plausible as smoking and alcohol drinking diversely influence HDL cholesterol and non-HDL cholesterol: HDL cholesterol is higher in drinkers than in non-drinkers and is lower in smokers than in non-smokers, while non-HDL cholesterol is lower in drinkers than in non-drinkers and is higher in smokers than in non-smokers [32]. In a preliminary study analysing all subjects (n = 37 693) included in the original database, HbA1c level was slightly but significantly higher in light and heavy smokers than in non-smokers, and the odds ratios of light and heavy smokers vs. non-smokers for hyperglycaemia and diabetes were slightly but significantly higher than a reference level of 1.00. The above positive associations between smoking and glycaemic status tended to be stronger in heavy smokers than in light smokers (data not shown). These results suggest that smoking causes insulin resistance. Therefore, there is a possibility that insulin resistance is involved in the influence of smoking on lipids as the triglycerides:HDL cholesterol ratio and the lipid accumulation product have been shown to be associated with insulin resistance [33–35]. Although the exact mechanism(s) underlying the effects of smoking on the lipid profile remains to be elucidated, lipoprotein lipase activity has been demonstrated to be reduced in smokers compared with that in non-smokers [36]. A previous meta-analysis study for general populations showed that the change in the level of LDL cholesterol (1.7%) caused by smoking was not significant and was smaller than the changes in the levels of triglycerides (9.1%) and HDL cholesterol (–5.7%) by smoking [2], agreeing with the finding of the present study that LDL cholesterol level was not different among non-, light and heavy smokers in patients with diabetes. The ratio of large to small LDL particles has been shown to be lower in smokers than in non-smokers [37]. This also agrees with the finding of the present study that the triglycerides:HDL cholesterol ratio, which is known to reflect small dense LDL particles [11], was higher in smokers than in non-smokers. Therefore, the relationships between smoking and the lipid profile are thought to be similar in persons with and persons without diabetes. Needless to say, not smoking is strongly recommended for patients with diabetes in order to prevent cardiovascular complications. In addition to a variety of pathophysiological mechanisms explaining smoking-induced acceleration of atherosclerosis [1], an increase in cardiometabolic risk attributable to exacerbation of the blood lipid profiles by
876
smoking is involved in the increased risk of cardiovascular disease in patients with diabetes. Adiposity is associated with both blood lipid levels and smoking status and possibly confounds the relationships between smoking and lipid-related indices. Therefore, BMI was included in explanatory variables and covariates in the multivariate analyses for LDL cholesterol:HDL cholesterol ratio, triglycerides:HDL cholesterol ratio, non-HDL cholesterol and non-HDL cholesterol:HDL cholesterol ratio. The lipid accumulation product is calculated by using levels of waist circumference and triglycerides. Both BMI and waist circumference are anthropometry variables related to adiposity and were strongly associated with each other (Pearson’s correlation coefficient between them: 0.902, P < 0.001). In order to avoid using two adiposity-related variables in each analysis, BMI was not used as an explanatory variable or covariate in the multivariate analyses for the lipid accumulation product. Inclusion of BMI in explanatory variables did not greatly influence the odds ratios for high lipid-related indices and did not change the tendencies of the relationships between smoking and lipid indices (data not shown). There are limitations of this study. The subjects were divided into smoking groups by their current status of smoking, but no information on ex-smokers was available. In this study, diabetes was diagnosed by the level of HbA1c and a history of drug therapy for diabetes, as assessed in the questionnaire. Thus, there is a possibility of an informational bias regarding the diagnosis of diabetes. In addition, the type of diabetes was not identified for each subject, and both patients with Type 1 diabetes and patients with Type 2 diabetes were included in the subjects. However, for most of the subjects, the type is expected to be Type 2 diabetes because the prevalence of Type 2 diabetes is speculated to be > 100 times higher than that of Type 1 diabetes in middle-aged Japanese men according to statistics [38]. Because of the small size of the population, further analysis for drinker subgroups classified by amount of daily alcohol intake was not performed. Although age, BMI, regular exercise and drug therapy for diabetes were adjusted in multivariate analyses, there are other possible confounders; for example, diet, nutrition and socio-economic status, for the relationships between smoking and lipid-related indices. As the subjects of this study were Japanese men, there are possibilities of gender-specific difference and ethnic and/or racial difference in the relationships between smoking and lipid indices. In fact, the triglycerides:HDL cholesterol ratio has been reported to fail to predict insulin resistance in African-American women [39]. Moreover, gender differences in the relationships between smoking and lipids were reported: The mean increases in total cholesterol and LDL cholesterol in smokers, compared with non-smokers, were greater by twice and almost four times, respectively, in women than in men [40]. Because the original database used in this study included results of health check-up examinations, data for only common examinations, such as
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
triglycerides, HDL cholesterol and LDL cholesterol for evaluation of blood lipids, were available in this study. There is, to my knowledge, no evidence indicating that the three lipid-related indices for this investigation are superior to other lipid examinations including LDL particle size, HDL particle size, apolipoprotein B (ApoB) and ApoAII for discriminating cardiovascular risk. Therefore, further studies are needed to investigate relationships between smoking and other blood lipid markers in patients with diabetes. Finally, as this study is cross-sectional in its design, further studies are needed to clarify causal relationships between smoking and cardiometabolic disorders including poor lipid profiles. In conclusion, there are positive associations between smoking and lipid-related indices in patients with diabetes, which are weaker in drinkers than in non-drinkers. As lipid indices are known to be good predictors of cardiovascular disease, smoking is thought to accelerate atherosclerotic progression through increasing cardiometabolic risk in patients with diabetes.
Funding sources
None.
Competing interests
None declared.
References 1 Ambrose JA, Barua RS. The pathophysiology of cigarette smoking and cardiovascular disease: an update. J Am Coll Cardiol 2004; 43: 1731–1737. 2 Craig WY, Palomaki GE, Haddow JE. Cigarette smoking and serum lipid and lipoprotein concentrations: an analysis of published data. Br Med J 1989; 298: 784–788. 3 Gepner AD, Piper ME, Johnson HM, Fiore MC, Baker TB, Stein JH. Effects of smoking and smoking cessation on lipids and lipoproteins: outcomes from a randomized clinical trial. Am Heart J 2011; 161: 145–151. 4 Zaratin ACM, Quint~ao ECR, Sposito AC, Nunes VS, Lottenberg AM, Morton RE et al. Smoking prevents the intravascular remodeling of high-density lipoprotein particles: implications for reverse cholesterol transport. Metabolism 2004; 53: 858–862. 5 The Emerging Risk Factors Collaboration, Seshasai SR, Kaptoge S, Thompson A, Di Angelantonio E, Gao P, Sarwar N et al. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med 2011; 364: 829–841. 6 Mooradian AD. Dyslipidemia in type 2 diabetes mellitus. Nat Clin Pract Endocrinol Metab 2009; 5: 150–159. 7 Dunn FL. Management of dyslipidemia in people with type 2 diabetes mellitus. Rev Endocr Metab Disord 2010; 11: 41–51. 8 Kannel WB. Lipids, diabetes, and coronary heart disease: insights from the Framingham Study. Am Heart J 1985; 110: 1100–1107. 9 Gaziano JM, Hennekens CH, O’Donnell CJ, Breslow JL, Buring JE. Fasting triglycerides, high-density lipoprotein, and risk of myocardial infarction. Circulation 1997; 96: 2520–2525. 10 Jeppesen J, Hein HO, Suadicani P, Gyntelberg F. Relation of high TG-low HDL cholesterol and LDL cholesterol to the
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
11
12
13
14
15
16
17 18
19
20
21
22 23 24
25
26
27
28
incidence of ischemic heart disease: An 8-year follow-up in the Copenhagen Male Study. Arterioscler Thromb Vasc Biol 1997; 17: 1114–1120. Dobi asov a M, Frohlich J. The plasma parameter log (TG/HDL-C) as an atherogenic index: correlation with lipoprotein particle size and esterification rate in apoB-lipoprotein-depleted plasma (FERHDL). Clin Biochem 2001; 34: 583–588. Kahn HS. The ‘lipid accumulation product’ performs better than the body mass index for recognizing cardiovascular risk: a population-based comparison. BMC Cardiovasc Disord 2005; 5: 26. Kahn HS. The lipid accumulation product is better than BMI for identifying diabetes: a population-based comparison. Diabetes Care 2006; 29: 151–153. Wehr E, Pilz S, Boehm BO, M€ arz W, Obermayer-Pietsch B. The lipid accumulation product is associated with increased mortality in normal weight postmenopausal women. Obesity (Silver Spring) 2011; 19: 1873–1880. Castelli WP, Doyle JT, Gordon T, Hames CG, Hjortland MC, Hulley SB et al. Alcohol and blood lipids. The Cooperative Lipoprotein Phenotyping Study. Lancet 1977; 2: 153–155. Langer RD, Criqui MH, Reed DM. Lipoproteins and blood pressure as biological pathways for effect of moderate alcohol consumption on coronary heart disease. Circulation 1992; 85: 910– 915. Van de Wiel A. The effect of alcohol on postprandial and fasting triglycerides. Int J Vasc Med 2012; 2012: 862504. Wakabayashi I. Increased body mass index modifies associations between alcohol intake and blood cholesterol profile. Eur J Clin Invest 2012; 42: 179–185. Wakabayashi I. Inverse association between triglycerides-to-HDL-cholesterol ratio and alcohol drinking in middle-aged Japanese men. J Stud Alcohol Drugs 2012; 73: 998–1004. Wakabayashi I. Relationship between alcohol intake and lipid accumulation product in middle-aged men. Alcohol Alcohol 2013; 48: 535–542. Veenstra J, Schenkel JA, van Erp-Baart AM, Brants HA, Hulshof KF, Kistemaker C et al. Alcohol consumption in relation to food intake and smoking habits in the Dutch National Food Consumption Survey. Eur J Clin Nutr 1993; 47: 482–489. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2010; 33: S62–69. Anonymous. Metabolic syndrome—definition and diagnostic criteria in Japan. J Jpn Soc Int Med 2005; 94: 794–809 (in Japanese). Miida T, Nishimura K, Okamura T, Hirayama S, Ohmura H, Yoshida H et al. A multicenter study on the precision and accuracy of homogeneous assays for LDL-cholesterol: comparison with a beta-quantification method using fresh serum obtained from non-diseased and diseased subjects. Atherosclerosis 2012; 225: 208–215. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). J Am Med Assoc 2001; 285: 2486–2497. Wakabayashi I, Daimon T. A strong association between lipid accumulation product and diabetes in Japanese women and men. J Atheroscler Thromb 2014; 21: 282–288. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972; 18: 499–502. Kashiwagi A, Kasuga M, Araki E, Oka Y, Hanafusa T, Ito H et al., Committee on the Standardization of Diabetes Mellitus-Related Laboratory Testing of Japan Diabetes Society (JDS). International
877
DIABETICMedicine
29
30
31
32
33
34
Smoking and lipid-related indices in patients with diabetes Wakabayashi
clinical harmonization of glycated hemoglobin in Japan: From Japan Diabetes Society to National Glycohemoglobin Standardization Program values. Diabetol Int 2012; 3: 8–10. Frost PH, Havel RJ. Rationale for use of non-high-density lipoprotein cholesterol rather than low-density lipoprotein cholesterol as a tool for lipoprotein cholesterol screening and assessment of risk and therapy. Am J Cardiol 1998; 81: 26B–31B. Sniderman A, McQueen M, Contois J, Williams K, Furberg CD. Why is non-high-density lipoprotein cholesterol a better marker of the risk of vascular disease than low-density lipoprotein cholesterol? J Clin Lipidol 2010; 4: 152–155. Eliasson B, Cederholm J, Eeg-Olofsson K. Svensson AM, Zethelius B, Gudbj€ ornsdottir S; National Diabetes Register. Clinical usefulness of different lipid measures for prediction of coronary heart disease in type 2 diabetes: a report from the Swedish National Diabetes Register. Diabetes Care 2011; 34: 2095–2100. Wakabayashi I, Groschner K. Age-dependent associations of smoking and drinking with non-high-density lipoprotein cholesterol. Metabolism 2010; 59: 1074–1081. McLaughlin T, Reaven G, Abbasi F, Lamendola C, Saad M, Waters D et al. Is there a simple way to identify insulin-resistant individuals at increased risk of cardiovascular disease? Am J Cardiol 2005; 96: 399–404. Kim JS, Kang HT, Shim JY, Lee HR. The association between the triglyceride to high-density lipoprotein cholesterol ratio with insulin resistance (HOMA-IR) in the general Korean population: based on the National Health and Nutrition
878
35
36
37
38
39
40
Examination Survey in 2007–2009. Diabetes Res Clin Pract 2012; 97: 132–138. Taverna MJ, Martınez-Larrad MT, Frechtel GD, Serrano-Rıos M. Lipid accumulation product: a powerful marker of metabolic syndrome in healthy population. Eur J Endocrinol 2011; 164: 559– 567. Freeman DJ, Caslake MJ, Griffin BA, Hinnie J, Tan CE, Watson TD et al. The effect of smoking on post-heparin lipoprotein and hepatic lipase, cholesteryl ester transfer protein and lecithin: cholesterol acyl transferase activities in human plasma. Eur J Clin Invest 1998; 28: 584–591. Griffin BA, Freeman DJ, Tait GW, Thomson J, Caslake MJ, Packard CJ et al. Role of plasma triglyceride in the regulation of plasma low density lipoprotein (LDL) subfractions: relative contribution of small, dense LDL to coronary heart disease risk. Atherosclerosis 1994; 106: 241–253. International Diabetes Federation. Diabetes Atlas. 4th edn. 2009Available at www.diabetesatlas.org. Last accessed on 25 July 2013. Sumner AE, Harman JL, Buxbaum SG, Miller BV 3rd, Tambay AV, Wyatt SB et al. The triglyceride/high-density lipoprotein cholesterol ratio fails to predict insulin resistance in African-American women: an analysis of Jackson Heart Study. Metab Syndr Relat Disord 2010; 8: 511–514. Cullen P, Schulte H, Assmann G. Smoking, lipoproteins and coronary heart disease risk. Data from the M€ unster Heart Study (PROCAM). Eur Heart J 1998; 19: 1632–1641.
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research: Metabolism Smoking and lipid-related indices in patients with diabetes mellitus I. Wakabayashi Department of Environmental and Preventive Medicine, Hyogo College of Medicine, Nishinomiya, Hyogo, Japan Accepted 4 March 2014
Abstract Aims Lipid-related indices, including the ratio of LDL cholesterol to HDL cholesterol, the ratio of triglycerides to HDL cholesterol and lipid accumulation product, are known to be good discriminators for cardiovascular disease. The aim of this study was to clarify the relationships between smoking and the lipid indices in patients with diabetes. Methods Subjects were those who had been diagnosed as having diabetes mellitus at annual health check-ups at their places of work (n = 2563). The subjects were divided into three groups of non-smokers, light smokers (≤ 20 cigarettes/ day) and heavy smokers (> 20 cigarettes/day). The relationships between smoking and the lipid indices were investigated.
Both in all subjects and in the subjects without a habit of alcohol drinking, the LDL cholesterol:HDL cholesterol ratio and the log-transformed triglyerides:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking, and the log-transformed lipid accumulation product was significantly higher in heavy smokers than in non-smokers. In the non-alcohol drinking subjects, the odds ratios of heavy smokers vs. non-smokers for high LDL cholesterol:HDL cholesterol ratio [2.32 (95% CI 1.40–3.84)], for high triglycerides:HDL cholesterol ratio [1.69 (95% CI 1.06–2.69)] and for high lipid accumulation product [1.65 (95% CI 1.02–2.67)] were significantly higher than a reference level of 1.00. The associations between smoking and the lipid indices were weaker in alcohol drinkers than in non-drinkers.
Results
Conclusions In patients with diabetes, the levels of lipid-related indices were higher in smokers than in non-smokers, and cardiometabolic disorders, reflected by high lipid indices, are thought to be involved in the proneness of smokers to develop atherosclerotic cardiovascular disease.
Diabet. Med. 31, 868–878 (2014)
Introduction Cigarette smoking, a major health hazard, is a well-known risk factor for atherosclerotic cerebro- and cardiovascular diseases and has been demonstrated to increase inflammation, thrombosis and oxidation of LDL by previous clinical and experimental studies [1]. In addition, smoking influences blood lipid profile: smokers have significantly higher triglyceride levels and lower HDL cholesterol levels [2]. Thus, alteration in the lipid profile is also thought to promote atherosclerosis in smokers. Smoking cessation has been shown to improve HDL cholesterol, total HDL and large HDL particle concentrations, despite weight gain, but not to improve LDL cholesterol and total LDL particle concentrations [3]. Smokers have been shown to have higher HDL
Correspondence to: Ichiro Wakabayashi. E-mail: [email protected]
868
phospholipids, lower hepatic lipase activity, lower phospholipid transfer protein activity, higher cholesteryl ester transfer protein mass, and lower endogenous cholesteryl ester transfer protein activity [4]. Thus, smoking induces changes in many steps of the reverse cholesterol transport system, which might be involved in the atherogenicity of smoking. Patients with diabetes mellitus are prone to suffer from cardiovascular disease, which is a leading cause of death for the patients [5]. Higher triglyceride and lower HDL cholesterol levels and an increase in atherogenic small dense LDL particles have been shown in patients with diabetes compared with persons without diabetes [6,7]. Therefore, cigarette smoking is expected to further accelerate atherosclerotic progression in patients with diabetes, in part, through exacerbation of the lipid profile. The ratio of LDL cholesterol to HDL cholesterol is a classical atherogenic index for cardiovascular disease [8]. In
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
What’s new? • In addition to the ratio of LDL cholesterol to HDL cholesterol that is a classical atherogenic index, the lipid accumulation index and the ratio of triglycerides to HDL cholesterol have recently been proposed to be good discriminators for cardiovascular disease. • Smoking is a major risk factor for cardiovascular disease, and the harmful influence of smoking on cardiovascular health is, in part, explained by its effects on lipid profile. • This study demonstrated that smoking was positively associated with the above lipid-related indices in patients with diabetes independent of age, regular exercise and drug therapy for diabetes. addition, there are more recently proposed lipid-related indices, such as the ratio of triglycerides to HDL cholesterol and the lipid accumulation product, determined by using waist circumference and levels of triglycerides. The triglycerides:HDL cholesterol ratio has been demonstrated to be better for predicting the risk of myocardial infarction than the LDL cholesterol:HDL cholesterol ratio [9,10] and has been proposed to reflect small dense LDL particles [11]. The lipid accumulation product has been shown to be a better discriminator for diabetes than BMI [12,13] and to be associated with risk of cardiovascular disease [12,14]. However, it remains to be determined whether and how these lipid indices are influenced by smoking in patients with diabetes. The purpose of this study was therefore to clarify the relationships between smoking and lipid-related indices in patients with diabetes. Alcohol drinking is known to potently influence the levels of the blood lipids, including HDL cholesterol, LDL cholesterol and triglycerides [15–17], and has been reported to modify lipid indices as well [18– 20], and smokers are generally more inclined to drink than non-smokers [21]. Accordingly, considering the possibility of confounding by alcohol drinking, the relationships between smoking and lipid indices were investigated both in all subjects and in subjects of different drinker categories.
Subjects and methods Subjects
A cross-sectional study was performed using a local population-based database. The subjects in the original database of the health check-up were male workers aged from 35 to 70 years (n = 37 693) who had received periodic health examinations at their places of work in Yamagata Prefecture in Japan. All of the subjects were of Japanese origin. Subjects who were receiving treatment for any illness were requested to state the names of the diseases in a questionnaire at the
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
health check-up. History of anti-diabetic drug therapy, including insulin injection, was asked in the questionnaire. Subjects with diabetes (n = 2563) were extracted from the database according to the definition of diabetes given below. Subjects with diabetes were defined as those showing high HbA1c levels (≥ 47.5 mmol/mol [6.5%]), according to the recent criteria for diagnosis of diabetes by the American Diabetes Association [22], and/or having a current history of drug therapy for diabetes. Those receiving treatment for dyslipidaemia [n = 458 (15.2%)] were excluded from the study. This study was approved by the Ethics Committee of Yamagata University School of Medicine. Histories of alcohol consumption, cigarette smoking, regular exercise (almost every day, exercising for 30 min or longer per day) and illness were also surveyed by questionnaires. The subjects were divided into three groups by average cigarette consumption (non-smokers; light smokers, 20 or fewer cigarettes per day; heavy smokers, more than 20 cigarettes per day). Average alcohol consumption of each subject per week was reported on the questionnaires. Frequency of habitual alcohol drinking was asked in the questionnaire as ‘How frequently do you drink alcohol?’. Frequency of weekly alcohol drinking was categorized as ‘every day’ (regular drinkers), ‘sometimes’ (occasional drinkers) and ‘never’ (non-drinkers).
Measurements
Height and body weight were measured with the subjects wearing light clothes at the health check-up. BMI was calculated as weight in kilograms divided by the square of height in metres. Waist circumference was measured at the navel level according to the recommendation of the definition of the Japanese Committee for the Diagnostic Criteria of Metabolic Syndrome [23]. Fasting blood was sampled from each subject in the morning, and the serum triglyceride, HDL cholesterol and LDL cholesterol levels were measured by enzymatic methods using commercial kits: Pureauto S TG-N, Cholestest N-HDL and Cholestest LDL (Sekisui Medical Co. Ltd, Tokyo, Japan), respectively. The coefficients of variation for the reproducibility of each measurement were ≤ 3% for triglycerides, ≤ 5% for HDL cholesterol and ≤ 5% for LDL cholesterol. Measurement of LDL cholesterol by the method used in the present study has been shown to be satisfactory for subjects with hypertriglyceridaemia as well as for non-hypertriglyceridaemic subjects [24]. Cut-off values that are often used for high triglycerides and low HDL cholesterol are 150 mg/dl (1.69 mmol/l) and 40 mg/dl (1.03 mmol/l), respectively, according to the clinical definition of the metabolic syndrome for men by the National Cholesterol Education Program (NCEP) Adult Treatment Panel III (ATP III) [25]. The cut-off value for high LDL cholesterol that is generally used in Japan is 140 mg/dl (3.62 mmol/l) according to the Japan Atherosclerosis Society Guidelines for Prevention of Atherosclerotic Cardiovascular Diseases 2012.
869
DIABETICMedicine
Smoking and lipid-related indices in patients with diabetes Wakabayashi
As there have been no confirmed cut-off values for high ratio of LDL cholesterol (mg/dl) to HDL cholesterol (mg/dl) and high ratio of triglycerides (mg/dl) to HDL cholesterol (mg/ dl), these cut-off values were calculated to be 3.5 and 3.75, respectively, by using the above cut-off values for low HDL cholesterol, high LDL cholesterol and high triglycerides. The lipid accumulation product was determined by using triglyceride level and waist circumference as follows [12]: lipid accumulation product = triglycerides (mmol/l) 9 [waist circumference (cm) – 65]. The cut-off value of lipid accumulation product used was 37.2 [26]. Because results for total cholesterol were not included in the database used, non-HDL cholesterol was therefore estimated by calculation of total cholesterol using Friedewald’s formula [27] as non-HDL cholesterol (mg/dl) = LDL cholesterol (mg/dl) + 0.2 9 triglycerides (mg/dl) when triglycerides level is less than 400 mg/dl (4.52 mmol/l). Subjects showing triglyceride levels of 400 mg/dl or higher [n = 195 (7.6%)] were excluded in the analysis of non-HDL cholesterol. The cut-off value for high non-HDL cholesterol used in this study was 170 mg/dl (4.40 mmol/l) according to the Japan Atherosclerosis Society guidelines, and the cut-of value for high ratio of non-HDL cholesterol (mg/dl) to HDL cholesterol (mg/dl) (non-HDL cholesterol:HDL cholesterol ratio) was calculated to be 4.25 by using the cut-off values for low HDL cholesterol and high non-HDL cholesterol. HbA1c was measured by the National Glycohemoglobin Standardization Program (NGSP)-approved technique using the latex cohesion method with a commercial kit (Determiner HbA1c; Kyowa Medex, Tokyo, Japan). The coefficient of variation for reproducibility of HbA1c measurement was ≤ 5%. As the standards of HbA1c used for measurement are different in the NGSP method and the Japan Diabetes Society (JDS) method, the HbA1c values were calibrated by using a formula proposed by the Japan Diabetes Society [28]: HbA1c (NGSP) (%) = 1.02 9 HbA1c (JDS) (%) + 0.25(%). The HbA1c values are presented using both International Federation of Clinical Chemistry and Laboratory Medicine (IFCC) units (mmol/mol) and Diabetes Control and Complications Trial (DCCT) units (%).
Statistical analysis
Statistical analyses were performed using a computer software program (SPSS version 16.0 J for Windows; SPSS Inc., Chicago, IL, USA). The percentages of drinkers, subjects exercising regularly and subjects receiving drug therapy for diabetes were compared between each pair of groups using the v2-test for independence. In univariate analysis, the means of each variable were compared among the groups by using analysis of variance (ANOVA) followed by Scheffe’s F-test as a post-hoc test. In multivariate analysis, the mean levels of each variable were compared by using analysis of covariance (ANCOVA) followed by Student’s t-test after Bonferroni correction. As triglyceride, triglycerides:HDL cholesterol ratio and lipid accumulation product levels did
870
not show normal distributions, these parameters were compared among the groups non-parametrically by using the Kruskal–Wallis test, followed by the Steel–Dwass test as a post-hoc test in univariate analysis (ANOVA), or were used after log-transformation in multivariate analysis (ANCOVA). In logistic regression analysis, the adjusted odds ratios for high LDL cholesterol:HDL cholesterol ratio, high triglycerides:HDL cholesterol ratio, high lipid accumulation product, high non-HDL cholesterol or high non-HDL cholesterol:HDL cholesterol ratio were calculated. Age, alcohol drinking, regular exercise and drug therapy for diabetes were used as other explanatory variables or covariates in multivariate analyses. A three-level description (non, occasional and regular drinkers) was used for alcohol drinking as an explanatory variable. In analyses of variables other than waist circumference and lipid accumulation product, BMI was also added to the explanatory variables and covariates. Probability (P) values less than 0.05 were defined as significant.
Results Profiles of the subject groups
Table 1 shows profiles of the subject groups divided by status of smoking. Age tended to be younger with an increase in the amount of smoking. BMI and waist circumference were slightly but significantly higher and larger, respectively, in heavy smokers than in non-smokers. The proportion of subjects doing exercise regularly and that of subjects receiving drug treatment for diabetes tended to be lower with an increase in the amount of smoking. The levels of HbA1c and triglycerides tended to be higher with an increase in the amount of smoking. HDL cholesterol level tended to be lower with an increase in the amount of smoking, while there was no significant difference in the level of LDL cholesterol among non-, light and heavy smokers. LDL cholesterol:HDL cholesterol ratio and triglycerides:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking. The lipid accumulation product was significantly higher in heavy smokers than in non-smokers, while there was no significant difference in the lipid accumulation product between non-smokers and light smokers.
Comparison of mean levels of variables determining lipid-related indices among non-smokers, light smokers and heavy smokers in each drinker group and in all subjects
Figure 1 shows the mean levels of each variable determining the lipid indices after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of variables other than waist circumference. In all of the drinker groups, as well as in the overall subjects, HDL cholesterol level tended to be lower with an increase in the amount of smoking (Fig. 1a), while
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 1 Characteristics of the groups of non-smokers, light smokers and heavy smokers in subjects with diabetes
Number Age (years) BMI (kg/m2) Waist circumference (cm) Alcohol drinkers (%) Occasional drinkers (%) Regular drinkers (%) Regular exercise (%) Therapy for diabetes (%) HbA1c [mmol/mol (%)] Triglycerides (mg/dl) (mmol/l) HDL cholesterol (mg/dl) (mmol/l) LDL cholesterol (mg/dl) (mmol/l) LDL cholesterol:HDL cholesterol ratio Triglycerides:HDL cholesterol ratio Lipid accumulation product
Non-smokers
Light smokers
Heavy smokers
Overall subjects
1231 54.7 6.8 25.58 3.92 88.4 9.8 73.5 36.5 37.0 14.9 61.5 58.2 16.7 (7.48 1.52) 138 (87, 223) 1.56 (0.98, 2.52) 53.6 14.0 1.39 0.36 117.7 31.2 3.04 0.81 2.34 0.86 2.62 (1.56, 4.76) 36.8 (18.5, 64.2)
850 52.8 7.3† 25.30 4.11 87.9 10.2 75.9 35.2 40.7 11.4* 54.2† 60.4 17.5 (7.67 1.60)* 146 (95, 223) 1.65 (1.07, 2.52) 51.3 13.9† 1.33 0.36† 120.3 32.6 3.11 0.84 2.52 0.96† 2.99 (1.70, 5.24)* 37.4 (19.5, 65.3)
482 51.5 7.4† 26.11 4.30* 90.6 10.6† 76.0 26.8 49.2 9.1† 48.3† 62.1 17.2 (7.83 1.58)† 163 (102, 262)† 1.84 (1.15, 2.96)† 49.6 13.6† 1.28 0.35† 121.3 34.9 3.14 0.90 2.64 1.07† 3.44 (1.92, 6.19)† 46.8 (24.2, 81.3)†
2563 53.4 7.2 25.59 4.07 88.7 10.1 74.7 34.2 40.5 12.6 56.6 59.6 17.1 (7.61 1.57) 146 (92, 228) 1.65 (1.04, 2.57) 52.1 14.0 1.35 0.36 119.2 32.4 3.08 0.84 2.46 0.94 2.88 (1.64, 5.15) 38.4 (19.9, 67.1)
Shown are number of subjects in each group, percentage of subjects with alcohol drinking, regular exercise or drug therapy for diabetes, mean with standard deviation of each variable, and median with 25 and 75 percentile values of each variable. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
there were no significant differences in LDL cholesterol level among non-smokers, light smokers and heavy smokers (Fig. 1b). Significant differences in HDL cholesterol from non-smokers were found in the heavy smokers of the non-drinker group, regular drinker group and overall subjects and in light smokers of the non-drinker group and overall subjects (Fig. 1a). In the non-drinker group and overall subjects, log-transformed triglycerides tended to be higher with an increase in the amount of smoking, although a significant difference was found only between non-smokers and heavy smokers in the non-drinker group (Fig. 1c). There was no significant difference in log-transformed triglycerides among non-, light and heavy smokers of the occasional and regular drinker groups (Fig. 1c). In the regular drinker group and overall subjects, waist circumference was significantly smaller in light smokers than in non-smokers, while in the occasional drinker group, waist circumference was significantly larger in heavy smokers than in non-smokers (Fig. 1d).
to be higher with an increase in the amount of smoking (Fig. 2a). In the non-drinker group and overall subjects, there were significant differences in LDL cholesterol:HDL cholesterol ratio between light smokers and non-smokers and between heavy smokers and non-smokers (Fig. 2a). In the non-drinker group and overall subjects, log-transformed triglycerides:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking, while no significant difference was found in smokers and non-smokers of the occasional and regular drinker groups (Fig. 2b). In the non-drinker group and overall subjects, the log-transformed lipid accumulation product level was significantly higher in heavy smokers than in non-smokers, while this difference was not found in the occasional and regular drinker groups (Fig. 2c). In the regular drinker group, the log-transformed lipid accumulation product level was significantly lower in light smokers than in non-smokers (Fig. 2c).
Adjusted odds ratios for high LDL cholesterol:HDL cholesterol Comparison of mean levels of lipid-related indices among non-smokers, light smokers and heavy smokers in each
ratio, triglycerides:HDL cholesterol ratio or lipid accumulation product of light or heavy smokers vs. non-smokers in each
drinker group and in overall subjects
drinker group and in overall subjects
Figure 2 shows the mean levels of each lipid index after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of variables other than the lipid accumulation product. In all of the drinker groups as well as in overall subjects, the LDL cholesterol:HDL cholesterol ratio tended
Table 2 shows the odds ratios for a high lipid-related index of light or heavy smokers vs. non-smokers after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the odds ratios for high LDL cholesterol:HDL cholesterol ratio and high triglycerides:HDL cholesterol ratio. In all of the drinker
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
871
Smoking and lipid-related indices in patients with diabetes Wakabayashi
DIABETICMedicine
(a)
(b)
Non-smokers Light smokers Heavy smokers
60
135
1.5
3.4
† † 1.3
50
†
1.2
†
45
LDL cholesterol (mg/dl)
1.4
3.2
120 3
115 110
2.8 105
1.1 40
125
2.6
100
Non
Occasional
Regular
Overall
Non
Occasional
Regular
Overall
Drinker category
Drinker category (c)
(d) 2.3
94
*
† 92
Waist circumference (cm)
2.25
Log [triglycerides (mg/dl)]
LDL cholesterol (mmol/l)
†
55
HDL cholesterol (mmol/l)
HDL cholesterol (mg/dl)
130
2.2
2.15
2.1
2.05
2
90
*
88
† 86 84 82 80
Non
Occasional
Regular
Overall
Drinker category
Non
Occasional
Regular
Overall
Drinker category
FIGURE 1 Comparison of mean levels of variables determining lipid-related indices [(a) HDL cholesterol, (b) LDL cholesterol, (c) log-transformed triglycerides, (d) waist circumference] among non-smokers, light smokers and heavy smokers in overall subjects and subjects in each drinker group. The means of each variable were calculated after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of variables other than waist circumference. Alcohol drinking was also used as a covariate for analysis of overall subjects. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
groups, as well as in overall subjects, the odds ratios for high LDL cholesterol:HDL cholesterol ratio in heavy smokers vs. non-smokers were significantly higher than a reference level of 1.00. The odds ratio for high LDL cholesterol:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking. In the non-drinker group, the odds ratios of heavy smokers vs. non-smokers for high triglycerides:HDL cholesterol ratio and high lipid accumulation product were also significantly higher than the reference level. In the regular drinker group, the odds ratio of light smokers vs. non-smokers for the high lipid accumulation product was significantly lower than the reference level.
872
Comparison of mean levels of non-HDL cholesterol and non-HDL cholesterol:HDL cholesterol ratio in non-smokers, light smokers and heavy smokers and adjusted odds ratios for high non-HDL cholesterol or high non-HDL cholesterol: HDL cholesterol ratio of light or heavy smokers vs. non-smokers in each drinker group and in overall subjects
The odds ratio vs. non-smokers for high non-HDL cholesterol was significantly higher than a reference level of 1.00 only in heavy smokers of the non-drinker group (Table 3). Similarly, the mean non-HDL cholesterol level of the non-drinker group was significantly higher in heavy smokers
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
LDL cholesterol:HDL cholesterol ratio
†
(b) 0.7
Non-smokers Light smokers Heavy smokers
†
† 3
†
†
2.5
2
1.5
Non
Occasional
Regular
Log(triglycerides:HDL cholesterol ratio)
(a) 3.5
0.65 0.6 0.55
†
0.5 0.45 0.4 0.35 0.3
Overall
*
Non
Drinker category
Occasional
Regular
Overall
Drinker category
(c) 1.8
Log(lipid accumulation product)
† 1.7
* 1.6
*
1.5
1.4
1.3
Non
Occasional
Regular
Overall
Drinker category
FIGURE 2 Comparison of mean levels of lipid-related indices [(a) LDL cholesterol:HDL cholesterol ratio, (b) log-transformed triglycerides:HDL cholesterol ratio, (c) log-transformed lipid accumulation product] among non-smokers, light smokers and heavy smokers in overall subjects and subjects in each drinker group. The means of each variable were calculated after adjustment for age, regular exercise and drug therapy for diabetes. In addition, BMI was adjusted for calculation of the means of LDL cholesterol:HDL cholesterol ratio and log-transformed triglycerides:HDL cholesterol ratio. Alcohol drinking was also used as a covariate for analysis of overall subjects. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
than in non-smokers, while no significant difference was found between light or heavy smokers and non-smokers in the occasional and regular drinker groups (Fig. 3a). In all of the drinker categories, the odds ratios of light and heavy smokers vs. non-smokers for high non-HDL cholesterol: HDL cholesterol ratio were significantly higher than the reference level (Table 3). In non-drinkers and overall subjects, mean non-HDL cholesterol:HDL cholesterol ratio was significantly higher in light and heavy smokers than in
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
non-smokers. A similar trend was found in occasional and regular drinkers, although there was no significant difference in the non-HDL cholesterol:HDL cholesterol ratio of light or heavy smokers and that of non-smokers (Fig. 3b).
Discussion This study, for the first time, demonstrates that smoking is associated with lipid-related indices, including the LDL
873
Smoking and lipid-related indices in patients with diabetes Wakabayashi
DIABETICMedicine
Table 2 Odds ratios for high LDL cholesterol:HDL cholesterol ratio, high triglycerides:HDL cholesterol ratio or high lipid accumulation product of the light and heavy smoker groups vs. the non-smoker group in overall subjects and subjects in each drinker group Non-smokers Odds ratios for high LDL cholesterol:HDL cholesterol ratio Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall subjects 1.00 Odds ratios for high triglycerides/HDL cholesterol ratio Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall subjects 1.00 Odds ratios for high lipid accumulation product Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall subjects 1.00
Light smokers
Heavy smokers
1.48 1.94 1.30 1.61
(0.95–2.31) (1.26–2.97)† (0.70–2.39) (1.23–2.12)†
2.32 2.27 1.96 2.10
(1.40–3.84)† (1.35–3.84)† (1.02–3.76)* (1.53–2.87)†
1.33 1.04 1.04 1.09
(0.91–1.93) (0.77–1.43) (0.76–1.42) (0.90–1.32)
1.69 0.92 1.32 1.27
(1.06–2.69)* (0.60–1.40) (0.93–1.87) (1.01–1.60)*
1.08 1.00 0.72 0.89
(0.74–1.56) (0.73–1.35) (0.54–0.97)* (0.74–1.07)
1.65 1.13 1.02 1.19
(1.02–2.67)* (0.74–1.72) (0.73–1.42) (0.95–1.49)
Adjusted odds ratios with their 95% confidence intervals indicated in the parentheses are shown. Adjusted odds ratios for high LDL cholesterol:/HDL cholesterol ratio, high triglycerides:HDL cholesterol ratio or high lipid accumulation product of each smoker group vs. non-smokers were calculated using age, regular exercise and therapy for diabetes as other explanatory variables. BMI was also used as an explanatory variable for calculation of adjusted odds ratios for high LDL cholesterol:HDL cholesterol ratio and high triglycerides:HDL cholesterol ratio. Alcohol drinking was also used as an explanatory variable for calculation of odds ratios in overall subjects. Symbols denote significantly higher or lower odds ratios compared with a reference level of 1.00 (*P < 0.05; †P < 0.01).
(b) 170
Non-smokers Light smokers Heavy smokers
4.3
165 4.2 160 4.1 155
4 3.9
150
3.8 145 3.7 140
Non
Occasional
Regular
Overall
Drinker category
Non-HDL cholesterol (mmol/l)
Non-HDL cholesterol (mg/dl)
*
Non-HDL cholesterol:HDL cholesterol ratio
(a)
4.5
† 4
† 3.5
† †
3
2.5
2 Non
Occasional
Regular
Overall
Drinker category
FIGURE 3 Comparison of mean levels of non-HDL cholesterol (a) and non-HDL cholesterol:HDL cholesterol ratio (b) among non-smokers, light smokers and heavy smokers in overall subjects and subjects in each drinker group. The means of each variable were calculated after adjustment for age, BMI, history of regular exercise and history of drug therapy for diabetes. Alcohol drinking was also used as a covariate for analysis of overall subjects. Symbols denote significant differences from non-smokers (*P < 0.05; †P < 0.01).
cholesterol:HDL cholesterol ratio, the triglycerides:HDL cholesterol ratio and the lipid accumulation product, in patients with diabetes. Both in covariance analysis (Fig. 2) and in logistic regression analysis (Table 2), the three lipid-related indices in non-drinkers were positively associated with heavy drinking. In covariance analysis, the LDL cholesterol:HDL cholesterol ratio and the triglycerides:HDL
874
cholesterol ratio were also significantly higher in light smokers than in non-smokers and tended to be higher with an increase in the amount of smoking in non-drinkers (Fig. 2). Thus, there was a dose–response relationship between smoking and risk of high lipid indices. In addition to a proneness to dyslipidaemia induced by diabetes [6,7], the results of the present study indicated that exacerbation of
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
DIABETICMedicine
Table 3 Odds ratios vs. non-smokers for high non-HDL cholesterol or high non-HDL cholesterol:HDL cholesterol ratio in overall subjects and subjects in each drinker group Non-smokers Odds ratios for high non-HDL cholesterol Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall 1.00 Odds ratios for high non-HDL cholesterol:HDL cholesterol ratio Non-drinkers 1.00 Occasional drinkers 1.00 Regular drinkers 1.00 Overall 1.00
Light smokers
Heavy smokers
1.39 1.06 1.06 1.14
(0.93–2.08) (0.75–1.49) (0.74–1.53) (0.92–1.40)
2.11 0.96 0.96 1.19
(1.30–3.44)† (0.61–1.52) (0.63–1.47) (0.92–1.53)
2.11 1.70 1.80 1.85
(1.36–3.26)† (1.14–2.55)* (1.03–3.12)* (1.43–2.40)†
3.47 1.86 2.26 2.35
(2.08–5.81)† (1.12–3.09)* (1.23–4.15)† (1.73–3.19)†
Adjusted odds ratios with their 95% confidence intervals indicated in the parentheses are shown. Adjusted odds ratios for high non-HDL cholesterol and high non-HDL cholesterol:HDL cholesterol ratio were calculated using age, BMI, regular exercise and therapy for diabetes as other explanatory variables. Alcohol drinking was also used as an explanatory variable for calculation of odds ratios in overall subjects. Symbols denote significantly higher odds ratios compared with a reference level of 1.00 (*P < 0.05; †P < 0.01).
the blood lipid profile by smoking is involved in acceleration of atherosclerotic progression, resulting in an increase in cardiovascular risk in smokers with diabetes. The above-mentioned findings on the relationships between smoking and the three lipid-related indices were not changed when those receiving anti-dyslipidaemic drug therapy were included in subjects for analysis (data not shown). The three lipid-related indices used in this study consist of two variables out of HDL cholesterol, LDL cholesterol, triglycerides and waist circumference. Thus, the influence of smoking on the lipid indices depends on the relationship of smoking with each of the above variables. HDL cholesterol levels tended to be lower with an increase in the amount of smoking, while LDL cholesterol levels were not different among non-smokers, light smokers and heavy smokers. Thus, the inverse association between HDL cholesterol and smoking resulted in a positive association between smoking and the LDL cholesterol:HDL cholesterol ratio. In non-drinkers and overall subjects, log-transformed triglyceride levels tended to be higher with an increase in the amount of smoking (Fig. 1c). Thus, in addition to the inverse association between HDL cholesterol and smoking, the positive association between smoking and triglycerides may contribute to the positive association between smoking and the triglycerides:HDL cholesterol ratio. In non-drinkers and overall subjects, waist circumference was not significantly different in the smoker groups and the non-smoker group, except for a slightly smaller waist circumference in light smokers than in non-smokers of overall subjects (Fig. 1d). Nonetheless, there was a positive association between heavy smoking and lipid accumulation product (Fig. 2c), which may be attributable to the positive association of smoking with triglycerides. In summary, the HDL-lowering and triglycerides-elevating effects of smoking are mainly involved in the modification of levels of the lipid-related indices in smokers with diabetes. Because the lipid indices have been shown to be considerably modified by alcohol drinking [15–17], the relationª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
ships between smoking and lipid indices were investigated in subjects of the different alcohol drinking categories. The odds ratio of heavy smokers vs. non-smokers for high LDL cholesterol:HDL cholesterol ratio was significantly higher in all of the drinker groups and in the overall subjects. The odds ratio and mean level of LDL cholesterol:HDL cholesterol ratio tended to be higher with an increase in the amount of smoking in all of the drinker groups and in overall subjects. The differences in the LDL cholesterol:HDL cholesterol ratio between light smokers and non-smokers and between heavy smokers and non-smokers were greater in the non-drinker group than in the occasional and regular drinker groups. There were positive associations of heavy smoking with the triglycerides:HDL cholesterol ratio and lipid accumulation product in non-drinkers, but not in occasional and regular drinkers. Thus, the positive associations between smoking and the lipid-related indices are weaker in drinkers than in non-drinkers. This finding is plausible because HDL cholesterol is known to be higher in drinkers than in non-drinkers [15,16] and the HDL cholesterol-lowering effect of smoking is thought to be reduced in drinkers. In fact, the differences in HDL cholesterol levels between light smokers and non-smokers and between heavy smokers and non-smokers were smaller in the occasional and regular drinker groups than in the non-drinker group (Fig. 1a). In addition, the level of log-transformed triglycerides of the non-drinker group was significantly higher in heavy smokers than in non-smokers, whereas this difference was not found in the drinker groups (Fig. 1c). Therefore, in the non-drinker group, higher triglycerides in heavy smokers than in non-smokers contribute to a higher triglycerides:HDL cholesterol ratio and higher lipid accumulation product in heavy smokers than in non-smokers. Although no significant association was found between smoking and LDL cholesterol, there was a positive association between heavy smoking and non-HDL cholesterol, which was found in non-drinkers but not in drinkers. In
875
DIABETICMedicine
Smoking and lipid-related indices in patients with diabetes Wakabayashi
addition, there was a positive association between heavy smoking and non-HDL cholesterol:HDL cholesterol ratio in drinkers and non-drinkers, and the association was stronger in non-drinkers than in drinkers. Therefore, smoking is positively associated with non-HDL cholesterol and the non-HDL cholesterol:HDL cholesterol ratio, which are useful markers to predict adverse cardiovascular outcomes [29–31], and these associations are also weakened by alcohol drinking. The confounding effects of alcohol are plausible as smoking and alcohol drinking diversely influence HDL cholesterol and non-HDL cholesterol: HDL cholesterol is higher in drinkers than in non-drinkers and is lower in smokers than in non-smokers, while non-HDL cholesterol is lower in drinkers than in non-drinkers and is higher in smokers than in non-smokers [32]. In a preliminary study analysing all subjects (n = 37 693) included in the original database, HbA1c level was slightly but significantly higher in light and heavy smokers than in non-smokers, and the odds ratios of light and heavy smokers vs. non-smokers for hyperglycaemia and diabetes were slightly but significantly higher than a reference level of 1.00. The above positive associations between smoking and glycaemic status tended to be stronger in heavy smokers than in light smokers (data not shown). These results suggest that smoking causes insulin resistance. Therefore, there is a possibility that insulin resistance is involved in the influence of smoking on lipids as the triglycerides:HDL cholesterol ratio and the lipid accumulation product have been shown to be associated with insulin resistance [33–35]. Although the exact mechanism(s) underlying the effects of smoking on the lipid profile remains to be elucidated, lipoprotein lipase activity has been demonstrated to be reduced in smokers compared with that in non-smokers [36]. A previous meta-analysis study for general populations showed that the change in the level of LDL cholesterol (1.7%) caused by smoking was not significant and was smaller than the changes in the levels of triglycerides (9.1%) and HDL cholesterol (–5.7%) by smoking [2], agreeing with the finding of the present study that LDL cholesterol level was not different among non-, light and heavy smokers in patients with diabetes. The ratio of large to small LDL particles has been shown to be lower in smokers than in non-smokers [37]. This also agrees with the finding of the present study that the triglycerides:HDL cholesterol ratio, which is known to reflect small dense LDL particles [11], was higher in smokers than in non-smokers. Therefore, the relationships between smoking and the lipid profile are thought to be similar in persons with and persons without diabetes. Needless to say, not smoking is strongly recommended for patients with diabetes in order to prevent cardiovascular complications. In addition to a variety of pathophysiological mechanisms explaining smoking-induced acceleration of atherosclerosis [1], an increase in cardiometabolic risk attributable to exacerbation of the blood lipid profiles by
876
smoking is involved in the increased risk of cardiovascular disease in patients with diabetes. Adiposity is associated with both blood lipid levels and smoking status and possibly confounds the relationships between smoking and lipid-related indices. Therefore, BMI was included in explanatory variables and covariates in the multivariate analyses for LDL cholesterol:HDL cholesterol ratio, triglycerides:HDL cholesterol ratio, non-HDL cholesterol and non-HDL cholesterol:HDL cholesterol ratio. The lipid accumulation product is calculated by using levels of waist circumference and triglycerides. Both BMI and waist circumference are anthropometry variables related to adiposity and were strongly associated with each other (Pearson’s correlation coefficient between them: 0.902, P < 0.001). In order to avoid using two adiposity-related variables in each analysis, BMI was not used as an explanatory variable or covariate in the multivariate analyses for the lipid accumulation product. Inclusion of BMI in explanatory variables did not greatly influence the odds ratios for high lipid-related indices and did not change the tendencies of the relationships between smoking and lipid indices (data not shown). There are limitations of this study. The subjects were divided into smoking groups by their current status of smoking, but no information on ex-smokers was available. In this study, diabetes was diagnosed by the level of HbA1c and a history of drug therapy for diabetes, as assessed in the questionnaire. Thus, there is a possibility of an informational bias regarding the diagnosis of diabetes. In addition, the type of diabetes was not identified for each subject, and both patients with Type 1 diabetes and patients with Type 2 diabetes were included in the subjects. However, for most of the subjects, the type is expected to be Type 2 diabetes because the prevalence of Type 2 diabetes is speculated to be > 100 times higher than that of Type 1 diabetes in middle-aged Japanese men according to statistics [38]. Because of the small size of the population, further analysis for drinker subgroups classified by amount of daily alcohol intake was not performed. Although age, BMI, regular exercise and drug therapy for diabetes were adjusted in multivariate analyses, there are other possible confounders; for example, diet, nutrition and socio-economic status, for the relationships between smoking and lipid-related indices. As the subjects of this study were Japanese men, there are possibilities of gender-specific difference and ethnic and/or racial difference in the relationships between smoking and lipid indices. In fact, the triglycerides:HDL cholesterol ratio has been reported to fail to predict insulin resistance in African-American women [39]. Moreover, gender differences in the relationships between smoking and lipids were reported: The mean increases in total cholesterol and LDL cholesterol in smokers, compared with non-smokers, were greater by twice and almost four times, respectively, in women than in men [40]. Because the original database used in this study included results of health check-up examinations, data for only common examinations, such as
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
Research article
triglycerides, HDL cholesterol and LDL cholesterol for evaluation of blood lipids, were available in this study. There is, to my knowledge, no evidence indicating that the three lipid-related indices for this investigation are superior to other lipid examinations including LDL particle size, HDL particle size, apolipoprotein B (ApoB) and ApoAII for discriminating cardiovascular risk. Therefore, further studies are needed to investigate relationships between smoking and other blood lipid markers in patients with diabetes. Finally, as this study is cross-sectional in its design, further studies are needed to clarify causal relationships between smoking and cardiometabolic disorders including poor lipid profiles. In conclusion, there are positive associations between smoking and lipid-related indices in patients with diabetes, which are weaker in drinkers than in non-drinkers. As lipid indices are known to be good predictors of cardiovascular disease, smoking is thought to accelerate atherosclerotic progression through increasing cardiometabolic risk in patients with diabetes.
Funding sources
None.
Competing interests
None declared.
References 1 Ambrose JA, Barua RS. The pathophysiology of cigarette smoking and cardiovascular disease: an update. J Am Coll Cardiol 2004; 43: 1731–1737. 2 Craig WY, Palomaki GE, Haddow JE. Cigarette smoking and serum lipid and lipoprotein concentrations: an analysis of published data. Br Med J 1989; 298: 784–788. 3 Gepner AD, Piper ME, Johnson HM, Fiore MC, Baker TB, Stein JH. Effects of smoking and smoking cessation on lipids and lipoproteins: outcomes from a randomized clinical trial. Am Heart J 2011; 161: 145–151. 4 Zaratin ACM, Quint~ao ECR, Sposito AC, Nunes VS, Lottenberg AM, Morton RE et al. Smoking prevents the intravascular remodeling of high-density lipoprotein particles: implications for reverse cholesterol transport. Metabolism 2004; 53: 858–862. 5 The Emerging Risk Factors Collaboration, Seshasai SR, Kaptoge S, Thompson A, Di Angelantonio E, Gao P, Sarwar N et al. Diabetes mellitus, fasting glucose, and risk of cause-specific death. N Engl J Med 2011; 364: 829–841. 6 Mooradian AD. Dyslipidemia in type 2 diabetes mellitus. Nat Clin Pract Endocrinol Metab 2009; 5: 150–159. 7 Dunn FL. Management of dyslipidemia in people with type 2 diabetes mellitus. Rev Endocr Metab Disord 2010; 11: 41–51. 8 Kannel WB. Lipids, diabetes, and coronary heart disease: insights from the Framingham Study. Am Heart J 1985; 110: 1100–1107. 9 Gaziano JM, Hennekens CH, O’Donnell CJ, Breslow JL, Buring JE. Fasting triglycerides, high-density lipoprotein, and risk of myocardial infarction. Circulation 1997; 96: 2520–2525. 10 Jeppesen J, Hein HO, Suadicani P, Gyntelberg F. Relation of high TG-low HDL cholesterol and LDL cholesterol to the
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
DIABETICMedicine
11
12
13
14
15
16
17 18
19
20
21
22 23 24
25
26
27
28
incidence of ischemic heart disease: An 8-year follow-up in the Copenhagen Male Study. Arterioscler Thromb Vasc Biol 1997; 17: 1114–1120. Dobi asov a M, Frohlich J. The plasma parameter log (TG/HDL-C) as an atherogenic index: correlation with lipoprotein particle size and esterification rate in apoB-lipoprotein-depleted plasma (FERHDL). Clin Biochem 2001; 34: 583–588. Kahn HS. The ‘lipid accumulation product’ performs better than the body mass index for recognizing cardiovascular risk: a population-based comparison. BMC Cardiovasc Disord 2005; 5: 26. Kahn HS. The lipid accumulation product is better than BMI for identifying diabetes: a population-based comparison. Diabetes Care 2006; 29: 151–153. Wehr E, Pilz S, Boehm BO, M€ arz W, Obermayer-Pietsch B. The lipid accumulation product is associated with increased mortality in normal weight postmenopausal women. Obesity (Silver Spring) 2011; 19: 1873–1880. Castelli WP, Doyle JT, Gordon T, Hames CG, Hjortland MC, Hulley SB et al. Alcohol and blood lipids. The Cooperative Lipoprotein Phenotyping Study. Lancet 1977; 2: 153–155. Langer RD, Criqui MH, Reed DM. Lipoproteins and blood pressure as biological pathways for effect of moderate alcohol consumption on coronary heart disease. Circulation 1992; 85: 910– 915. Van de Wiel A. The effect of alcohol on postprandial and fasting triglycerides. Int J Vasc Med 2012; 2012: 862504. Wakabayashi I. Increased body mass index modifies associations between alcohol intake and blood cholesterol profile. Eur J Clin Invest 2012; 42: 179–185. Wakabayashi I. Inverse association between triglycerides-to-HDL-cholesterol ratio and alcohol drinking in middle-aged Japanese men. J Stud Alcohol Drugs 2012; 73: 998–1004. Wakabayashi I. Relationship between alcohol intake and lipid accumulation product in middle-aged men. Alcohol Alcohol 2013; 48: 535–542. Veenstra J, Schenkel JA, van Erp-Baart AM, Brants HA, Hulshof KF, Kistemaker C et al. Alcohol consumption in relation to food intake and smoking habits in the Dutch National Food Consumption Survey. Eur J Clin Nutr 1993; 47: 482–489. American Diabetes Association. Diagnosis and classification of diabetes mellitus. Diabetes Care 2010; 33: S62–69. Anonymous. Metabolic syndrome—definition and diagnostic criteria in Japan. J Jpn Soc Int Med 2005; 94: 794–809 (in Japanese). Miida T, Nishimura K, Okamura T, Hirayama S, Ohmura H, Yoshida H et al. A multicenter study on the precision and accuracy of homogeneous assays for LDL-cholesterol: comparison with a beta-quantification method using fresh serum obtained from non-diseased and diseased subjects. Atherosclerosis 2012; 225: 208–215. Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Executive summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). J Am Med Assoc 2001; 285: 2486–2497. Wakabayashi I, Daimon T. A strong association between lipid accumulation product and diabetes in Japanese women and men. J Atheroscler Thromb 2014; 21: 282–288. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma, without use of the preparative ultracentrifuge. Clin Chem 1972; 18: 499–502. Kashiwagi A, Kasuga M, Araki E, Oka Y, Hanafusa T, Ito H et al., Committee on the Standardization of Diabetes Mellitus-Related Laboratory Testing of Japan Diabetes Society (JDS). International
877
DIABETICMedicine
29
30
31
32
33
34
Smoking and lipid-related indices in patients with diabetes Wakabayashi
clinical harmonization of glycated hemoglobin in Japan: From Japan Diabetes Society to National Glycohemoglobin Standardization Program values. Diabetol Int 2012; 3: 8–10. Frost PH, Havel RJ. Rationale for use of non-high-density lipoprotein cholesterol rather than low-density lipoprotein cholesterol as a tool for lipoprotein cholesterol screening and assessment of risk and therapy. Am J Cardiol 1998; 81: 26B–31B. Sniderman A, McQueen M, Contois J, Williams K, Furberg CD. Why is non-high-density lipoprotein cholesterol a better marker of the risk of vascular disease than low-density lipoprotein cholesterol? J Clin Lipidol 2010; 4: 152–155. Eliasson B, Cederholm J, Eeg-Olofsson K. Svensson AM, Zethelius B, Gudbj€ ornsdottir S; National Diabetes Register. Clinical usefulness of different lipid measures for prediction of coronary heart disease in type 2 diabetes: a report from the Swedish National Diabetes Register. Diabetes Care 2011; 34: 2095–2100. Wakabayashi I, Groschner K. Age-dependent associations of smoking and drinking with non-high-density lipoprotein cholesterol. Metabolism 2010; 59: 1074–1081. McLaughlin T, Reaven G, Abbasi F, Lamendola C, Saad M, Waters D et al. Is there a simple way to identify insulin-resistant individuals at increased risk of cardiovascular disease? Am J Cardiol 2005; 96: 399–404. Kim JS, Kang HT, Shim JY, Lee HR. The association between the triglyceride to high-density lipoprotein cholesterol ratio with insulin resistance (HOMA-IR) in the general Korean population: based on the National Health and Nutrition
878
35
36
37
38
39
40
Examination Survey in 2007–2009. Diabetes Res Clin Pract 2012; 97: 132–138. Taverna MJ, Martınez-Larrad MT, Frechtel GD, Serrano-Rıos M. Lipid accumulation product: a powerful marker of metabolic syndrome in healthy population. Eur J Endocrinol 2011; 164: 559– 567. Freeman DJ, Caslake MJ, Griffin BA, Hinnie J, Tan CE, Watson TD et al. The effect of smoking on post-heparin lipoprotein and hepatic lipase, cholesteryl ester transfer protein and lecithin: cholesterol acyl transferase activities in human plasma. Eur J Clin Invest 1998; 28: 584–591. Griffin BA, Freeman DJ, Tait GW, Thomson J, Caslake MJ, Packard CJ et al. Role of plasma triglyceride in the regulation of plasma low density lipoprotein (LDL) subfractions: relative contribution of small, dense LDL to coronary heart disease risk. Atherosclerosis 1994; 106: 241–253. International Diabetes Federation. Diabetes Atlas. 4th edn. 2009Available at www.diabetesatlas.org. Last accessed on 25 July 2013. Sumner AE, Harman JL, Buxbaum SG, Miller BV 3rd, Tambay AV, Wyatt SB et al. The triglyceride/high-density lipoprotein cholesterol ratio fails to predict insulin resistance in African-American women: an analysis of Jackson Heart Study. Metab Syndr Relat Disord 2010; 8: 511–514. Cullen P, Schulte H, Assmann G. Smoking, lipoproteins and coronary heart disease risk. Data from the M€ unster Heart Study (PROCAM). Eur Heart J 1998; 19: 1632–1641.
ª 2014 The Author. Diabetic Medicine ª 2014 Diabetes UK
