Arterial stiffness plays a key role in the pathophysiology of cardiovascular (CV) disease and is recognized as an independent risk factor for CV disease (1- 2). Pulse wave velocity (PWV) is a representative measure of arterial stiffness and has been used as a marker of not only vascular damage but also CV risk (1,3). Several epidemiological studies have demonstrated that smoking is a major risk factor for cardiovascular events (4- 5). Acute smoking impairs regional arterial distensibility and increases PWV (6- 7), and chronic smokers exhibit impairments of arterial distensibility and elevated PWV (8- 9). However, no prospective study has determined whether continuous smoking might accelerate the progression of arterial stiffening with advancing age and, if so, whether there might be a dose–response relationship. Moreover, it is not clear whether inflammation, one of the mechanisms underlying the atherogenic action of smoking (4,10), contributes to the progression of arterial stiffening attributed to smoking.

Accordingly, the present prospective study was conducted on a large cohort of middle-age Japanese workers to determine: 1) the effect of smoking status on the progression of arterial stiffening with age; 2) the relationship between the amount of smoking and the age-associated progression of arterial stiffening; and 3) involvement of inflammation in the progression of arterial stiffening attributed to smoking.

This cohort study was initiated on the employees of a single large construction company in the year 2000 when annual health checkups were mandated by the Occupational Health and Safety Law. In addition to the evaluation of routine clinical parameters such as blood pressure and serum cholesterol, the checkup included atherosclerotic risk factors and brachial-ankle pulse wave velocity (baPWV). Measurement of the serum C-reactive protein (CRP) levels was begun from the year 2002. Among all the study subjects, those who were working at the company headquarters were invited to undergo follow-up measurements of baPWV and blood and urine examinations from 2007 to 2008. Verbal informed consent was obtained from all of the study participants before their participation in this study. The study was conducted with the approval of the Ethical Guidelines Committee of Tokyo Medical University. Some related data have been presented elsewhere (11).

All the participants were provided feedback on the results of the annual health checkups. Along with other lifestyle modification recommendations, smoking cessation was recommended to the subjects who were identified as smokers. Subjects with positive atherosclerotic risk factors (body mass index [BMI] ≥25 kg/m², triglyceride ≥150 mg/dl, high-density lipoprotein [HDL] cholesterol <40 mg/dl, fasting plasma glucose ≥126 mg/dl, blood pressure ≥140/90 mm Hg, and/or total cholesterol ≥240 mg/dl) were advised to visit the health care center within the construction company and were provided with advice with regard to therapeutic lifestyle modifications by health professionals, in accordance with the guidelines of the Japanese Societies of Atherosclerosis, Diabetes Mellitus, and Hypertension (12- 17). Subjects requiring medications were prescribed appropriate drugs at either the health care center or at other clinics. Each patient was given the freedom to choose his/her own doctor for such treatment.

Subjects with the following conditions were excluded from the present study: 1) unusually low or unreliable baPWV values (due to ankle/brachial systolic blood pressure index [ABI] <0.95, atrial fibrillation, and/or regular hemodialysis) (11); and 2) serum CRP levels >10.0 mg/l (as the conventional clinical cutoff point for inflammation) (18).

A total of 2,357 subjects who were working at the company headquarters in the year 2002 were enrolled in the study protocol. Of these, 2,175 subjects were successfully followed up until 2007 to 2008, and 182 were excluded because of the lack of complete datasets (n = 18) and missing follow-up data (n = 164), due to moving of the subjects from the company headquarters to branch offices, layoffs, retirement, and so forth. Of the 2,175 subjects, 79 were excluded for the following reasons: ABI <0.95 (n = 17); atrial fibrillation (n = 12); serum CRP ≥10 mg/l (n = 48; 2 of them also had an ABI <0.95); and hemodialysis (n = 2). Finally, 2,096 subjects were successfully enrolled for the 5- to 6-year follow-up study.

Brachial-ankle pulse wave velocity was measured with a volume-plethysmographic apparatus (Form/ABI, Colin Company, Ltd., Komaki, Japan), as previously described (11,19). Briefly, electrocardiographic electrodes were placed on both wrists, and a microphone for the phonocardiogram was attached on the left chest. Electrocardiograms and phonocardiograms were used to provide timing markers for the device. Occlusion cuffs, which were connected to both the plethysmographic and oscillometric sensors, were tied around both the upper arms and ankles while the subjects were in the supine position. The brachial and post-tibial arterial pressures were measured by the oscillometric sensor. Ten-second recordings of the brachial and post-tibial arterial pressure waveforms recorded by the plethysmographic sensor were stored. The measurements were conducted after the subjects had rested for at least 5 min in the supine position in an air-conditioned room (maintained at 24°C) allocated exclusively for this study. Blood pressure was determined by the oscillometric sensor, and heart rate was also simultaneously recorded during the measurement of the baPWV. The interobserver and intraobserver coefficients of variation for baPWV measurements were 8% and 10% in our previous study (19).

For the analyses, the mean values of the baPWV and blood pressure measured on both sides (i.e., [value measured on the right side + value measured on the left side] / 2) were used. The annual rate of change in the baPWV during the study period was calculated as: value at the final examination minus value at the baseline examination divided by the duration of follow-up (years).

Serum concentrations of triglyceride, total cholesterol, HDL-cholesterol, creatinine, and CRP as well as plasma glucose concentration were measured with standard enzymatic methods (Falco Biosystems Company, Ltd., Tokyo, Japan) (20). All the blood samples were obtained in the morning, after the patients had fasted overnight.

Blood pressure was measured in an office setting with the subjects in the seated position by the conventional cuff method with a mercury sphygmomanometer. Measurements were obtained twice after the subjects had rested for at least 5 min, and the mean of the 2 measurements was reported.

Smoking status of the subjects (i.e., never-smoker, former smoker, current smoker) were assessed with a questionnaire. Current and former smokers were requested to provide an average number of cigarettes smoked/day and the duration (number of years) of smoking. The subjects were classified into 5 different categories on the basis of their responses at the baseline and final examinations; never smokers (who had no history of smoking), former smokers (who had a past history of smoking), quitters (who were current smokers in the baseline survey and former smokers in the final survey), continuous light-to-moderate smokers (who continued to smoke ≤20 cigarettes/day at the time of both the baseline and the final surveys), and continuous heavy smokers (who continued to smoke >20 cigarettes/day at both the baseline and the final survey) (21). The questionnaire was not designed to assess the reason why former smokers quit smoking or the exact date of quitting smoking. Starters or relapsers (who were never-smokers/former smokers at the baseline survey and changed their smoking status to current smokers at the final survey) were excluded, because of a smaller number of subjects in this group (n = 42) and the difficulty in precisely assessing their smoking history.

We estimated the average number of cigarettes smoked/day (amount of smoking) in the subjects who continued to smoke throughout the study period, to examine the dose–response relationship between the amount of smoking and the rate of changes in baPWV. The average number of cigarettes smoked/day during the study period was calculated by the sum of the mean daily number of cigarettes smoked reported at each annual checkup divided by the duration of follow-up (in years).

Data were expressed as mean ± SD unless otherwise indicated. The differences in the continuous variables across the 5 smoker categories were assessed by 1-way analysis of variance (ANOVA), and the differences in the measured values between the baseline and final examinations were assessed by the paired t test. In the ANOVA, Scheffe's test was applied for post-hoc multiple comparisons.

The categorical variables were categorized as follows: sex (female = 0 and male = 1); alcohol intake (nondrinker = 0, light-to-moderate drinker [ethanol consumption, 1 to 29 g/day] = 1, and heavy drinker [ethanol consumption, over 30 g/day] = 2); medication for coronary heart disease, stroke, hypertension, diabetes mellitus, and/or dyslipidemia (for each medication: no medication = 0, and taking medication = 1); and persistent elevation of the serum CRP level during the study period (i.e., serum CRP levels in the highest tertile range at both the baseline and final examinations) (no persistent elevation = 0, and persistent elevation = 1). Pearson's chi-square test was applied for assessment of the differences in the categorical variables across the 5 smoking status categories (post-hoc comparisons among subgroups were not conducted). McNemar's nonparametric test was applied for assessment of the differences in the medications prescribed between the baseline and final examinations, and Friedman's nonparametric test was applied for assessment of the differences in the amount of alcohol intake between the baseline and final examinations.

In the general linear model analyses, the annual rate of change in baPWV during the study period was entered as a dependent variable, the 5 smoker categories were entered as a fixed factor, and others were entered as covariates for adjustment. Two different adjustment models were applied. In Model 1, the covariates for adjustments were the baseline values of factors already known to be linked to arterial stiffness and/or atherosclerosis (1,3,11,19- 20) (i.e., age; sex; body mass index; alcohol intake; mean blood pressure; heart rate; serum total cholesterol; serum HDL-cholesterol; serum triglyceride; serum creatinine; fasting blood glucose; and medication for coronary heart disease, stroke, hypertension, diabetes mellitus, and/or dyslipidemia). In Model 2, the covariates for adjustments were all the variables included in Model 1 plus the changes in these variables (except for sex) during the study period. In the analysis without adjustments for covariates, Scheffe's test was applied for post hoc multiple comparisons. In the analysis with adjustments for covariates (Model 1 and 2), Bonferroni's test was applied for post hoc pair-wise comparisons.

We specifically tested the association between serum CRP and the annual rate of changes in baPWV, because CRP might mediate the effect of smoking on the progression of arterial stiffening. Because serum CRP levels were skewed, the analyses were conducted with and without log-transformation of serum CRP levels. Multivariate linear regression analysis with the aforementioned adjustments (Model 1 and 2) were used to test the influence of serum CRP as well as persistently elevated serum CRP on the annual rate of changes in baPWV. Finally, we conducted multivariate regression analysis with the Model 1 and 2 adjustments mentioned in the preceding text among continuous smokers, to test the dose–response relationship between the amount of smoking and the change in baPWV.

All the analyses were conducted with the IBM/SPSS software (version 17.0J for Windows, IBM/SPSS, Inc., Chicago, Illinois). The p values <0.05 were considered to denote statistical significance.

Among the 2,096 subjects, 42 were identified as relapsers and therefore were excluded from the analyses. The data of the remaining 2,054 subjects (age 40 ± 8 years, 1,648 men and 406 women) were included for the analyses. The follow-up period (study duration) was 6 years for 1,734 of the 2,054 subjects and 5 years for the remaining 320 subjects.

(Table 1) shows the clinical characteristics and changes in the body measurements during the study period in the 5 smoking categories. The paired t test demonstrated that baPWV increased significantly during the study period in all 5 smoking categories. The ANOVA with Scheffe's test demonstrated that the baPWV at the baseline examination was higher in the quitters and continuous heavy smokers than in the never-smokers. The mean age at the baseline examination was higher in the former smokers and continuous heavy smokers than in the other 3 categories of smokers. However, the group differences in baPWV remained significant even after the adjustments for covariates, including age (p < 0.05).

We examined, in the entire subject population (n = 2,054), whether continuous smoking could influence the rate of progression in arterial stiffening with advancing age. (Figure 1) shows the crude and adjusted values of the annual rate of increase in baPWV during the study period across the 5 different smoker categories. In (Figure 1), the upper panel presents the crude comparison with the Scheffe's method. Continuous heavy smokers were significantly different from never-smokers in the pair-wise comparison. Continuous heavy smokers were also significantly different from other 4 groups in the contrast comparison. In an adjusted model (Model 1), post-hoc pair-wise comparisons by Bonferroni's test demonstrated that the rate of arterial stiffening was significantly greater in the continuous heavy smokers than in the never-smokers and former smokers (Figure 1, middle panel). In further adjustment including changing scores (Model 2), the rate of arterial stiffening was significantly greater in the continuous heavy smokers than in the never-smokers, former smokers, and quitters (Figure 1, lower panel). Because there were fewer women than men in the present study, the same analyses were conducted with only male data. The same results shown in (Figure 1) were obtained even if we confined the analyses to men only (n = 1,648) (data not shown).

As shown in (Table 1), ANOVA with the Scheffe's test demonstrated that the serum CRP concentrations were greater in the continuous heavy smokers than in the never-smokers at both the baseline and final examinations. The results of the Pearson's chi-square test demonstrated that the prevalence of subjects with persistently elevated serum CRP during the study period (as defined in the statistical analysis section) differed among the 5 smoking categories (i.e., the prevalence rate in the continuous smokers was 34%, whereas that in the never-smokers was 14%).

Univariate analyses in the entire subject population (n = 2,054) revealed that the annual rate of increase in baPWV was not significantly associated with baseline serum CRP levels, baseline log-transformed serum CRP levels, changes in serum CRP levels during the study period, and changes in log-transformed serum CRP levels. The annual rate of increase in baPWV was significantly correlated only with persistent elevations in serum CRP levels (Table 2). However, in the multivariate linear regression analysis with adjustments using Model 1 and Model 2, the relationships between the annual rate of increases in baPWV and the persistent elevation of the serum CRP did not reach statistical significance (Table 3).

The completed questionnaire on the smoking status was available for all the study years (except in year 2005) in 493 of the 528 continuous smokers. On univariate analyses, there was a significant positive relationship between the annual rate of increases in baPWV and the number of cigarettes smoked/day (r = 0.19, p < 0.01). Furthermore, multivariate linear regression analyses with adjusted Model 1 and Model 2 demonstrated that the number of cigarettes smoked/day was a significant independent correlate with the annual rate of increases in baPWV (Table 4). The results were essentially the same when the sample was limited only to men (n = 480) (data not shown).

To the best of our knowledge, the present study was the first prospective study to examine the effects of smoking status as well as the amount of smoking on the rate of progression in arterial stiffening with aging. The results revealed that the annual rate of increases in baPWV was significantly greater in the continuous heavy smokers than in the never-smokers. These findings suggest that continuous heavy smoking accelerates the age-associated progression of stiffening of the large conduit arteries and could contribute to the age-related increases in CV disease.

A dose–response relationship between cigarette consumption and risks of CV events has been reported (22- 24). Baldassarre et al. (23) described, on the basis of the results of their cross-sectional study, the existence of a significant relationship between the amount of smoking and the severity of carotid atherosclerosis, as assessed by ultrasound examination. The present subanalysis, in which information about the number of cigarettes smoked/day was successfully obtained each year (except year 2005) throughout the study period, demonstrated a significant relationship between the number of cigarettes smoked/day and the annual rate of changes in baPWV. These results suggest the existence of a dose–response relationship between the amount of continuous smoking and age-associated accelerations of arterial stiffening.

Smoking is thought to be involved in the initiation and progression of atherosclerosis via several different mechanisms (4,10,25), and some studies have suggested the involvement of inflammation in the accelerated progression of arterial stiffening in smokers (3,20). In the present study, serum CRP levels were higher in the continuous smokers than in the never-smokers. Although persistently elevated serum CRP levels had a significant association with the progression of arterial stiffening on the basis of the univariate analyses, such associations did not reach statistical significance when the multivariate linear regression model was applied. Serum CRP is a marker of systemic inflammation but not a robust marker of vascular inflammation (26). Recently, Schumacher et al. (27) suggested that CRP does not have a causal role in the progression of arterial stiffening. Thus, further studies are needed to determine the contribution of smoking-related vascular inflammation to the progression of arterial stiffening.

Increased aortic stiffness acts as a CV risk factor via several different mechanisms, including increased cardiac afterload, impaired coronary blood flow, increased arterial wall stress, and microvascular damage (1- 3). Carotid-femoral PWV is the most established index for assessment of the aortic stiffness (1,3), but baPWV has been increasingly used in the research and clinical settings because its measurement is extremely simple, allowing repeated measurements in a large number of study subjects. Although baPWV is a marker of stiffness of both large and middle-sized arteries (11,19), baPWV demonstrates a close correlation not only with carotid-femoral PWV (28) but also with aortic PWV, as assessed by a direct catheter-method (19). Furthermore, increased baPWV has been shown as a marker of prognosis in subjects with acute coronary syndrome and end-stage renal disease (29- 30).

If continuous smoking itself was a significant determinant of the progression of arterial stiffening, baPWV at the baseline examination should be greater in the continuous heavy smokers than in the former smokers. However, baseline baPWV was not different between the 2 groups. The precise explanation for this apparent discrepancy is unknown. Part of the reason might be related to the questionnaire used to evaluate the smoking status. Several studies have suggested that questionnaire surveys to determine the smoking status might underestimate the prevalence of smoking (31), and no biochemical confirmation of quitting smoking was carried out in the present study. An alternative explanation is that the effects of chronic smoking on arterial stiffness would persist for a long time after the smoking cessation. Indeed, 5 to 6 years of no smoking did not induce any noticeable arterial de-stiffening in the former smokers or in the quitters. Jatoi et al. (8) suggested, on the basis of the results of their cross-sectional study, that it might take more than 1 decade to reverse the smoking-induced arterial stiffening. Another possibility is that smoking is simply one of the various factors that affect arterial stiffening. Indeed, after the adjustments for a number of confounders (Model 1), the association between arterial stiffening and cigarette smoking remained statistically significant but weakened. This would suggest that other pertinent factors (e.g., blood pressure) might be exerting important influences on arterial stiffening.

There are several study limitations in the present study that should be emphasized. First, the subjects were relatively young (mean age of approximately 40 years), and there were fewer women than men. Therefore, the confirmation of the present findings in older subjects and/or in women is warranted. Second, the effects of passive smoking were not examined in the present study.

The present study was the first prospective study to suggest that continuous smoking could accelerate the age-associated progression of structural stiffening of the large- to middle-sized arteries. Our findings also support a dose–response relationship between cigarette consumption and accelerated arterial stiffening. The elevated serum CRP levels associated with continuous smoking do not seem to be directly associated with the acceleration of arterial stiffening in continuous smokers.