Pregnancy is associated with various modulations of the

Transcription

1 ORIGINAL Endocrine ARTICLE Research Development of Autoimmune Overt Hypothyroidism Is Highly Associated With Live Births and Induced Abortions but Only in Premenopausal Women Allan Carlé, Inge Bülow Pedersen, Nils Knudsen, Hans Perrild, Lars Ovesen, Lone Banke Rasmussen, and Peter Laurberg Department of Endocrinology (A.C., I.B.P., P.L.), Aalborg University Hospital, DK-9000 Aalborg, Denmark; Endocrine Unit (N.K., H.P.), Medical Clinic I, Bispebjerg Hospital, DK-2400 Copenhagen, Denmark; Department of Internal Medicine (L.O.), Slagelse Hospital, DK-4200 Slagelse, Denmark; and National Food Institute (L.B.R.), Technical University of Denmark, DK-2800 Copenhagen, Denmark Context: The 1-year postpartum period is often accompanied by increased risk for thyroid disease. Objective: The objective of the study was to investigate the role of reproductive risk factors in the development of autoimmune overt hypothyroidism in the years after the 1-year postpartum period. Design, Setting, and Subjects: In a population study, we included Danish women with new autoimmune overt hypothyroidism not diagnosed within the first year after a pregnancy (n 117; median age 53.0 y) and age- and region-matched euthyroid controls from the same population (n 468). Main Outcome Measures: In conditional multivariate logistic regression models, we analyzed the associations between the development of autoimmune hypothyroidism and age at menarche/ menopause, years of menstruations, pregnancies, spontaneous and induced abortions, live births, and years on oral contraceptives and postmenopausal hormone replacement therapy, also taking various possible confounders into account. Results: In multivariate regression models with no event as reference, the odds ratios (ORs) for hypothyroidism [95% confidence interval (CI)] after one/two/three or more live births were 1.72 ( )/3.12 ( )/4.51 ( ) and 1.02 ( )/2.70 ( ) after one/two or more induced abortions. Findings were valid only for women having hypothyroidism diagnosed before the age of 55 years. We found no association between disease development and other reproductive risk factors investigated. Conclusions: Previous live births and induced abortions were major risk factors for the development of autoimmune overt hypothyroidism in women aged up to 55 years. The increased risk for hypothyroidism after giving birth extends longer than just to the 1-year postpartum period, and numbers of previous pregnancies should be taken into account when evaluating the risk of hypothyroidism in a young women. (J Clin Endocrinol Metab 99: , 2014) ISSN Print X ISSN Online Printed in U.S.A. Copyright 2014 by the Endocrine Society Received December 20, Accepted March 10, First Published Online April 2, 2014 Pregnancy is associated with various modulations of the immune system that may influence the risk for development of autoimmune disorders. During pregnancy, a lowering of the risk has been reported for some autoimmune diseases such as Graves disease and autoimmune hypothyroidism (1). On the other hand, the risk of autoimmune disease after delivery (1, 2) and after abortion (3) may be even higher than it was before pregnancy. How- Abbreviations: BMI, body mass index; CI, confidence interval; DanThyr, Danish Investigation on Iodine Intake and Thyroid Diseases; HRT, postmenopausal estrogen hormone replacement therapy; OC, oral contraceptive; OR, odds ratio; TgAb, thyroglobulin antibody; TPOAb, thyroid peroxidase antibody. doi: /jc J Clin Endocrinol Metab, June 2014, 99(6): jcem.endojournals.org 2241

2 2242 Carlé et al Live Birth and Induced Abortions and Hypothyroidism J Clin Endocrinol Metab, June 2014, 99(6): ever, not all studies agree on the increased risk of autoimmune diseases after delivery (2 4) or after abortion (4), and as shown by Jorgensen et al (2), different autoimmune diseases may respond unequally to parity. Use of oral contraceptives (OCs) in young age (5, 6) and estrogens in the postmenopausal period (6) has also been associated with altered risk for some autoimmune diseases, as is also the case for age at menarche (5 7) and menopause (6, 7). Several studies have evaluated the role of reproductive factors for later development of thyroid disease (2, 8 16) or the occurrence of thyroid autoantibodies in serum (7 9, 17 20). Although a positive association has been suggested in some reports, no firm conclusion has been reached. Studies had variable study designs and definitions of thyroid disease. It may be of note that the first person to be treated for hypothyroidism with thyroid hormone by Murray in 1891 (21) was a 46-year-old woman who had given birth to nine children. Moreover, it has been shown that women who had an episode of transient postpartum thyroiditis are at risk of developing permanent hypothyroidism within some years (22). To study the associations between various reproductive risk factors in women and development of autoimmune overt hypothyroidism in detail, we used population-based data collected as part of the monitoring of the Danish iodization program (23). Subjects and Methods The Danish Investigation of Iodine Intake and Thyroid Diseases (DanThyr) was established in We registered all women newly diagnosed with spontaneously developed autoimmune overt hypothyroidism in a population by a prospective monitoring of all thyroid function tests performed. The study period was March 1997 to December 2000, before iodization of salt became mandatory in Denmark. This encompassed person-years of observation (women: person-years). For case-control analyses, we simultaneously recruited randomly selected civilians from the same population to participate in two cross-sectional surveys. Patients Two well-defined geographical areas were chosen for the present study, an area in and around the city of Aalborg, Denmark, with moderate iodine deficiency [population under surveillance, n ; of these were women; urinary iodine excretion 45 g/l in subjects taking no mineral supplements (24)] and in Copenhagen with only mild iodine deficiency [n ; of these were women; urinary iodine excretion 61 g/l (24)]. A register was linked to databases of the four diagnostic laboratories responsible for all thyroid function testing performed in the two study areas. All subjects identified by the register for the first time with an elevated serum TSH in combination with a low T 4 estimate were individually evaluated to verify or disprove new overt autoimmune hypothyroidism. We contacted the physician requesting the blood sample and scrutinized all available hospital registers and records. Incident hypothyroidism was confirmed if patients were prescribed T 4 treatment or a second blood test confirmed untreated overt hypothyroidism a minimum of three weeks later. Hypothyroidism was classified into well-defined nosological subtypes (25), and only women diagnosed with primary autoimmune hypothyroidism were considered for the present study. Patients who had given birth within 1 year prior to diagnosing hypothyroidism were not included in the present study but were classified as suffering from postpartum thyroid dysfunction (25). In the study period, we successively identified 460 women newly diagnosed with primary autoimmune hypothyroidism (Figure 1). In selected periods, in which staff was available, we invited patients with autoimmune hypothyroidism (n 202, mostly women 70 y of age) to participate in a comprehensive investigation program as described in detail previously (26). Of those invited, 122 (60.4%) accepted and participated. We have previously described the surveillance program and the use of a register linked to laboratory databases (27), the diagnostic criteria used to identify patients with overt hypothy- Non-autoimmune hypothyroidism (n=77) Not invited (n=258) Non participants (n=80) 4 control subjects not available (n=5) The Danish Investigation of Iodine Intake and Thyroid Diseases (DanThyr) Population, n = 275,376 (women) Study period: Laboratory surveillance 1,055,608 py. Overt hypothyroidism (n=537) Autoimmune hypothyroidism (n=460) Invited (n=202) Participants (n=122) Overt autoimmune hypothyroidism (n=117 women) Cross-sectional studies of random groups (n=4,242) Survey I: n=3,712 Survey II: n=530 No previous thyroid disease (n=4,059) Euthyroid subjects (n=3,839) Age-, sex-, and region-matched controls (n=468 women) Previous overt hypo- or hyperthyroidism (n=183) Abnormal TSH (n=220) Controls not selected (n=3,371) Figure 1. Flow chart depicting the origin and selection of cases identified and their controls recruited at the same time from the same geographical area under study.

3 doi: /jc jcem.endojournals.org 2243 roidism, the algorithm for final verification of incident autoimmune hypothyroidism, and the classification of hypothyroidism into various subtypes of disease (25). Controls For each patient, we included from two simultaneously ongoing surveys four randomly selected control women matched on age and region (Figure 1). Survey I, a cross-sectional population study comprising 3712 women, was performed in Aalborg and Copenhagen in (28). Women aged 18 22, 25 30, 40 45, and years participated. Control subjects falling outside these age categories were invited from the same population in 1998 (survey II; women, n 530). Controls for the present study were selected among survey participants with TSH between 0.2 and 5.0 mu/l and with no previous thyroid disease. We were able to find 468 female controls (survey I II; n ) matching 117 women diagnosed with autoimmune hypothyroidism. We were not able to identify four matching controls for the remaining five cases, who were excluded from the present study. Investigational program Patients and controls underwent the same program, which was performed by the same staff during the same period of time. The sources of data for the present study were the responses from patients and controls to the comprehensive questionnaires. Participants had their height and weight measured, and body mass index (BMI) was calculated as weight measured in kilograms/ (height in meters) 2. At the visit, a blood sample was drawn and stored at 20 C until all samples were analyzed in random order at study end. Questionnaires Participants filled out questionnaires concerning information on reproductive and other risk factors that may be associated with development of autoimmune hypothyroidism. The number of pregnancies, live births, and abortions (induced or spontaneous) were specified. Women also gave information on the use of premenopausal OCs and postmenopausal hormone replacement therapy (HRT). The time of first menstruation and menopause was answered, and reproductive span (years of menstruations) was calculated. We had full information on reproductive factors in the 3839 women with no current or previous thyroid disease (Figure 1). Of these, 1022 were postmenopausal, 943 specified age at menopause, and 888 (95.2%) reported that they had experienced menopause at or before the age of 55 years. Participants filled out questions on alcohol consumption (units per week). Smoking habits were asked for in terms of never, previous, or current smoking. Previous smokers were asked to specify time since quitting. Participants also gave information on family history of hypothyroidism and current and previous diseases and medication. We evaluated the influence of cardiovascular comorbidity (defined present in case of ever history of acute myocardial infarction, angina pectoris, cardiac arrhythmia, hypertension, or cerebral stroke) and noncardiovascular comorbidity (epilepsy, diabetes mellitus, asthma, chronic obstructive pulmonary disease, or gastrointestinal ulcer). Educational status was answered into five categories: basic school with no vocational education, vocational education up to 2 years, 3 4 years of vocational education, vocational education for more than 4 years, and less than vocational education. Blood specimen analyses Patients and controls had blood drawn at the investigation Thyroid autoantibodies (thyroid peroxidase antibody [TPOAb]; thyroglobulin antibody [TgAb]) were measured after study end (26, 29). Subjects with antibody concentration above the functional sensitivity given by the manufacturer (TPOAb: 30 ku/l, TgAb: 20 ku/l) were regarded as antibody positive. Serum TSH concentrations were measured in serum drawn from controls. For all patients with incident hypothyroidism, we successively registered serum TSH and T 4 concentrations when hypothyroidism was diagnosed, ie, before any treatment. Details on TSH and T 4 analyses have been given previously (25, 27). Control subjects had median (interquartile range) serum TSH of 1.21 mu/l ( ). The hypothyroid patients included in the present study had pretreatment median serum TSH of 47.3 mu/l ( ) and serum T 4 of 40.0 nmol/l ( ). Statistical analysis We used IBM Statistical Package for Social Sciences version 15.0 for calculations and for statistical analyses. Data with no Gaussian distribution were expressed with median and interquartile range (25% and 75% percentiles). Groups of subjects were compared using a Mann-Whitney U test, Pearson s 2 test, Fishers exact test, or independent t test. Associations between having autoimmune hypothyroidism diagnosed and various reproductive variables were analyzed in conditional uni- and multivariate logistic regression models. Associations were expressed as odds ratios (OR) with 95% confidence intervals (CIs). Values of P.05 or an OR with 95% CI not including 1.0 were regarded as statistically significant. A conditional regression analysis requires no missing values, and the 33 of 6435 questions (0.51%) regarding reproductive data left unanswered by participants were filled out by means of nearest neighbor imputation (30). Explanatory variables were tested and no multicollinearity was found for those variables, which were simultaneously included in the regression models. In the regression models, we explored one reproductive factor at a time comparing a specific reproductive event vs no event. After having excluded subjects with no reproductive events, tests for linear trend were calculated using the ordinal categories for pregnancies, live births, and abortions as continuous variables. We concomitantly included also as explanatory variables the following nonreproductive factors: family history of hypothyroidism (ever/never), alcohol consumption (abstainers vs 1U consumed per week), all-cause comorbidity (ever/never), and education (basic school plus up to 2 y ofvocational schooling [eg, store employees, carpenters, or mason] vs more). We have previously shown that recent smoking cessation (within 2 y prior to development of hypothyroidism) was strongly associated with the development of autoimmune hypothyroidism (31). In a regression analysis, we tested both the smoking variable traditionally used (never, current, and previous smoking) and the dichotomized variable (recent quitters within 2 y vs therest). At the age of 55 years, 95% of all healthy control women had experienced menopause. In the analyses, this age was used as a proxy variable for menopause status. Ethical approval This study was approved by regional ethics committees in North Jutland and Copenhagen. Registry permission was ob-

4 2244 Carlé et al Live Birth and Induced Abortions and Hypothyroidism J Clin Endocrinol Metab, June 2014, 99(6): tained from the Danish Data Protection Agency. All participants gave their written informed consent. No conflicts of interest have occurred during implementation or completion of the study. Results Baseline characteristics of cases and controls The characteristics of 117 women newly diagnosed with autoimmune hypothyroidism and their individually age- and region-matched 468 control women recruited from the same population at the same period of time are given in Table 1. Patients had on average a 6.8-kg higher body weight and a 5 times higher thyroid autoantibody prevalence at the time of diagnosis. In addition, they were 4 times more likely to be recent smoking quitters, were more often alcohol abstainers, and had a higher prevalence of comorbidity. Reproductive history and hypothyroidism Reproductive history in hypothyroid women and their euthyroid controls is given in Table 2. Women diagnosed with autoimmune hypothyroidism had more often been pregnant, had more live births, and had more induced abortions. No difference was observed with regard to the number of spontaneous abortions. Neither did we find any difference in mean age at menarche (cases vs controls: 13.6 vs 13.7, P.79) or menopause (48.7 vs 47.9, P.30) or median years of menstruations (33.5 vs 33.0, P.32). The role of estrogen use was also explored, but no difference was observed between cases and controls with regard to the use of OCs or postmenopausal HRT. To evaluate the possible role of confounders and to examine statistical interaction between various risk factors, we performed uni- and multivariate conditional logistic testing (Table 3). For each reproductive event (preg- Table 1. Characteristics of Autoimmune Hypothyroid Women and Matched Control Women Cases (n 117) a Controls (n 468) a P Value b Age, y 53.0 ( ) 52.7 ( ) NS Inhabitancy Aalborg (moderate ID) 71 (60.7) 284 (60.7) Copenhagen (mild ID) 46 (39.3) 184 (39.3) NS Weight, kg 73.1 ( ) 66.3 ( ).001 Height, m 1.64 ( ) 1.65 ( ).56 BMI, kg/m ( ) 24.4 ( ).001 TPOAb, ku/l 4,396 (759 9,593) TPOAb, 30 ku/l 111 (94.9) 93 (19.9).001 TgAb, ku/l 117 (31 935) TgAb, 20 ku/l 94 (80.3) 85 (18.2).001 Education Basic school and up to 2 y vocational training 80 (68.4) 302 (64.5) More 37 (31.6) 166 (35.5).43 Smoking history Never smoker 47 (40.2) 194 (41.5) Previous smoker Recent quitter ( 2 y) 17 (14.5) 15 (3.2) Old quitter ( 2 y) 18 (15.4) 94 (20.1) Current smoker 33 (28.2) 164 (35.0).001 c Alcohol intake, U/wk 0 (abstainer) 31 (26.5) 71 (15.2) (61.5) 311 (66.5) (6.8) 65 (13.9) 21 6 (5.1) 21 (4.5).004 d Comorbidity All causes 66 (56.4) 201 (42.9).009 Cardiovascular e 49 (41.9) 120 (25.6).001 Noncardiovascular f 17 (14.5) 81 (17.3).47 Abbreviations: NS, not significant; TPOAb, TPOAb positive. a Depicted are number of participants (percentage) or medians (interquartile range, 25% 75% range). Some data were missing for weight (n 1), height (n 1), BMI (n 1), TPOAb (n 1), TgAb (n 1), and time since smoking cessation (two cases and one control subject). b Univariate comparison between groups. c Recent quitting within 2 yr vs rest. d Abstainers vs 1 10 vs 11 or more alcohol units per week. e Questionnaire obtained information on myocardial infarction, angina pectoris, cardiac arrhythmia, hypertension, or cerebral stroke. f Questionnaire obtained information on epilepsy, diabetes mellitus, asthma, chronic obstructive pulmonary disease, or gastrointestinal ulcers.

5 doi: /jc jcem.endojournals.org 2245 Table 2. Reproductive History of Autoimmune Hypothyroid Women and Control Women Cases (n 117) a Controls (n 468) a P Value Fertile status Premenopausal 45 (38.5) 208 (44.4) Postmenopausal 72 (61.5) 260 (55.6) HRT 9 (7.7) 45 (9.6) HRT 63 (53.8) 215 (45.9).30 b No. of pregnancies, mean c 0 8 (6.8) 77 (16.5) 1 10 (8.5) 70 (15.0) 2 32 (27.4) 118 (25.2) 3 35 (29.9) 108 (23.1) 4 32 (27.4) 95 (20.3).012 b Live births, mean c 0 11 (9.4) 95 (20.3) 1 21 (17.9) 90 (19.2) 2 47 (40.2) 165 (35.3) 3 24 (20.5) 91 (19.4) 4 14 (12.0) 27 (5.8).017 b Abortions, spontaneous 0 95 (81.2) 376 (80.3) 1 20 (17.1) 78 (16.7) 2 1 (0.9) 10 (2.1) 3 1 (0.9) 4 (0.9).84 b Abortions, induced 0 81 (69.2) 355 (75.9) 1 21 (17.9) 86 (18.4) 2 8 (6.8) 23 (4.0) 3 7 (6.0) 4 (0.9).003 b OCs Ever 76 (65.0) 329 (70.4) Present use 4 (3.5) 33 (7.2) Previous use 72 (61.5) 296 (63.2) Never 41 (35.0) 138 (29.6).25 b a Depicted are number of participants (percentage). b Univariate comparison between groups. c Independent t test comparing variables with Gaussian distribution. nancy, live birth, induced and spontaneous abortion), we compared those having experienced an event with women with no event. Results from the various regression models confirmed the univariate associations. Because age and a number of reproductive events showed statistical interaction, we stratified patients and controls according to age (Figure 2). The statistically significant association between number of live births or induced abortions and development of hypothyroidism was confined to women aged up to 55 years, whereas no association was observed in participants being older than 55 years. In the group of younger women, the risk of hypothyroidism was associated with previous pregnancies [1 vs none, OR (95% CI): 14.2 ( ), live births (1 vs none, OR 9.57 ( ), and induced abortions (2 vs 0 1, OR 5.17 ( ); 1 vs none, OR 1.91 ( )]. The associations were confirmed in regression models including categories of menopausal status reproductive risk factors (data not shown). In the group of younger women aged up to 55 years, we explored whether several pregnancy exposures were associated with higher risk for developing hypothyroidism (Figure 3). Clearly a dose-response pattern was observed. Discussion Pregnancies and live births We found profound associations between pregnancies, live births, and induced abortions and the subsequent development of autoimmune hypothyroidism, but only in premenopausal women. The risk increased with the number of pregnancies. Greer et al (18) demonstrated that women who had been pregnant three or more times were 3 times more often TPOAb-positive compared with nulliparous women. In a Danish register study (2), a positive association was observed between parity and a hospital diagnosis of Hashimoto s thyroiditis but no stronger association in those who had several children. In an Australian study (9) with several end points, the number of pregnancies was associated with self-reported thyroid disease but not with the prevalence of thyroid autoantibodies or with the prevalence of a raised serum TSH. In the Study of Health in Pomerania, Friedrich et al (19) found positive associations between both thyroid autoantibodies and thyroid hypoechogenic ultrasound findings and previous pregnancies and deliveries, but other studies found no such associations (10, 11, 13, 17). In a previous DanThyr study of 3283 euthyroid female volunteers, Bülow Pedersen et al (17) found no association between previous pregnancy and the prevalence of TPOAb or TgAb. Moreover, Strieder et al (10) studied a heterogeneous group of patients with autoimmune thyroid disease (Graves disease, postpartum thyroid dysfunction, hypothyroidism, and silent thyroiditis) and found no association between disease and pregnancies or estrogen use. Similarly, Phillips et al (7) found no association between Hashimoto s thyroiditis and the number of pregnancies. Abortions We observed a positive association between the development of autoimmune hypothyroidism and induced abortions, whereas spontaneous abortion had no impact on later hypothyroidism. Spontaneous abortions are often caused by mutations or other failures taking place during the very early gestational period. On the other hand, the initial phase of pregnancy is normal in women having an induced abortion performed. Thus, it is to be anticipated that early pregnancy hormone and immunological changes had been less pronounced when spontaneous abortions had occurred.

6 2246 Carlé et al Live Birth and Induced Abortions and Hypothyroidism J Clin Endocrinol Metab, June 2014, 99(6): Table 3. Factors a Odds Ratios (95% Confidence Intervals) of Autoimmune Hypothyroidism in Women by Reproductive Risk Subjects (Cases/Controls), n Univariate Model Multivariate Model 1 b Multivariate Model 2 c Multivariate Model 3 d Pregnancies, n 0 8/77 1 (referent) 1 (referent) 1 (referent) 1 (referent) 1 10/ ( ) 1.98 ( ) 1.78 ( ) 1.72 ( ) 2 32/ ( ) e 4.03 ( ) e 3.63 ( ) f 3.12 ( ) f 3 67/ ( ) g 5.52 ( ) g 5.11 ( ) e 4.51 ( ) e P for trend h Live births 0 11/94 1 (referent) 1 (referent) 1 (referent) 1 (referent) 1 21/ ( ) f 2.54 ( ) 2.29 ( ) 2.03 ( ) 2 47/ ( ) e 3.81 ( ) e 3.67 ( ) e 3.33 ( ) f 3 38/ ( ) e 3.93 ( ) e 3.58 ( ) e 3.31 ( ) f P for trend h Abortions, induced 0 81/355 1 (referent) 1 (referent) 1 (referent) 1 (referent) 1 21/ ( ) 1.18 ( ) 1.12 ( ) 1.02 ( ) 2 15/ ( ) e 2.72 ( ) e 2.89 ( ) e 2.70 ( ) f Abortions, spontaneous 0 95/376 1 (referent) 1 (referent) 1 (referent) 1 (referent) 1 20/ ( ) 1.12 ( ) 1.08 ( ) 1.05 ( ) 2 2/ ( ) 0.53 ( ) 0.51 ( ) 0.60 ( ) a Conditional logistic uni- and multivariate regression performed with disease state (hypothyroidism vs control) as dependent variable. In both uniand multivariate regression, the various reproductive characteristics were explored one at a time, not simultaneously. b Explanatory variables included all-cause comorbidity, family history of hypothyroidism, and educational levels (basic school and 2 y additional schooling vs more). c Explanatory variables included all-cause comorbidity, family history of hypothyroidism, educational levels, alcohol consumption, and smoking habit (never, current, previous). d Explanatory variables included all-cause comorbidity, family history of hypothyroidism, educational levels, alcohol consumption, and smoking habit (recent quitting within 2yvsrest). e P.01. f P.05. g P.001. h Trend P values were calculated after exclusion of subjects with no reproductive events. Our positive finding between induced abortions and development of autoimmune thyroid disease is supported by Sgarbi et al (20), who investigated 675 women and reported both TPOAb and TgAb to be almost doubled in those who had experienced abortions. However, their study did not specify the type of abortion. In addition, Marqusee et al (32) reported on five women who experienced the combination of autoimmune thyroiditis and biochemical hyper- and/or hypothyroidism after both spontaneous and elective abortions. Reproductive span, menarche, and menopause In our study there was no association between development of autoimmune hypothyroidism and women s age at menarche or menopause. This is in accordance with a follow-up study of autoimmune diabetes patients performed by McCanlies et al (12). They compared euthyroid and subclinically hypothyroid patients and reported no association between Hashimoto s hypothyroidism and age of neither menarche nor menopause. On the other side, Phillips et al (7) observed a significant association between Hashimoto s thyroiditis and both early menarche and late menopause, suggesting that the number of reproductive years might play a role in disease development. Estrogens and hypothyroidism No association was observed between autoimmune hypothyroidism and the use of OCs in young age or HRT among postmenopausal women. In a previous DanThyr study (33), the use of estrogens was associated with a somewhat higher serum TSH, which is in accordance with the higher need of L-thyroxine for replacement when estrogens are taken (34). In accordance with our findings, Vestergaard et al (15) studied 628 patients diagnosed with autoimmune hypothyroidism and found no association between disease development and the use of OCs or postmenopausal HRT. Strieder et al (8) also reported no association between estrogen use and the development of autoimmune hypothyroidism studying 803 subjects pre-

7 doi: /jc jcem.endojournals.org p=0.027 p= Odds ratio (CI-95%) Figure 2. Association between reproductive factors and development of autoimmune overt hypothyroidism. Results are depicted as ORs and are shown for young ( 55 y old, mostly premenopausal) and old patients ( 55 y old, mainly postmenopausal). In the group of younger women, ORs for autoimmune hypothyroidism were; one/two/three or more pregnancies: 8.14 ( )/26.8 ( )/50.9 ( ); one/two/three or more live births: 6.98 ( )/13.5 ( )/ 13.2 ( ), one/two/three or more induced abortions: 1.27 ( )/5.55 ( ); all other associations depicted showed statistical nonsignificance. disposed to autoimmune thyroid disease. Three other studies found no association between estrogen use and development of Hashimoto s thyroiditis (7, 14, 35). Finally, Massoudi et al (16) found no association between HRT and prevalence of thyroid autoantibodies or subclinical hypothyroidism Number of pregnancies Figure 3. Association between the number of pregnancies and the risk of developing hypothyroidism in young women aged up to 55 years. Women with one pregnancy constituted the reference group in the univariate analysis. The OR (95% CI) for developing hypothyroidism was 2.90 ( ) after two to four pregnancies and 10.1 ( ) after five pregnancies or more. The P values indicate results of 2 testing between groups. Pathomechanisms involved Our study reveals that repeated pregnancy exposure may be an important risk factor for the development of autoimmune hypothyroidism before but not after the menopause. Because pregnancy does not increase the probability of harboring circulating thyroid autoantibodies (7, 9, 10, 17, 18, 20), we speculate that reproductive factors do not initiate the autoimmune process but may accelerate or even precipitate early hypothyroid disease development in subjects already harboring autoimmune thyroiditis. Some degree of autoimmune reaction to thyroid tissue is present in approximately 30% of all 20- to 50-year-old women of Caucasian ancestry as judged from lymphocytic infiltration of the thyroid gland (36). It is well established that profound changes of thyroid autoimmunity occurs during pregnancy, postpartum, and presumably also after abortion. We find it likely that this sequence of amelioration by worsening may involve a risk of permanent damage to the thyroid already affected by autoimmunity. Strengths and limitations The strength of the study is the inclusion of overtly autoimmune hypothyroid patients because they were diagnosed in the population. Many studies have dealt with participants from cross-sectional studies (7, 9, 10, 17, 19, 20), patients highly disposed for thyroid disease (14), pregnant women undergoing screening (11, 13, 18), follow-up study of type 1 diabetes patients (12), laboratory registers (7), general practitioners medical records (35), whereas others included patients referred to hospital (2, 15) which may introduce bias (37). The homogeneous group of patients suffering from overt hypothyroidism in the present study is in contrast to other studies including also patients with subclinical hypothyroidism (9, 11, 13, 33), thyroid autoantibody positive subjects (7, 17, 18, 20), subjects with ultrasound findings of a hypoechogenic thyroid gland (19), self-reported thyroid disease (9), valid (15) or dubious (2) hospital register diagnoses, or even a mixture of these conditions (8, 10, 12, 19). The case-control design does not directly allow for any causal conclusions. However, a number of criteria suggesting causal association such as temporality and dose dependency are met in the present study (38). Further-

8 2248 Carlé et al Live Birth and Induced Abortions and Hypothyroidism J Clin Endocrinol Metab, June 2014, 99(6): more, there are biological plausible explanations for our findings. Our matched case-control design offers the possibility of adjusting for many potential confounders, which is important, because reproductive history may be associated with smoking habits (19), alcohol intake (19), and education (19). A limitation of our study is the 60% participation rate. A group of patients were not invited (n 258) due to high age (n 104, median age 79.9 y), general practitioners information on patient disability (n 51, age 64.0 y), and no staff available for patient investigation (n 78, age 64.0 y). Some patients died shortly after having hypothyroidism diagnosed (n 14, age 77.3 y), whereas 11 patients were not invited due to a mixture of other reasons. Among those invited, some self-selection was present because participants (n 122, age 53.4 y) were younger than those declining participation (n 80, age 66.0 y). This may hamper the external validity of the study. Finally, CIs were rather broad due to statistical analyses after stratification into groups containing a relatively low number of patients. Information on reproductive factors was obtained by questionnaires. Previous studies have revealed robust validity and reproducibility of reproductive data collected in this way and have shown a moderate to good agreement between information obtained from various registers and self-reported information on pregnancies (39, 40), deliveries (39, 40), abortions (41), age at menarche (40), age at menopause (42), use of OCs (39, 40), and postmenopausal HRT (39). Unfortunately, we had no information on breast-feeding or age at each reproductive event. In conclusion, previous pregnancy plays a major role in development of autoimmune overt hypothyroidism in premenopausal women, and the number of previous pregnancies should be taken into account when evaluating the risk of hypothyroidism in a young women. Acknowledgments We are indebted to the general practitioners in Copenhagen and Northern Jutland and to the clinical chemical laboratories at Aalborg Hospital, Bispebjerg Hospital, and Frederiksberg Hospital as well as the Laboratory of General Practitioners in Copenhagen for their helpful collaboration in identifying patients. Address all correspondence and requests for reprints to: Allan Carlé, MD, PhD, Department of Endocrinology, Aalborg University Hospital, DK-9000 Aalborg, Denmark. carle@dadlnet.dk. This study was part of DanThyr, and it was supported by the following grants: the IMK General Foundation; the Danish Council for Independent Research; the Ministry of Food, Agriculture, and Fisheries; the Danish Agency for Science, Technology, and Innovation; the Institute for Clinical Medicine, University of Aarhus; and the Aase og Ejnar Danielsens Foundation. Disclosure Summary: The authors have nothing to disclose. References 1. Davies TF. The thyroid immunology of the postpartum period. Thyroid. 1999;9: Jorgensen KT, Pedersen BV, Nielsen NM, Jacobsen S, Frisch M. Childbirths and risk of female predominant and other autoimmune diseases in a population-based Danish cohort. J Autoimmun. 2012; 38:J81 J Ulff-Moller CJ, Jorgensen KT, Pedersen BV, Nielsen NM, Frisch M. Reproductive factors and risk of systemic lupus erythematosus: nationwide cohort study in Denmark. J Rheumatol. 2009;36: Jorgensen KT, Pedersen BV, Jacobsen S, Biggar RJ, Frisch M. National cohort study of reproductive risk factors for rheumatoid arthritis in Denmark: a role for hyperemesis, gestational hypertension and pre-eclampsia? Ann Rheum Dis. 2010;69: Pedersen M, Jacobsen S, Klarlund M, et al. Environmental risk factors differ between rheumatoid arthritis with and without autoantibodies against cyclic citrullinated peptides. Arthritis Res Ther. 2006;8:R Costenbader KH, Feskanich D, Stampfer MJ, Karlson EW. Reproductive and menopausal factors and risk of systemic lupus erythematosus in women. Arthritis Rheum. 2007;56: Phillips DI, Lazarus JH, Butland BK. The influence of pregnancy and reproductive span on the occurrence of autoimmune thyroiditis. Clin Endocrinol (Oxf). 1990;32: Strieder TG, Prummel MF, Tijssen JG, Endert E, Wiersinga WM. Risk factors for and prevalence of thyroid disorders in a cross-sectional study among healthy female relatives of patients with autoimmune thyroid disease. Clin Endocrinol (Oxf). 2003;59: Walsh JP, Bremner AP, Bulsara MK, et al. Parity and the risk of autoimmune thyroid disease: a community-based study. J Clin Endocrinol Metab. 2005;90: Strieder TG, Tijssen JG, Wenzel BE, Endert E, Wiersinga WM. Prediction of progression to overt hypothyroidism or hyperthyroidism in female relatives of patients with autoimmune thyroid disease using the Thyroid Events Amsterdam (THEA) score. Arch Intern Med. 2008;168: Casey BM, Dashe JS, Spong CY, McIntire DD, Leveno KJ, Cunningham GF. Perinatal significance of isolated maternal hypothyroxinemia identified in the first half of pregnancy. Obstet Gynecol. 2007;109: McCanlies E, O Leary LA, Foley TP, et al. Hashimoto s thyroiditis and insulin-dependent diabetes mellitus: differences among individuals with and without abnormal thyroid function. J Clin Endocrinol Metab. 1998;83: Casey BM, Dashe JS, Wells CE, et al. Subclinical hypothyroidism and pregnancy outcomes. Obstet Gynecol. 2005;105: Effraimidis G, Strieder TG, Tijssen JG, Wiersinga WM. Natural history of the transition from euthyroidism to overt autoimmune hypo- or hyperthyroidism: a prospective study. Eur J Endocrinol. 2011;164: Vestergaard P, Rejnmark L, Weeke J, et al. Smoking as a risk factor for Graves disease, toxic nodular goiter, and autoimmune hypothyroidism. Thyroid. 2002;12: Massoudi MS, Meilahn EN, Orchard TJ, et al. Prevalence of thyroid antibodies among healthy middle-aged women. Findings from the thyroid study in healthy women. Ann Epidemiol. 1995;5: Bulow PI, Laurberg P, Knudsen N, et al. Lack of association between thyroid autoantibodies and parity in a population study argues against microchimerism as a trigger of thyroid autoimmunity. Eur J Endocrinol. 2006;154:39 45.

9 doi: /jc jcem.endojournals.org Greer LG, Casey BM, Halvorson LM, Spong CY, McIntire DD, Cunningham FG. Antithyroid antibodies and parity: further evidence for microchimerism in autoimmune thyroid disease. Am J Obstet Gynecol. 2011;205: Friedrich N, Schwarz S, Thonack J, John U, Wallaschofski H, Volzke H. Association between parity and autoimmune thyroiditis in a general female population. Autoimmunity. 2008;41: Sgarbi JA, Kasamatsu TS, Matsumura LK, Maciel RM. Parity is not related to autoimmune thyroid disease in a population-based study of Japanese-Brazilians. Thyroid. 2010;20: Murray GR. The life history of the first case of myxoedema treated by thyroid extract. Br Med J. 1920; Premawardhana LD, Parkes AB, Ammari F, et al. Postpartum thyroiditis and long-term thyroid status: prognostic influence of thyroid peroxidase antibodies and ultrasound echogenicity. J Clin Endocrinol Metab. 2000;85: Laurberg P, Jorgensen T, Perrild H, et al. The Danish investigation on iodine intake and thyroid disease, DanThyr: status and perspectives. Eur J Endocrinol. 2006;155: Rasmussen LB, Ovesen L, Bulow I, et al. Dietary iodine intake and urinary iodine excretion in a Danish population: effect of geography, supplements and food choice. Br J Nutr. 2002;87: Carle A, Laurberg P, Pedersen IB, et al. Epidemiology of subtypes of hypothyroidism in Denmark. Eur J Endocrinol. 2006;154: Carle A, Laurberg P, Knudsen N, et al. Thyroid peroxidase and thyroglobulin auto-antibodies in patients with newly diagnosed overt hypothyroidism. Autoimmunity. 2006;39: Pedersen IB, Laurberg P, Arnfred T, et al. Surveillance of disease frequency in a population by linkage to diagnostic laboratory databases. A system for monitoring the incidences of hyper- and hypothyroidism as part of the Danish iodine supplementation program. Comput Methods Programs Biomed. 2002;67: Knudsen N, Bulow I, Jorgensen T, Laurberg P, Ovesen L, Perrild H. Comparative study of thyroid function and types of thyroid dysfunction in two areas in Denmark with slightly different iodine status. Eur J Endocrinol. 2000;143: Pedersen IB, Knudsen N, Jorgensen T, Perrild H, Ovesen L, Laurberg P. Thyroid peroxidase and thyroglobulin autoantibodies in a large survey of populations with mild and moderate iodine deficiency. Clin Endocrinol (Oxf). 2003;58: Chen J, Shao J. Nearest Neighbor Imputation for Survey Data. J Offic Stat. 2000;16: Carle A, Pedersen IB, Knudsen N, et al. Smoking cessation is followed by a sharp but transient rise in the incidence of overt autoimmune hypothyroidism a population-based case-control study. Clin Endocrinol (Oxf). 2012;77(5): Marqusee E, Hill JA, Mandel SJ. Thyroiditis after pregnancy loss. J Clin Endocrinol Metab. 1997;82: Knudsen N, Bulow I, Laurberg P, Perrild H, Ovesen L, Jorgensen T. Low goitre prevalence among users of oral contraceptives in a population sample of 3712 women. Clin Endocrinol (Oxf). 2002;57: Arafah BM. Increased need for thyroxine in women with hypothyroidism during estrogen therapy. N Engl J Med. 2001;344: Frank P, Kay CR. Incidence of thyroid disease associated with oral contraceptives. Br Med J. 1978;2: Okayasu I, Hara Y, Nakamura K, Rose NR. Racial and age-related differences in incidence and severity of focal autoimmune thyroiditis. Am J Clin Pathol. 1994;101: Carle A, Laurberg P, Pedersen IB, et al. Mainly the younger hypothyroid patients are referred to hospital evidence for referral bias. J Clin Epidemiol. 2009;62: Hill AB. The environment and disease: association or causation? Proc R Soc Med. 1965;58: Lucas R, Azevedo A, Barros H. Self-reported data on reproductive variables were reliable among postmenopausal women. J Clin Epidemiol. 2008;61: Lin SS, Glaser SL, Stewart SL. Reliability of self-reported reproductive factors and childhood social class indicators in a case-control study in women. Ann Epidemiol. 2002;12: Wilcox AJ, Horney LF. Accuracy of spontaneous abortion recall. Am J Epidemiol. 1984;120: den Tonkelaar I. Validity and reproducibility of self-reported age at menopause in women participating in the DOM-project. Maturitas. 1997;27: