Active smoking and macrocytosis in the general population: Two population-based cohort studies.

Macrocytosis, an elevated mean corpuscular volume (MCV) of erythrocytes, is a highly prevalent phenomenon in adult individuals.1 MCV is the measurement of the average volume of red blood cells, and macrocytosis is defined as a MCV exceeding 100 fL. Currently, in textbooks and guidelines a myriad causes are being mentioned for macrocytosis, with vitamin B12 and folate deficiency, alcohol use, myeloid dysplastic syndromes, and liver disease as the most prominent ones.2 In the 70s, a number of papers have reported a positive association between smoking and MCV.3, 4 This has nowadays, however, not resulted in inclusion of cigarette smoking as an important cause of macrocytosis in textbooks and guidelines. Hence, in the current study, we aimed to investigate the association between smoking, assessed by both questionnaire and 24‐hour urinary cotinine excretion, as objective measurement of nicotine exposure, with MCV in 2 large population‐based cohorts.

First, we analyzed data from the Lifelines cohort study. Lifelines is a large multi‐disciplinary prospective population‐based cohort study which examines, in a unique 3‐generation design, the health and health‐related behaviors of persons living in the north of The Netherlands. For the present study, we included 131 886 of the 167 729 subjects (aged 18‐93 years) of whom hematology indices, drinking and smoking behavior were available. Second, we analyzed data from the Prevention of Renal and Vascular End‐Stage Disease (PREVEND) study, a prospective, population‐based cohort of Dutch men and women aged 28‐75 years. For current analyses, we used data from the second survey (n = 6894) and excluded missing data on smoking behavior (n = 86), resulting in 6808 participants eligible for analyses. Smoking status was categorized as never, former, and current (<6, 6‐20, or >20 cigarettes/d). To exclude possible misclassification or under‐ or overestimation of number of cigarettes smoked per day as determined by questionnaire, 24‐hour urinary cotinine levels were measured. Alcohol use was categorized as no alcohol use, 1 U of alcohol per month to 1 U/wk, >1 U/wk to 7 U of alcohol per week, >1 U/d to 3 U of alcohol per day, or > 3 U of alcohol per day. Details of the Lifelines cohort and PREVEND study regarding clinical examination, biochemical measurements, data description and statistical analyses are described in the Supporting Information. Similarly, baseline demographics and clinical characteristics of the included 131 886 community‐dwelling participants and 6808 PREVEND participants are shown in Supporting Information Tables S1 and S2.

Of the 131 886 Lifelines participants (age 45 ± 13 years, 40% males), 47% were nonsmokers, 33% were former smokers and 20% were current smokers. Of the current smokers, 28% smoked <6 cigarettes per day, 55% smoked 6‐20 cigarettes per day and 18% smoked > 20 cigarettes per day. Hemoglobin levels were higher in current smokers (14.3 ± 1.2 g/dL) compared with nonsmokers (14.0 ± 1.3 g/dL, P < .001). Similarly, MCV levels were higher in current smokers (91.4 ± 4.3 fL) compared with nonsmokers (89.2 ± 4.0 fL, P < .001, Figure 1A). Macrocytosis was present in 494 (1.9%) of current smokers compared with 166 (0.3%) of nonsmokers (P < .001, Figure 1B).

Figure 1

Association of smoking and 24‐hour urinary cotinine excretion levels with mean corpuscular volume and macrocytosis. A, The association between smoking status and MCV in the lifelines cohort. Reported P‐values are shown in respect to reference category of nonsmokers. B, The prevalence of macrocytosis for each smoking status in the lifelines cohort. Reported P‐values are shown in respect to reference category of nonsmokers. C, The association between smoking status and MCV in the PREVEND study. Reported P‐values are shown in respect to reference category of nonsmokers. D, The association between 24‐hour urinary cotinine excretion levels and MCV by means of restricted cubic splines. Three knots have been specified at the 10th, 50th, and 90th of 24‐hour urinary cotinine percentiles. The 95% CIs are indicated by the shaded areas. Twenty‐four urinary cotinine levels have been natural log transformed. Abbreviations: MCV, mean corpuscular volume; PREVEND, prevention of renal and vascular end‐stage disease. *P < .05 **P < .01 ***P < .001

In univariable linear regression analysis, current smoking, compared with nonsmoking, was positively associated with MCV (β = 0.24, P < .001). In multivariable regression analysis, performed in the whole cohort, current smoking compared with nonsmoking, remained positively associated with MCV (β = 0.23, P < .001), independent of adjustment for age, sex, estimated glomerular filtration rate (eGFR), body mass index (BMI), and alcohol use. Multivariable regression analysis was also performed in a subgroup of participants from whom also gamma‐glutamyltransferase (GGT), alanine aminotransferase (ALAT), free thyroxine (FT4), and high‐sensitivity C‐reactive protein (hs‐CRP) were available (n = 36 109) with the same result (β = 0.23, P < .001).

Similarly, in logistic regression, smoking was a strong determinant of macrocytosis (OR 6.25, 95% CI 5.2‐7.51; P < .001 in the total cohort, OR 6.00, 95% CI 4.12‐8.73; P < .001 in the subgroup of n = 36 109), independent of adjustment for potential confounders.

Hereafter, we divided current smoking into categories of cigarettes smoked per day. In multivariate analysis, all smoking categories (<6 cigarettes [β = 0.07, P < .01], 6‐20 cigarettes [β = 0.22, P < .001], and >20 cigarettes [β = 0.19, P < .001]) were associated with MCV, independent of adjustment for potential confounders. The association remained the same after adjustment for GGT, ALAT, FT4, and hs‐CRP (<6 cigarettes [β = 0.06, P < .001), 6‐20 cigarettes [β = 0.22, P < .001], and >20 cigarettes (β = 0.21, P < .001]).

Of the 6808 subjects (age 53 ± 12 years, 50% males) in the PREVEND study, 29% were nonsmokers, 43% were former smokers, and 28% were current smokers. Of the latter, 16% smoked <6 cigarettes per day, 70% smoked 6‐20 cigarettes per day, and 14% smoked >20 cigarettes per day. Hemoglobin levels were higher in current smokers (13.9 ± 1.2 g/dL) compared with nonsmokers (13.6 ± 1.3 g/dL, P < .001). Similarly, MCV levels were higher in current smokers (92.3 ± 4.7 fL) compared with nonsmokers (89.2 ± 4.3 fL, P < .001, Figure 1C). Macrocytosis was present in 73 (4%) of current smokers compared with 8 (0.4%) of nonsmokers (P < .001).

In univariable linear regression analysis, current smoking, compared with nonsmoking, was positively associated with MCV (β = 0.30, P < .001). In multivariable analysis, current smoking, compared with nonsmoking, remained positively associated with MCV (β = 0.24, P < .001), independent of adjustment for age, sex, eGFR, BMI, hs‐CRP, alcohol use, GGT, ALAT, FT4, vitamin B12, and folic acid. Similarly, in logistic regression, smoking was a strong determinant of macrocytosis (OR, 8.54, 95% CI 2.57‐28.37; P < .001), independent of adjustment for potential confounders.

Hereafter, we divided current smoking into categories of cigarettes smoked per day. In multivariate analysis, smoking <6 cigarettes (β = 0.03, P = .06), was not associated with MCV, whereas smoking 6‐20 cigarettes (β = 0.24, P < .001), and smoking >20 cigarettes per day (β = 0.13, P < .001) remained, compared with nonsmoking, associated with MCV, independent of adjustment for potential confounders.

As sensitivity analysis, we repeated in the PREVEND study the analysis with 24‐hour urinary cotinine excretion levels as objective reflection of smoking. Twenty‐four hour urinary cotinine excretion was strongly correlated with current smoking (β = 0.82, P < .001). Similar to the primary analysis, we identified a strong positive association between 24‐hour urinary cotinine excretion and MCV (β = 0.26, P < .001, Figure 1D). The association remained independent of adjustment for potential confounders (β = 0.23, P < .001).

In this study, we have shown that smoking, assessed both by means of a self‐administered questionnaire and by 24‐hour urinary cotinine excretion levels, was strongly positively associated with MCV. Importantly, this association was independent of known causes of macrocytosis, including alcohol use. A few years ago, McNamee et al.5 and O'Reilly et al.6 reinvestigated the association between smoking as unrecognized cause of macrocytosis and showed that cigarette smoking was a significant risk factor for macrocytosis, independent of other known causes. Unfortunately, at present cigarette smoking is still not mentioned in textbooks and major guidelines, and clinicians are generally unaware of this association. The major drawback of the previously performed studies was that smoking status was assessed by means of a self‐administered questionnaire, which might still be regarded as a subjective measurement of smoking status. In this study, we underline the importance of this association, and we are the first to utilize an objective measurement that is, urinary cotinine excretion levels, for the current association. The latter combined with the large patient populations can be regarded also as the major strength of this study. Due to the observational design of this study, we cannot discern potential mechanisms for the strong association between smoking and MCV. Finally, despite the extensive number of factors for which we adjust, residual confounding can still not be excluded.

In conclusion, smoking is an important determinant of MCV levels and macrocytosis, independent of prominent causes such as alcohol intake, liver disease, vitamin B12, and folic acid deficiency. Smoking should be included in current guidelines regarding known causes of an elevated MCV, and the current study might draw more attention to the mechanism by which smoking causes macrocytosis independent of alcohol intake.

CONFLICT OF INTEREST

Nothing to report.

AUTHOR CONTRIBUTIONS

All authors read and approved the final version of the manuscript. M.F.E., H.J.C.M.W., G.H. and S.J.L.B. contributed to the study design. M.F.E. and H.J.C.M.W performed the statistical analysis. M.F.E., H.J.C.M.W., L.M.K., M.M.vd.K., P.vd.M., B.H.R.W., C.A.J.M.G., J.E.K‐R., D.J.T., G.H. and S.J.L.B contributed to the interpretation of the data and analysis. M.F.E. and H.J.C.M.W. wrote the first draft and all authors edited the paper.

Supporting Information Appendix S1: Supporting Information.

Click here for additional data file.