| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Division of Endocrinology and Metabolism (R.M.B., P.W.L.) and Welch Center for Prevention, Epidemiology, and Clinical Research (B.C.A., N.R.P.), Johns Hopkins Medical Institutions, Baltimore, Maryland 21287
Address all correspondence and requests for reprints to: Dr. Ruth M. Belin, Division of Endocrinology and Metabolism, Johns Hopkins University School of Medicine, 1830 East Monument Street, Suite 333, Baltimore, Maryland 21287. E-mail: rbelin2{at}jhmi.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
-interferon or IL-2 (2, 3, 4), histological evidence of both B cells and cytotoxic T cells in autoimmune thyroiditis (5), specificity of activated helper T cells for thyroid antigens (6, 7), expression on affected thyrocytes of the major histocompatibility complex class II proteins needed for antigen presentation to helper T cells (8, 9), and creation of animal models through immunization with thyroid antigens (10). Interaction between genetic predispositions and external exposures influences the expression of thyroid autoimmunity and associated thyroid dysfunction. The genetics of autoimmune thyroid disease and the importance of many predisposing genetic loci have been elucidated in familial studies (11, 12, 13) and linkage analyses (14, 15, 16). Several environmental influences have been associated with the development of hypothyroidism in individuals with autoimmune thyroiditis, including increased dietary iodine content (17, 18), lithium and amiodarone exposure (19, 20), and cigarette smoke (21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35).
The relationship between cigarette smoke and thyroid function is complex. Smoke interferes with thyroid gland hormonogenesis (23, 36, 37, 38, 39, 40) and with peripheral thyroid hormone action (30, 41). However, stimulatory effects of smoking on the thyroid are suggested by increased serum thyroglobulin and T3 concentrations and decreased serum TSH levels (23, 28, 39). Nicotine does not appear to be the causative agent because nicotine infusion in rats had no effect on serum T4, T3, or TSH (42). Smoking can exacerbate established mild hypothyroidism (28). Several studies have shown that cigarette smoke impairs thyroid function in women with iodine deficiency or autoimmune thyroiditis, whereas other studies have not confirmed the association between smoking and Hashimoto’s thyroiditis (21, 22, 23, 24, 25, 25, 28, 31, 33, 34, 35). Similarly, variability is demonstrated in studies evaluating smoking and the development of postpartum thyroiditis, another autoimmune thyroid condition (26, 27, 29, 32). The case-control design of prior studies makes them prone to selection bias. Assessment of smoke exposure by questionnaires raises concern about recall bias. Individuals who have been diagnosed with a disease thought to be associated with smoking may be more likely to describe a positive smoking history.
A population-based study was conducted to more definitively assess the relationship between cigarette smoke exposure and common thyroid abnormalities in the U.S. population and to determine whether such a relationship is influenced by age, gender, race-ethnicity, or iodine status.
|
Subjects and Methods |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The study design was a national cross-sectional assessment using data collected from 1988–1994 for the Third National Health and Nutrition Examination Survey (NHANES III). NHANES III used complex, multistage, stratified, clustered samples of civilian, noninstitutionalized subjects, ranging from 2 months to 90 yr of age, to represent the entire U.S. population. Sample weighting compensates for the differential probabilities of selection and nonresponse and allows estimation of prevalence in the civilian U.S. population. Description of informed consent, appropriate treatment of human subjects, and statistical methods of NHANES III are detailed in the plan and operations manual (43). The data are publicly available on CD-ROM issued by the National Center for Health Statistics of the Centers for Disease Control and Prevention.
To be eligible for this analysis, subjects had complete data available regarding age, gender, race-ethnicity, iodine status, smoke exposure, and thyroid tests. Among 39,695 participants, 18,148 who were age 12 yr or older underwent thyroid tests. Of these, 16,046 individuals ranging from 12–90 yr of age met the criteria. Four hundred and fifty-four subjects from this group were taking medications that could affect thyroid testing (desiccated thyroid, levothyroxine sodium, liotrix, methimazole, potassium iodide, potassium iodide-theophyline, propylthiouracil, and thyroglobulin). The proportion of smokers was evaluated in these 454 subjects. The remaining 15,592 individuals not taking thyroid-related medications were then analyzed.
Measurements
Participant characteristics included age in years, gender, and self-reported race-ethnicity. Race-ethnicity was classified as non-Hispanic white, non-Hispanic black, Mexican-American, or other. Recent dietary iodine status was estimated by the ratio of random urine iodine (Sandell-Koltoff reaction) to urinary creatinine (Jaffé alkaline picrate method) measurements, which has been shown to approximate 24-h urinary iodine excretion (44, 45). Consistent with prior analysis of NHANES III, a concentration less than 50 µg iodine/g creatinine was classified as low urinary iodine, and a concentration greater than 500 µg iodine/g creatinine was classified as high urinary iodine (44).
Smoke exposure was assessed with serum cotinine measurements (enzyme immunoassay screen, STC, Inc., Bethlehem, PA; liquid chromatography-tandem mass spectrometry confirmation by SCIEX, PerkinElmer, Wellesley, MA), a metabolite of nicotine often applied as a marker of smoke exposure due to its longer half-life than nicotine. The half-life of cotinine is about 24 h compared with nicotine’s half-life of 30 min. The levels do not usually change once a sample is collected (46, 47, 48). Based on prior studies validating use of cotinine measurements to reflect reported home/work smoke exposure and reported active smoking in NHANES III, a cut-off level of more than 15 ng/ml designated active smoking and a level of 15 ng/ml or less designated nonactive smoking (49). Cotinine measurements were also evaluated as a continuous variable. Additional dose-response analyses evaluated three categories of smoke exposure: active smoking (cotinine > 15 ng/ml); mild smoke exposure, such as that associated with passive smoke (cotinine 0.05–15 ng/ml); and no detectable smoke (cotinine < 0.05 ng/ml). Ever-smokers, or those with a history of prior or current smoking, met any one of the following criteria: serum cotinine level greater than 15 ng/ml, self-reported lifetime history of smoking more than 100 cigarettes, smoking longer than 1.5 yr, or smoking more than 10 cigarettes/d. Persons with cotinine levels of 15 ng/ml or less who self-reported smoking more than 100 cigarettes, longer than 1.5 yr, or more than 10 cigarettes/d in the past were classified as prior smokers.
Outcome measures were the presence of 1) anti-TPOAb level of 0.5 IU/ml or more or antithyroglobulin antibody (TgAb) level of 1.0 IU/ml or more, 2) TSH level greater than 4.5 mU/liter, 3) TSH level less than 0.1 mU/liter, and 4) TSH level of 0.1–0.4 mU/liter. TPOAb and TgAb were measured by a highly sensitive, direct RIA system (Kronus, San Clemente, CA). The normal range is less than 0.5 IU/ml for TPOAb and less than 1.0 IU/ml for TgAb. Serum TSH was measured with a chemiluminescence immunometric assay (Nichols Institute Diagnostics, San Juan Capistrano, CA), with a working range of 0.01–50 mU/liter. The reference normal range was 0.4–4.6 mU/liter.
Data analysis and statistical methods
Weighted analyses using the survey command to account for the complex sampling design of NHANES III were performed using STATA 7.0 (50). The distributions of exposure variables, participant characteristics, and outcome variables were first examined. The presence, direction, magnitude, and independence of the association between participant smoking exposure and each thyroid outcome were evaluated using Pearson’s 
2 test for comparison of proportions and using odds ratios (ORs) derived from logistic regression models. Individuals with suppressed TSH measurements below 0.4 mU/liter were excluded from analyses evaluating TSH elevation. Similarly, individuals with TSH elevation greater than 4.5 mU/liter were excluded from analyses of TSH suppression. The same methods were used to assess associations between participant characteristics and each thyroid outcome. Participant characteristics that were statistically significantly associated with thyroid outcomes were considered potential confounders of the relationship between smoke and thyroid disorders. Alcohol was not found to be a statistically significant confounder. Multivariate logistic regression analyses evaluated the independent contribution of smoke to the risk of thyroid disease while controlling for potential confounders identified previously. Adjusted rates were calculated based on adjusted ORs derived from logistic regression and the unadjusted frequency in the reference group (51). Comparison of proportions with Pearson’s 2 analyses and logistic regression models were repeated in persons stratified by gender, age categories, race-ethnicity categories, and iodine categories.
Variance inflation factors showed no suggestion of colinearity of covariates. Multivariate analyses were performed in a stepwise manner to determine the incremental contribution of adding each variable. Interaction terms were included to determine whether the relationship between smoke and thyroid outcome differed in various categories. Sensitivity analysis adding a variable representing glucocorticoid use to the multivariate logistic regression models did not change the relationships appreciated between smoking status and thyroid outcome. Also, subgroup analysis in individuals reporting no glucocorticoid use showed no change in the associations between active smoke and thyroid outcomes.
|
Results |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The characteristics of the study population stratified by smoking status are described in Table 1
. Smokers were younger (mean age [95% confidence interval (CI)] = 40 yr [39–41 yr]) than nonsmokers (mean age [95% CI] = 44 yr [42–46 yr]). There were more male smokers than women smokers, and there was a statistically significant different distribution of race-ethnicity among smokers compared with nonactive smokers. There was no statistically significant difference in the proportion of smokers with iodine deficiency compared with nonsmokers.
|
Overall, the frequency of thyroid autoantibody presence was 16%, 95% CI = [15–17%]. The prevalence of TSH elevation greater than 4.5 mU/liter was 4.6%, 95% CI = [3.9–5.1%].
Thyroid autoantibody presence
Association between smoking status and presence of thyroid autoantibodies. Fewer smokers (11%, 95% CI = [10–13%]) had TPOAb and/or TgAb presence compared with nonsmokers (18%, 95% CI = [17–19%]). Smokers were found to have 43% lower odds of presence of thyroid autoantibodies compared with nonsmokers (OR [95% CI] = 0.57 [0.48–0.67]). Several characteristics were found to be potential confounders of the relationship between smoke and thyroid autoantibody status. After adjusting for the other characteristics, there were 2% higher odds of thyroid autoantibody presence with each year increase in age (OR [95% CI] = 1.019 [1.017–1.025]) and 200% higher odds in females compared with males (OR [95% CI] = 2.0 [1.7–2.3]). Lower odds of thyroid autoantibody presence were associated with being non-Hispanic black (66% lower compared with whites, OR [95% CI] = 0.34 [0.29–0.41]) or Mexican-American (20% lower odds, OR [95% CI] = 0.8 [0.7–0.95]). Lower odds were suggested in subjects with iodine deficiency compared with those with normal iodine (unadjusted OR [95% CI] = 0.5 [0.4–0.8]; adjusted OR [95% CI] = 0.7 [0.5–1.0]).
After taking into account characteristics associated with thyroid autoantibody presence, a smaller proportion of smokers compared with nonsmokers was still found to have thyroid autoantibodies. As shown in Fig. 1, an adjusted proportion of 13%, 95% CI = [12–15%] of smokers compared with 18%, 95% CI = [17–19%] of nonsmokers was found to have thyroid autoantibodies. The relationships persisted upon analyzing the association between smoke exposure and presence of TPOAb (independent of TgAb status) and upon analyzing smoke exposure and presence of TgAb (independent of TPOAb status).
|
summarizes the evaluation for dose relationships between smoke exposure and thyroid autoantibody presence. The odds of having thyroid autoantibodies present was lower by 1.1% for every 10 ng/ml increase in serum cotinine (OR [95% CI] = 0.989 [0.983–0.995]) after adjustment for age, gender, race-ethnicity, and dietary iodine status. A dose response was also supported by evaluation of categories of smoke exposure. Active smokers compared with individuals without any detectable smoke exposure had 59% lower odds of presence of thyroid autoantibodies and 50% lower odds after adjusting for participant characteristics. Subjects with serum cotinine levels suggestive of mild smoke exposure (0.05–15 ng/ml) compared with those without detectable smoke exposure had 30% lower odds of having thyroid autoantibody present. After adjustment for participant characteristics, subjects with mild smoke exposure had an estimated 20% lower odds compared with those without detectable smoke exposure, but this was not statistically significant (OR [95% CI] = 0.8 [0.6–1.0]).
|
Association of smoke and thyroid autoantibodies in groups stratified by age, gender, race-ethnicity, and urinary iodine.
Table 3 summarizes unadjusted and adjusted ORs assessing the relationship between active smoking and thyroid autoantibody presence in subjects stratified by participant characteristics.
|
Inclusion of an interaction term (reflecting active smoking status and non-Hispanic black race-ethnicity) in the multivariate model confirmed the observation that the OR was attenuated in non-Hispanic blacks compared with the other race-ethnicity groups (P < 0.03). Addition of an interaction term (reflecting active smoking status and age category) evaluating whether the OR differs in individuals older than 60 yr compared with those less than 40 yr of age was not statistically significant (P < 0.2).
The numbers of Mexican-Americans and other race-ethnicities were too small to determine precise association. An interaction term to assess an altered association among Mexican-Americans compared with non-Hispanic whites was not significant (P < 0.2).
Elevated serum TSH concentration
Association between smoke exposure and presence of serum TSH elevation.
Fewer smokers (2.6%, 95% CI = [2.0–3.2%]) had TSH levels greater than 4.5 mU/liter compared with nonsmokers (5.4%, 95% CI = [4.7–6.3%]). The OR of TSH elevation in smokers compared with nonsmokers was 0.5, 95% CI = [0.4–0.6]). Characteristics associated with greater odds of TSH elevation (after adjustment for the remaining characteristics) were increasing age (3.5% increased odds with each year; OR [95% CI] = 1.035 [1.028–1.041]), female gender (60% increased odds compared with males; OR [95% CI] = 1.6 [1.3–2.0]), and high urinary iodine (70% increased odds compared with individuals with normal urine iodine; OR [95% CI] = 1.7 [1.2–2.5]. Non-Hispanic black race ethnicity was associated with 60% lower odds of TSH elevation (OR [95% CI] = 0.4 [0.3–0.6]). The relationship between smoke exposure and the presence of TSH elevation remained statistically significant after adjustment for participant characteristics. Figure 1
illustrates the lower adjusted prevalence of TSH levels greater than 4.5 mU/liter in smokers compared with nonsmokers.
Association between dose of smoke exposure and presence of serum TSH elevation.
The distribution of serum TSH concentration is summarized in Fig. 2 for three subgroups: individuals without any smoke exposure, individuals with mild smoke exposure, and active smokers. This demonstrates that active smokers have a TSH distribution that appears more narrow (less variance) and shifted toward lower levels. For each range of TSH values, the percentage of subjects with mild smoke exposure is intermediate between the percentage of active smokers and the percentage of those without detectable exposure. Inclusion of serum cotinine measurement as a continuous variable in the multivariate model showed that after adjusting for participant characteristics, the odds of having elevated TSH levels were lower by 1.4% for every 10 ng/ml increase in serum cotinine (Table 2). When analyzed by category of smoke exposure, subjects with mild smoke exposure (cotinine 0.05–15 ng/ml) were associated with 40% lower odds of elevated TSH levels compared with individuals with undetectable cotinine (OR [95% CI] = 0.6 [0.4–0.7]). A trend remained, but the association was not statistically significant, after taking gender, age, race-ethnicity, and iodine status into account (OR [95% CI] = 0.8 [0.6–1.0]).
|
Association between smoke exposure and presence of serum TSH elevation by age, gender, race-ethnicity, and iodine status.
Table 3 summarizes the unadjusted and adjusted associations between active smoking and TSH elevation in subjects stratified by participant characteristics. Smokers compared with nonsmokers had statistically significantly lower odds of elevated TSH levels among participants aged 40–59 yr, those 60 yr or older, males, females, non-Hispanic whites, and subjects with normal urinary iodine.
The association between smoke and TSH elevation was attenuated among younger subjects, in whom there was no statistically significant difference between smokers and nonactive smokers. The estimated 7% lower odds (OR [95% CI] = 0.93 [0.59–1.47]) were smaller in magnitude than those found in the other persons (estimates ranging from 30–60% lower odds). However, there was no statistically significant interaction in the multivariate model (P < 0.1). Among Mexican-Americans, there were 30% greater odds estimated in Mexican-American smokers compared with Mexican-American nonactive smokers (OR [95% CI] = 1.3 [0.8–2.2]). Inclusion of an interaction term (reflecting active smoking status and Mexican-American race-ethnicity) in the multivariate model was statistically significant, confirming a different association between smoke exposure and the presence of TSH elevation among Mexican-Americans compared with non-Hispanic whites (P < 0.005). There was no statistically significant different association among non-Hispanic blacks compared with non-Hispanic whites (P < 0.2).
Association between active smoking status and subnormal serum TSH concentration
TSH less than 0.1 mU/liter. There was no statistically significant difference in the proportion of active smokers with TSH levels below 0.1 mU/liter (0.6%, 95% CI = [0.4–0.9%]) compared with nonactive smokers with TSH levels below 0.1 mU/liter (0.3%, 95% CI = [0.10–0.49%]). The adjusted OR for smokers compared with nonactive smokers with TSH levels below 0.1 mU/liter was 0.49, 95% CI = [0.21–1.17]. The trend toward fewer TSH measurements below 0.1 mU/liter in active smokers was consistent in the stratified analyses of women, men, subjects less than 40 yr of age, subjects more than 60 yr of age, non-Hispanic whites, non-Hispanic blacks, Mexican-Americans, individuals with normal urinary iodine, and subjects with high urinary iodine, but not in individuals aged 40–60 yr. There were too few outcomes in the other race-ethnicity category and in the low urinary iodine groups to provide a precise assessment.
TSH levels of 0.1–0.4 mU/liter.
The proportion of active smokers with TSH levels of 0.1–0.4 mU/liter was 2.2%, 95% CI = [1.4–3.0%] compared with 1.2%, 95% CI = [0.85–1.4%] of nonactive smokers with TSH concentrations of 0.1–0.4 mU/liter. However, Pearson’s 2 test for difference in distribution of smokers and nonsmokers in subjects with TSH levels of 0.1–0.4 mU/liter (vs. normal TSH) was statistically significant (P < 0.005). ORs derived from the univariate and multivariate logistic regression analyses showed statistically significant 200% greater odds of presence of TSH levels of 0.1–0.4 mU/liter in active smokers compared with nonactive smokers (unadjusted OR [95% CI] = 1.8 [1.2–2.7]; adjusted OR [95% CI] = 2.0 [1.3–2.9]).
The relationship between serum cotinine measurement (as a continuous variable) and the presence of subnormal TSH measurement was evaluated. A 2% higher odds of having a low normal TSH level was associated with every 10 ng/ml increase in cotinine (unadjusted OR [95% CI] = 1.02 [1.01–1.03]; adjusted OR [95% CI] = 1.02 [1.01–1.03]).
The higher odds of the presence of low normal TSH in active smokers was seen among subgroups of women, non-Hispanic whites, non-Hispanic blacks, Mexican-Americans, subjects less than 40 yr of age, and subjects with normal or low iodine, but not among men, subjects 40–60 yr of age, or those older than 60 yr.
Association between history of prior or current smoking and thyroid disorders
The prior analyses (of subjects not taking thyroid-related medications) were repeated, redefining exposure as any present or past history of smoking, to evaluate the possibility that laboratory abnormalities may be more frequent in nonsmokers due to confounding by ex-smokers with thyroid abnormalities. This showed a persistent negative association between ever-smoking and decreased thyroid autoantibodies (OR [95% CI] = 0.8 [0.7–0.9]). There was no statistically significant relationship between ever-smoking and TSH elevation (OR [95% CI] = 0.9 [0.7–1.2]) or between ever-smoking and subnormal TSH concentration. There were no statistically significant associations between a history of prior, but not current, smoking and any thyroid outcome. Thus, the higher number of subjects with TSH elevation in the nonsmoker group is probably not attributable to cessation of smoking in individuals who develop TSH abnormalities.
To further evaluate that possibility, the proportion of smokers among the 454 individuals taking thyroid-related medication was assessed. This showed a smaller percentage of smokers (16.6%, 95% CI = [11.7–21.5%]) than the proportion of smokers among the individuals not reporting thyroid-altering medications (32.2%, 95% CI = [30.5–33.8%]). This was true among subjects less than 40 yr of age and those 40–60 yr of age. There was no statistically significant difference among subjects older than 60 yr, although there was suggestion of a smaller proportion of smokers among subjects not taking thyroid medications. Among those taking thyroid-altering medications, smoking was not statistically significantly related to the presence of thyroid autoantibodies (adjusted OR [95% CI] = 1.8 [0.7–4.2]), TSH elevation (adjusted OR [95% CI] = 0.7 [0.2–2.3]), or presence of TSH levels below 0.1 mU/liter (adjusted OR [95% CI] = 0.96 [0.4–2]).
Association between smoke exposure and thyroid disorders in subjects with thyroid autoantibodies present
Among the individuals with detectable thyroid autoantibodies not taking thyroid-related medications, there remained statistically significant lower odds of TSH elevation in smokers compared with nonsmokers after adjustment for important participant characteristics (OR [95% CI] = 0.6 [0.4–0.97]).
|
Discussion |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
NHANES III provided an opportunity for more definitive analysis of the relationships between active smoking and thyroid disorders than has been performed to date. The dataset was large, with adequate sampling for conclusions regarding race-ethnicity. It included detailed information about confounders. Its design had the advantage of studying individuals from the general population, rather than being limited to those who were identified due to high risk for thyroid disorders, established thyroid abnormality, or interest in medical attention. Finally, the study design avoided recall bias by employing a laboratory measurement to assess exposure.
Our findings of lower odds of TSH levels greater than 4.5 mU/ml in smokers (OR [95% CI] = 0.6 [0.5–0.8]) after adjustment for age, gender, race-ethnicity, and iodine status were consistent with a smaller, cross-sectional study of 4649 Danish individuals. In the Danish study, the prevalence of mild hypothyroidism (TSH > 3.6 mU/liter) was lower among self-reported smokers, with an OR of 0.47 after adjustment for age, gender, and iodine status (95% CI = [0.33–0.67]), but there was inadequate power to demonstrate a relationship between cigarette smoke and more substantial degrees of thyroid dysfunction (TSH > 5.0 mU/liter) (28). In contrast, the present analysis included a larger number of individuals sampled, providing adequate power to detect statistically significant effects of smoke in various ranges of TSH and permitting analyses of different race-ethnicities. Furthermore, the current study used laboratory measurement of smoke exposure rather than self-reported determination.
Our findings of lower prevalence of thyroid autoantibodies in smokers compared with nonsmokers were also consistent with a prior cross-sectional study that found a negative relationship between smoking and presence of anti-TPOAb among 803 healthy female relatives of individuals with autoimmune thyroid disease (34).
Our study differs from other reports describing higher or equivalent prevalences of hypothyroidism in smokers. One report from a location of relative iodine deficiency showed more hypothyroidism among smokers compared with controls (35). In NHANES III, there was inadequate power to detect statistically significant relationships among individuals with low urinary iodine. However, smoking appeared to be associated with a trend for less development of thyroid autoantibodies and more prevalent TSH elevation. In contrast, among individuals with normal iodine status in NHANES III, there was a statistically significant relationship between smoking and a lower prevalence of both thyroid autoantibodies and TSH elevation. These data are consistent with the hypothesis that decreased thyroid iodide transport and organification in smokers protect against development of autoantibodies, but predispose iodine-deficient individuals to hypothyroidism (52). If the curve relating iodine status to thyroid autoimmunity is sigmoidal, less frequent development of thyroid autoantibodies in smokers would be most dramatic in individuals with normal dietary iodine and would be difficult to show in extremes of low or high dietary iodine, where the risk plateaus.
A Swedish longitudinal study showed increased hypothyroidism among exsmokers and no association in active smokers, but adjustment for confounders was not reported (31). This raises the possibility that smokers are more likely to be diagnosed and treated than have a decreased likelihood of disease. Two characteristics of the present study argue against this explanation: first, the rate of smoking among subjects taking thyroid-altering medications was lower than that in the remaining persons; second, analysis of former smokers demonstrated no statistically significant associations with thyroid abnormalities.
Multiple studies from iodine sufficient areas, largely case-control in design, have failed to detect associations between cigarette smoke and hypothyroidism (21, 23, 33). The discrepancy between those findings and the present ones may relate to variations in study design. Many of the studies assessed the effect of smoking in individuals already exhibiting the presence of antibodies. In NHANES III, an association between smoking and lower prevalence of TSH elevation persisted among subjects with positive antibodies. Furthermore, several prior studies assessed the rate of smoking in individuals brought to medical attention for hypothyroidism; a strength of the NHANES III analysis is its focus on undiagnosed disease in the population. It is possible that smokers may be less likely to develop antibodies or TSH elevation, but that once these abnormalities present, they are associated with a more severe, complicated course than in nonsmokers (30). Our findings of an estimated 80% greater odds of thyroid autoantibodies in smokers compared with nonsmokers taking thyroid-related medications would be consistent with this hypothesis (although this was not statistically significant).
Because smoking is negatively associated with both thyroid autoantibody production and serum TSH elevation, its effects are unlikely to be attributable to an altered pituitary set-point alone. Smoking appears to be associated with an alteration of the autoimmune process widely believed to underlie most spontaneous thyroid gland dysfunction. Decreased thyroid autoimmunity may result from smoke’s interference with iodide transport and organification, decreased TSH secretion, or smoke’s effects on immune function. Several previous studies have associated cigarette smoking with decreased humoral and cell-mediated immunity and inhibited prostaglandin synthesis (53, 54, 55). This has been thought to be the basis for the possible protective effect of smoking on joint inflammation in early stages of rheumatoid arthritis (56). Similarly, active smoking has been found to be associated with lower incidence and severity of ulcerative colitis (57). Decreased TSH secretion in smokers and an antiestrogenic effect of smoke have been previously postulated to explain the lower incidence of thyroid cancer reported in smokers (58, 59). It is conceivable that these effects could protect the gland from the development of thyroid autoantibodies. In addition to its effects on autoimmunity, tobacco smoke appears to have additional, mild effects on hormonogenesis. The greater frequency of low normal TSH concentrations in smokers may be due to the mild thyroid autonomy and growth seen in smokers or to an altered pituitary set-point. The lower prevalence of elevated TSH concentrations may be due to both the effects of smoke on hormonogenesis and the decreased autoimmune hypothyroidism in smokers. Strieder and colleagues (34) hypothesized that decreased autoimmune destruction in smokers at risk for Graves’ disease may explain others’ findings of increased hyperthyroidism among active smoking relatives of individuals with autoimmune thyroid disease. It is unclear whether this would apply to individuals not at increased genetic risk for autoimmune thyroid disease.
In conclusion, active tobacco smoke exposure is associated with a lower risk of having thyroid autoantibodies and hypothyroidism in the general, noninstitutionalized, U.S. population evaluated in NHANES III. It is also associated with a higher risk of having low normal TSH concentrations. Our findings and those of previous investigators may be best explained by distinct effects of smoking on the expression of thyroid disease: an immunomodulatory action of smoke that reduces the likelihood of developing thyroid gland-specific autoimmunity; a mild stimulatory effect of smoke on thyroid hormonogenesis, which is less apparent once autoimmune thyroiditis has been initiated; and an inhibitory of effect of smoke on iodide transport and organification, which is more apparent in iodine deficiency. Future longitudinal studies could clarify these complex relationships between smoke exposure and thyroid disorders. Animal models could also help define the underlying pathogenic processes. Our findings suggest that it may be worthwhile to evaluate experimentally whether the elimination of smoke exposure could help prevent or ameliorate abnormalities of TSH suggestive of mild hyperthyroidism. Furthermore, the results may justify research to identify a putative ingredient in smoke capable of modulating autoimmune responses.
|
Footnotes |
|---|
Abbreviations: CI, Confidence interval; OR, odds ratio; TgAb, antithyroglobulin antibody; TPOAb, thyroperoxidase antibody.
Received March 2, 2004.
Accepted September 7, 2004.
|
References |
|---|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
therapy. Clin Endocrinol (Oxf) 44:635–642[CrossRef][Medline]
This article has been cited by other articles:
 |
|
 |
|
  C. L. Meinhold, E. Ron, S. J. Schonfeld, B. H. Alexander, D. M. Freedman, M. S. Linet, and A. Berrington de Gonzalez Nonradiation Risk Factors for Thyroid Cancer in the US Radiologic Technologists Study Am. J. Epidemiol., January 15, 2010; 171(2): 242 - 252. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Boas, J. L. Forman, A. Juul, U. Feldt-Rasmussen, N. E. Skakkebaek, L. Hilsted, M. Chellakooty, T. Larsen, J. F. Larsen, J. H Petersen, et al. Narrow intra-individual variation of maternal thyroid function in pregnancy based on a longitudinal study on 132 women Eur. J. Endocrinol., December 1, 2009; 161(6): 903 - 910. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Effraimidis, J. G. P. Tijssen, and W. M. Wiersinga Discontinuation of Smoking Increases the Risk for Developing Thyroid Peroxidase Antibodies and/or Thyroglobulin Antibodies: A Prospective Study J. Clin. Endocrinol. Metab., April 1, 2009; 94(4): 1324 - 1328. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Shields, A. Hill, M. Bilous, B. Knight, A. T. Hattersley, R. W. Bilous, and B. Vaidya Cigarette Smoking during Pregnancy Is Associated with Alterations in Maternal and Fetal Thyroid Function J. Clin. Endocrinol. Metab., February 1, 2009; 94(2): 570 - 574. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. O. Asvold, T. Bjoro, T. I. L. Nilsen, and L. J. Vatten Tobacco Smoking and Thyroid Function: A Population-Based Study Arch Intern Med, July 9, 2007; 167(13): 1428 - 1432. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Volzke, C. Schwahn, H. Wallaschofski, and M. Dorr The Association of Thyroid Dysfunction with All-Cause and Circulatory Mortality: Is There a Causal Relationship? J. Clin. Endocrinol. Metab., July 1, 2007; 92(7): 2421 - 2429. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. S. Metsios, A. D. Flouris, A. Z. Jamurtas, A. E. Carrillo, D. Kouretas, A. E. Germenis, K. Gourgoulianis, T. Kiropoulos, M. N. Tzatzarakis, A. M. Tsatsakis, et al. A Brief Exposure to Moderate Passive Smoke Increases Metabolism and Thyroid Hormone Secretion J. Clin. Endocrinol. Metab., January 1, 2007; 92(1): 208 - 211. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. H. Brix, P. S. Hansen, A. B. Rudbeck, J. B. Hansen, A. Skytthe, K. O. Kyvik, and L. Hegedus Low Birth Weight Is Not Associated with Thyroid Autoimmunity: A Population-Based Twin Study J. Clin. Endocrinol. Metab., September 1, 2006; 91(9): 3499 - 3502. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. E Krassas and W. Wiersinga Smoking and autoimmune thyroid disease: the plot thickens. Eur. J. Endocrinol., June 1, 2006; 154(6): 777 - 780. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
