In the present study, performed in a large population-based cohort of almost 60,000 individuals, we investigated the relationship between smoking and the individual components of metabolic syndrome, and the association between smoking and levels of apolipoproteins and estimated lipoprotein particle size. Such a comprehensive and large-scale analysis has not been performed to date. We demonstrated that in both men and women smoking is associated with a greater prevalence of MetS, irrespective of their BMI. The largest differences between current smokers and non-smokers were observed in the levels of HDL-C and triglycerides, and, to a lesser extent, in waist circumference. While there were no consistent associations between smoking status and either blood pressure or fasting blood glucose levels, there was a dose-dependent relationship between the amount of tobacco smoked and decreased HDL-C levels and increased triglyceride levels. We also found a clear dose-dependent association between the amount of tobacco smoked and reduced ratios of HDL-C/apoA1 and LDL-C/apoB. To our knowledge, we are the first to explore these associations between smoking and levels of apolipoproteins and lipoprotein particle size in such a large cohort of individuals, with rigorously standardized physical and laboratory measurements, while taking into account both sex and BMI levels.
Our analysis revealed that in both men and women the prevalence of MetS was higher in current smokers in each BMI group, than in the non-smokers within that BMI group. Several earlier small-scale studies have reported smoking to be associated with higher prevalence of MetS [24–27]. The positive dose–response relationship between the amount of tobacco smoked and the prevalence of MetS that we observed is also consistent with previous studies [10, 13, 26, 28]. However, when BMI was included in our analysis, the odds ratio for having MetS was higher among normal weight smoking subjects than those with higher BMI (Figure 3). This is probably related to the initial lower risk of subjects in this BMI group. Since previous studies have shown an excess of visceral fat to be a major contributor to metabolic abnormalities, overweight and obesity are known to be highly associated with MetS , with already a high prevalence of MetS observed in the obese non-smokers.
With our approach we have been able to calculate precisely the effects of smoking on the lipid parameters. Our data unequivocally show that despite the fact that obese men and women have a lower mean HDL-C than non-obese, the effects of heavy smoking are similar in all three BMI groups, with a consistent 0.10 to 0.14 mmol/l lower HDL-C for smoking men, and 0.15 to 0.21 mmol/l lower HDL-C in smoking women, in all three BMI groups (Table 4). The fact that we found current smoking to be mainly associated with lower levels of HDL-C, higher levels of triglycerides and larger waist circumference than the non-smoking status is consistent with earlier cross-sectional studies [28, 30]. This observation of a dose-dependent relationship between the daily amount of tobacco smoked and lower HDL-C and higher triglycerides confirms the results of previous reports [13, 31–33]. In our study, the magnitude of the effects of tobacco usage on HDL-C varied between 0.04 for light smoking in men, and 0.21 mmol/l for heavy smoking in normal weight women. The study by Chen et al., comprising 1,164 men, reported a similar dose–response relationship with the largest effect on HDL-C and triglycerides seen in those who smoked more than 40 cigarettes per day . Ishizaka et al. also reported a dose–response association between the number of cigarettes per day and prevalence of MetS in a cohort of 5,033 individuals, although they did not examine the influence of the amount of tobacco smoked on the individual MetS components . A recent review summarized the effects of smoking cessation on HDL-C levels: within a few weeks after stopping smoking, HDL-C levels start to increase, resulting in an overall increase of 0.2 mmol/l . Taken together, these and our data support the causal relationship between smoking and low HDL-C levels.
There are indications that current smoking is associated with increased abdominal obesity . In our study, although current smokers had a greater waist circumference than non-smokers, these differences were rather small. We also observed a consistent increase in the waist circumference with an increase in tobacco smoked in normal weight and overweight men, as well as in normal weight and obese women. Larger effects were especially seen among obese women, where the increase in waist circumference was 2.2 cm for moderate smokers and 6.4 cm for heavy smokers. One of the possible mechanisms that might explain these observations is a direct effect of smoking on cortisol production [12, 36]. Indeed, it was demonstrated more than three decades ago that smokers have higher fasting plasma cortisol levels than non-smokers [37, 38]. The increase in cortisol production leads to accumulation of abdominal fat , which, in turn, increases waist circumference.
Although some studies have indicated that smoking is related to reduced insulin sensitivity and the development of insulin resistance [12, 40] and type 2 diabetes [41, 42], in our population there was no consistent association between smoking and fasting blood glucose. This confirms the results obtained in other studies [13, 30, 43]. Ishizaka et al. found a higher prevalence of elevated blood glucose in smoking men, but not in women . Such discrepancy in the results may be due to the different cut-off values for elevated fasting glucose used in the present study (5.6 mmol/l) and that of Ishizaka et al. (6.1 mmol/l) .
While it is well established that acute smoking may cause a rise in blood pressure [44, 45], in the chronic situation smokers’ blood pressure is similar to or even lower than that of non-smokers [33, 44, 46], although Primatesta et al. found higher blood pressure in male smokers older than 45 years compared to never smokers . We found no association between smoking and blood pressure in any of the three BMI classes, even after correction for the use of blood pressure-lowering medication. In addition, we found similar blood pressure in smokers aged 45 and higher versus non-smokers (data not shown). Nevertheless, some studies have suggested that smoking may be a risk factor for developing hypertension  or for an increase in blood pressure during exercise , although in the latter study smoking cessation did not lead to reduced blood pressure. Weight changes after smoking cessation have been suggested to be involved in this paradox .
One of the new findings of our study is the association between smoking and alterations in levels of apolipoproteins and in the size of lipoprotein particles. Until now, only a limited number of studies have investigated the relationship between smoking and the levels of apoA1 and apoB, usually involving a small number of participants such as, for example, young adults [18, 50], middle-aged men [19, 51], or postmenopausal women . In addition, few studies have assessed the effects of smoking on lipoprotein particle size. In the Framingham study, smoking was associated with higher levels of small LDL particles . However, apoA1 and apoB measurement and standardization have considerably improved in the last decade, both because of the appearance of a legal and regulatory framework (the In Vitro Diagnostics (IVD)-directive 98/79/EC and the institution of the Joint Committee on Traceability in Laboratory Medicine (JCTLM)), technical improvements of equipment, and the availability of international reference materials . An additional milestone was the preparation, evaluation and introduction of value-assigned reference materials for monitoring trueness of apolipoprotein test results . ApoA1 is the main protein component of HDL-C particles, and higher levels of apoA1 are associated with lower risk of CVD . We observed that in current smokers plasma apoA1 levels were lower than in non-smokers. In addition, smoking was associated with lower HDL-C/apoA1 ratio, which is a strong indication of smaller HDL particle size. Such alterations of the HDL particle have been negatively associated with heart disease [54, 55]. While apoA1 is protective, apoB, the main protein component of LDL particles, reflects the atherogenic potential of LDL, and higher levels of apoB are associated with an increased risk of CVD . The fact that we found higher apoB levels and lower LDL-C/apoB ratios in current smokers than in non-smokers, indicates the presence of increased numbers of small, more dense LDL particles. Such particles have been found to increase the risk not only of atherosclerosis [56, 57], but also of coronary artery disease  and fatal myocardial infarction . Furthermore, in a 3-year follow-up study among Korean men without MetS, a low LDL-C/apoB ratio was independently associated with development of MetS . Taken together with our findings, the Korean study supports the conclusion that the presence of increased amounts of small, dense LDL particles can be considered both a risk factor for future cardiovascular disease and an early feature of metabolic syndrome.
Our study has several major strengths. Considering the number of participants recruited from the general population (N >59,000), this is the largest study reporting these results. Our large dataset also enabled to carefully calculate effect sizes, and to perform sufficiently powered subgroup analyses, in subjects of both sexes and in those with normal body weight, overweight, and obesity, which to our knowledge has never been performed before. All participants to the LifeLines Cohort Study have been well characterized, with rigorously standardized blood pressure and anthropometric measurements. In addition, all laboratory measurements of lipids and apolipoproteins have been carried out over a period of 5 years in fresh serum samples, in the same certified laboratory, with the same equipment, and the same rigorous quality assessment and control. This unprecedented sample size also provided us with sufficient statistical power to investigate contradictory associations reported previously.
There are also some limitations to our study. Firstly, since smoking status was based on self-administered questionnaires, we cannot exclude the possibility that misreporting led to some individuals being misclassified with regard to their current smoking status. Considering the large number of participants, we believe that misclassification has only very limited influence on the results obtained, and earlier studies also reported low misclassification rate of smoking status . We should point out that we were unable to identify individuals who had never smoked, nor could we fully take into account the duration of smoking. Secondly, apart from age we could not adjust for other possibly relevant risk factors that influence levels of HDL cholesterol and triglycerides, such as nutrition and alcohol consumption. As data collection for the LifeLines Cohort Study is still ongoing, we hope to be able to investigate the effects of such factors on MetS in the future.