OUP user menu

Independent association of various smoking characteristics with markers of systemic inflammation in men

Margit Fröhlich, Malte Sund, Hannelore Löwel, Armin Imhof, Albrecht Hoffmeister, Wolfgang Koenig
DOI: http://dx.doi.org/10.1016/S0195-668X(03)00260-4 1365-1372 First published online: 2 July 2003


Aims Aim of the study was to investigate the association between various markers of systemic inflammation and a detailed history of smoking in a large representative sample of the general population.

Methods and results The effects of chronic smoking on white blood cell (WBC) count, fibrinogen, albumin, plasma viscosity (PV), and high-sensitivity C-reactive protein (CRP) were measured in 2305 men and 2211 women, age 25–74 years, participating in the third MONICA Augsburg survey 1994/95. In men, current smokers showed statistically significantly higher values for WBC count, fibrinogen, PV, and CRP, compared to never smokers, with intermediate, but only slightly increased values for ex-smokers and for occasional smokers. No consistent associations were seen with albumin. Duration of smoking was positively associated with markers of inflammation as were pack-years of smoking. Conversely, duration of abstinence from smoking was inversely related to these markers. Except for WBC count, no such associations were found in women.

Conclusion Data from this large representative population show strong associations between smoking and various markers of systemic inflammation in men. They also show that cessation of smoking is associated with a decreased inflammatory response, which may represent one mechanism responsible for the reduced cardiovascular risk in these subjects.

  • Smoking
  • Inflammation
  • C-reactive protein
  • Fibrinogen
  • Plasma viscosity
  • White blood cell count

1 Introduction

Cigarette smoking is a powerful risk factor for atherosclerosis. A variety of studies have shown an association of smoking with coronary heart disease (CHD),1peripheral artery disease,2aortic aneurysms,3and stroke.4On the other hand, after smoking cessation a time-dependent decrease in the risk of acute myocardial infarction and stroke, to a level almost similar to persons who never smoked, has been reported.5In general, the full risk reduction was seen within 5–10 years after abstinence from smoking.

Mechanisms by which cigarette smoke may initiate or accelerate atherosclerosis and its complications are not fully understood. Smoking causes coagulation and lipid abnormalities,6and circulatory effects due to enhanced catecholaminerelease have been reported.7Recent work focuses on smoking-induced endothelial injury, and there is evidence that mainly free radicals, such as nitric oxide radicals, singlet oxygen and hydrogen peroxide, mediate cigarette-smoke induced endothelial damage.8Oxidative stress promotes a systemic acute phase response possibly by NF-κB activation9and cytokine induction,10and an association of cigarette smoking with a variety of acute phase markers such as inflammatory cytokines,11,12cell adhesion molecules,12,13C-reactive protein (CRP),14–23fibrinogen,18,24–27plasma viscosity (PV),24,28–30white blood cell (WBC)-count6,31–33and albumin16,34,35has been reported. Therefore, markers of inflammation may directly indicate smoking-induced inflammatory responses. Additionally, the association of these markers of inflammation with smoking may modulate the smoking-associated risk, since several of them are by itself strong and independent risk factors for CHD.36

Some of the above mentioned studies merely investigated the association of solitary inflammatory markers with smoking status. Only incomplete data are available taking into account other smoking characteristics, such as duration of smoking, smoked pack-years, and duration of abstinence from smoking. Thus, in the present large population-based study we aimed to systematically investigate the association of the smoking status (never, ex, occasional, regular) with a variety of markers of systemic inflammation (CRP, fibrinogen, PV, WBC-count, albumin), and also to assess the association of various smoking characteristics with these markers.

2 Methods

The MONICA Project was a WHO-coordinated observational long-term study with the main aims of measuring trends in cardiovascular morbidity and mortality and assessing their relationship to changes in risk factor levels and medical care.37,38It comprised acoronary event register and three independent cross-sectional studies in 5-year intervals. The present analysis involved all 2305 men and 2211 women aged 25–74 years among those participating in the third MONICA Augsburg survey in 1994/95. The participation rate was 74.9%. Smoking characteristics such as smoking status (never smoker, ex-smoker, occasional smoker, regular smoker), type of smoked tobacco product (cigarettes, cigars, cigarillos, pipe), duration of smoking (years), number of smoked cigarettes, and duration of abstinence from smoking (years), were recorded in detail by a standardized questionnaire according to the recommendations of the WHO.39Cigarette pack-years were computed as duration of smoking (years) multiplied by the number of smoked cigarettes, divided by 20. Blood pressure was measured by a Hawksley random zero sphygmomanometer. Body height and weight were measured in light clothing, and BMI was computed as weight divided by the square of height (kg/m2). Alcohol consumption was determined as described elsewhere.40

View this table:
Table 1

Baseline characteristics and markers of inflammation

Age (years)49.9(14.2)49.0(13.8)
BMI (kg/m2)27.4(3.7)26.6(5.1)
Total cholesterol (mmol/l)6.02(1.17)5.95(1.14)
HDL-cholesterol (mmol/l)1.24(0.37)1.56(0.43)
Systolic blood pressure (mmHg)135(18.0)129(20.4)
Diastolic blood pressure (mmHg)82(11.4)78(11.0)
Alcohol consumption (g/d)a25.7(2.9-40)7.5(0-11.4)
Diabetes n (%)120(5.0)103(4.2)
Markers of inflammation
CRP (mg/l)b1.33(3.05)1.43(3.26)
Fibrinogen (g/l)b2.90(0.71)2.92(0.71)
Plasma viscosity (mPa·s)b1.229 (0.066)1.222 (0.063)
WBC count (103/μl)b7.2(1.9)7.1(1.8)
Albumin (g/l)44.3(4.7)42.4(4.4)
  • a Median (interquartile range)

  • b Antilogs of ln-transformed data. Data expressed as mean (standard deviation), or percentages.

View this table:
Table 2

Smoking status, relative frequencies (%)

View this table:
Table 3

Distribution of smoking characteristics in the adult general populationa

SexAge(years)Cigarettes per day(n)Pack-yearsDuration of smoking(years)Abstinence from smoking(years)
  • a Data expressed as median (IQ-range)

View this table:
Table 4

Smoking characteristics and markers of systemic inflammation

SexSmoking characteristicsCRP (mg/l)Fibrinogen (g/l)PV (mPa·s)Albumin (g/l)WBC-count (103/μl)
Smoking status
P value*<0.001<0.0010.0080.031<0.001
P valuea0.254<0.0010.0380.864<0.001
Cigarettes per day
P valuea0.0210.0230.1180.296<0.001
P valuea0.6890.2880.7210.226<0.001
Cigarette pack-years
P valuea0.001<0.0010.0270.150<0.001
P valuea0.0450.3570.9690.115<0.001
Duration of smoking (years)
P valuea<0.001<0.0010.0290.364<0.001
P valuea0.7610.0470.0580.370<0.001
Abstinence from smoking (years)
P valuea0.0140.4850.0870.4400.044
P valuea0.6710.4860.4680.5030.122
  • a P values for testing the equality of the means. Means (adjusted for age, BMI, total cholesterol, HDL-cholesterol, alcohol consumption, systolic and diastolic blood pressure, education, physical activity, diabetes, CHD and creatinine) computed from normalized transformed values.

Non-fasting venous blood samples were drawn into EDTA with the subjects in a supine resting position with only short term venous occlusion and minimal suction. Red blood cell count, WBC count, haematocrit and platelet count were determined (Coulter Counter). EDTA blood was immediately centrifuged at 3000g for 15min, and the plasma aliquoted and stored at −70°C. PV was measured at 37°C in a Coulter-Harkness capillary viscometer (Coulter Electronics, Luton, UK).41Fibrinogen and albumin concentrations were determined by immunonephelometry (Dade Behring, Marburg, Germany). CRP concentration was measured by an immunoradiometric assay (range 0.05–10mg/l) calibrated with the WHO reference standard 85/506.42Total and HDL-cholesterol concentrations were measured by routine enzymatic methods (CHOD-PAP, Boehringer, Mannheim, Germany), and LDL cholesterol was calculated according to Friedewald.43

Routine laboratory tests, including blood cell counts, total cholesterol and HDL-cholesterol were subject to continuous internal and external quality control. Coefficients of variation for repeated measurements were 0.7% for PV, 5.0% for fibrinogen, 6.9% for albumin and 12% for CRP over all ranges.

2.1 Statistical analysis

Baseline characteristics and markers of inflammation of the study population are described as means and standard deviations, if not indicated otherwise. Values have been calculated separately for men and women. CRP, fibrinogen, PV, and WBC count values were transformed to natural logarithms for greater symmetry of the distribution. Numbers shown in the tables are back-transformed to the original scale by taking antilogs, thus representing geometric means. The following analyses were restricted to cigarette smokers because of the small numbers of cigar, cigarillo and pipe smokers. Relative frequencies of the smoking status were calculated separately for men and women and for different age groups. Medians and interquartile ranges of different smoking characteristics were calculated for the same categories. Adjusted means of different markers of inflammation (CRP, fibrinogen, PV, WBC count, albumin) were computed according to the smoking status (never, ex, occasional, regular smoke exposure) and in quintiles of different smoking characteristics (number of smoked cigarettes, duration of smoking, pack-years, duration of abstinence from smoking). Adjustment was performed for multiple possible confounding covariates such as BMI, total cholesterol, HDL-cholesterol, alcohol consumption,systolic and diastolic blood pressure, education, physical activity, diabetes, CHD and creatinine. The p-values represent probability values for testing the simultaneous equality of the means. All computations were performed with SAS software (SAS Windows 6.12). P-values <0.05 were considered statistically significant.

3 Results

Baseline characteristics of the study population and markers of inflammation are reported in Table 1for men and women separately. Men and women differ in their lipid profile with higher total cholesterol and lower HDL-cholesterol concentrations in men, and a substantially higher alcohol consumption in men. On the other side, women tend to have higher concentrations of CRP than men.

Smoking is more prevalent in men than in women (Table 2). In general, regular smoking is more common in younger individuals than in the older above age 55 years. There are also a large number of ex-smokers in this age group, particularly among men. Men are heavier smokers than women (Table 3), but there are no differences in smoking duration. The age at onset of smoking (derived from the duration of smoking) tends to be the same in both sexes, with a tendency to an earlier beginning (in the teenage years) in younger individuals.

The associations of smoking status and different smoking characteristics with markers of systemic inflammation are shown in Table 4. All means presented have been adjusted for a variety of potential confounders by multiple linear regression. In men, the positive acute phase markers CRP, fibrinogen, PV and WBC count are positively associated with smoking status, with highest plasma concentrations in regular smokers, slightly increased concentrations in occasional and ex-smokers, and lowest concentrations in never smokers. Comparing never smoking with regular smoking men, plasma concentrations of CRP are nearly doubled in regular smokers (1.03mg/l vs 1.92mg/l). In men, the number of smoked cigarettes and pack-years and duration of smoking are positively, whereas duration ofabstinence from smoking is inversely associated with these markers of systemic inflammation. This result is very consistent and statistically significant for all positive acute phase markers, except for the inverse association of fibrinogen and PV with the duration of abstinence from smoking and PV with the number of smoked cigarettes per day. There is no association of albumin—an unspecific negative acute phase marker—with parameters of smoking. Furthermore, except for WBC count, there is no association of smoking with these markers of inflammation in women.

4 Discussion

In this large, population-based study we found a consistent and statistically highly significant association of positive acute phase markers with smoking in men. On the other hand, acute phase markers were negatively associated with years since cessation of smoking. Except for WBC count no such associations were found in women.

Several previous studies have reported anassociation of smoking with CRP,14–23but only few analysed in detail the association of smokingcharacteristics, such as the number of smoked cigarettes,14pack-years17,19and abstinence from smoking,15,17,23with CRP in a large representative population-based sample. Comparing current smokers with never-smokers, the magnitude of the difference seems to be higher in men (e.g. ∼1.5mg/l in the Physicians Health Study14), compared to women (e.g. 0.08mg/l in the Women's Health Study12). This is in accordance with the missing difference in women reported here. All of the previously reported data were cross-sectional, except one.15In the course of this retrospective study, a non-significant decrease in CRP concentration (−0.21mg/l) after smoking cessation of one year was reported. Conversely, in the present study reported smoking cessation covers a longer period (up to >30 years), and there was a greater difference in CRP concentration in men of 0.67mg/l, comparing smoking abstainers in the highest quintile (30–55 years of abstinence) to current smokers.

Significantly higher fibrinogen concentrations in smokers24,26,27with a dose-response relationship have been shown by others.25In some reports PV values24,28,29parallel the changes in fibrinogen concentrations, as we also could show in the present analyses. This is not surprising, since fibrinogen is a major determinant of PV.44Again, smoking cessation was not associated with lower PV and fibrinogen concentrations in women, and this corresponds with data from a longitudinal study28in which men experienced a significant decrease in PV of −0.021mPa.s after 2 years of smoking cessation, and women did not (difference −0.009mPa·s, ns).

In the present study, WBC count represented the parameter most reliably associated with smoking in both men and women. The association of WBC count with smoking has been reported previously6,31–33, showing a dose-response relationship31,32and an association with smokingduration.32Overall, the magnitude of the effect of smoking on WBC count (difference 1.5–2×103/μl) is comparable throughout the studies.

In contrast to other studies34,35we found an inverse association of the negative acute phase marker albumin with smoking in men only. Both, Das et al.16and Dales et al.34revealed significantly lower albumin levels in smokers, compared to non-smokers, but once more the difference was more prominent among men (men: 7.9g/l, women:4g/l16). A dose-response relationship and an association of albumin with abstinence from smoking was not observed in the present study, nor was it reported by others. The relatively weak association of albumin with smoking may be due to the fact that besides inflammation, serum albumin concentrations are mainly determined by other conditions such as nutritional status and liver function.

There are several possible mechanisms by which smoking may be associated with systemic markers of inflammation. First, acute phase proteins may be surrogates for smoking-induced chronic inflammatory processes such as periodontitis,45airwayinflammation46and atherosclerosis itself.47Second, certain compounds of smoke such as free radicals11and phenol-rich glycoproteins48directly exert an inflammatory stimulus on macrophages which may trigger the production of inflammatory cytokines such as TNFα, IL-1, and IL-6.11And third, it is conceivable that there is an indirect effect by nicotine-induced catecholamine release7which modulates the systemic and local cytokinebalance.49

Although women are more sensitive to harmful effects of smoking,1in the present study except for WBC no association of smoking and smoking characteristics with markers of the acute phase response were observed in women, compared to men. Taking the above mentioned mechanisms into consideration, one may speculate that women experience less smoking-related inflammatory reactions, e.g. airway inflammation. This may be attributable to a different puffing behaviour in women. Generally, women use less untipped and more low-yield cigarettes,50and take smaller and shorter puffs.51However, the sex differences in self-reported inhaling habits disappear in pharmacological motivated heavy-smoking women (>20cigarettes/d).50This is in line with our observation of markedly elevated inflammatory parameters just in the 5th quintile (>21cigarettes/d) of cigarette consumption in women.

The major strength of the present analysis is the fact that data are based on a large representative sample (participation rate 74.9%) of the general population including both sexes and a large age range. However, the results may be limited due to the cross-sectional nature of the study which excludes establishing causal relationships. Furthermore, residual confounding for example by other lipid variables and homocysteine may represent a problem. However, this does not seem to be critical since age-adjusted values were not significantly altered in multivariate analyses. Finally, data may be biased by incorrect reporting of smoking habits, mainly because we did not perform objective measures of cigarette consumption, such as cotinine plasma concentrations.

In summary, we found a strong and consistent association of markers of systemic inflammation with smoking and various smoking characteristics, particularly in men. The low grade systemic acute phase response may contribute to the increased cardiovascular risk in smokers. Markers of systemic inflammation are inversely associated with years since cessation of smoking. Thus, our data provide further evidence that immediate smoking cessation should be strongly recommended in particular to those individuals with CHD. Different smoking patterns may explain the less consistent results in women.


The authors thank Mrs Gerlinde Trischler for her excellent technical assistance and Mark B. Pepys, FRS, of the Royal Free and University College Medical School, London, Great Britain for providing the monoclonal and polyclonal CRP antibodies.


View Abstract