ethnic differences in maternal thyroid parameters during pregnancy: the generation r study

Ethnic Differences in Maternal Thyroid Parametersduring Pregnancy: The Generation R Study

Tim I. M. Korevaar,* Marco Medici,* Yolanda B. de Rijke, Willy Visser,Sabine M. P. F. de Muinck Keizer-Schrama, Vincent W. V. Jaddoe,Albert Hofman, H. Alec Ross, W. Edward Visser, Herbert Hooijkaas,Eric A. P. Steegers, Henning Tiemeier, Jacoba J. Bongers-Schokking,Theo J. Visser, and Robin P. Peeters

The Generation R Study Group (T.I.M.K., M.M., V.W.V.J., W.E.V.) and Departments of Internal Medicine(T.I.M.K., M.M., Y.B.d.R., T.J.V., R.P.P.), Clinical Chemistry (Y.B.R.), Obstetrics and Gynecology (W.V.,E.A.P.S.), Epidemiology (V.W.V.J., A.H., H.T.), and Immunology (H.H.), Erasmus Medical Center, 3015GE, Rotterdam, The Netherlands; Departments of Pediatrics (S.M.P.F.d.M.K.-S., J.J.B.-S.) and Child andAdolescent Psychiatry (H.T.), Erasmus Medical Center, Sophia Children’s Hospital, 3015 GJ. Rotterdam,The Netherlands; and Department of Laboratory Medicine (A.R.), Radboud University, Nijmegen MedicalCenter, 6525 GA, Nijmegen, The Netherlands

Context: Abnormal maternal thyroid function during pregnancy is associated with various com-plications. International guidelines advocate the use of population-based trimester-specific ref-erence ranges for thyroid function tests. When unavailable, an upper TSH limit of 2.5 for the firsttrimester and 3.0 mU/L for the second and third trimesters is recommended. Although interindi-vidual differences in thyroid function tests can partially be explained by ethnicity, data on theinfluence of ethnicity on TSH and free T4 reference ranges during pregnancy are sparse.

Design: Serum TSH, free T4, T4, and TPO-antibody levels were determined during early pregnancyin 3944 women from the Generation R study, Rotterdam, The Netherlands.

Results: The study population consisted of 2765 Dutch, 308 Moroccan, 421 Turkish, and 450 Suri-namese women. Mean TSH levels were higher in Dutch and Turkish women than in Moroccan orSurinamese women (1.50–1.48 vs 1.29–1.33 mU/L; P � .01). Although no differences in free T4 wereseen, T4 was lowest in Dutch women (142 vs 150–156 nmol/L; P � .01). Turkish women had thehighest frequency of TPO-antibody positivity (9.3% vs 5.0–5.8%; P � .05) and of elevated TSH levelsin the second trimester (11.0% vs 3.8–7.3%; P � .01). A comparison of disease prevalence betweena population-based vs an ethnicity-specific reference range changed the diagnosis for 18% ofwomen who were initially found to have abnormal thyroid function test results.

Conclusions: We show ethnic differences in serum TSH, T4, and TPO-antibody positivity and foundsignificant diagnostic discrepancies depending on whether population or ethnicity-specific refer-ence ranges were used to diagnose thyroid disease. (J Clin Endocrinol Metab 98: 3678–3686, 2013)

Abnormal maternal thyroid function during pregnancyis associated with various maternal and child com-

plications such as preeclampsia, miscarriage, preterm de-livery, and impaired neurodevelopment of the child (1–3).Recent guidelines by The Endocrine Society and the Amer-

ican Thyroid Association (ATA) advocate the use of pop-ulation-based trimester-specific reference ranges to diag-nose thyroid dysfunction in pregnant women. Whentrimester-specific reference ranges are not available in thelaboratory, upper TSH limits of 2.5 mU/L during the first

ISSN Print 0021-972X ISSN Online 1945-7197Printed in U.S.A.Copyright © 2013 by The Endocrine SocietyReceived April 18, 2013. Accepted July 1, 2013.First Published Online July 8, 2013

* T.I.M.K. and M.M. contributed equally to the study.Abbreviations: BMI, body mass index; ESRR, ethnicity-specific reference ranges; fT4, freeT4; hCG, human chorionic gonadotrophin; SES, socioeconomic status; TPOAb, thyroidperoxidase antibody; TPRR, total population reference ranges; WHO, World HealthOrganization.

O R I G I N A L A R T I C L E

E n d o c r i n e C a r e

3678 jcem.endojournals.org J Clin Endocrinol Metab, September 2013, 98(9):3678–3686 doi: 10.1210/jc.2013-2005

trimester and 3.0 mU/L during the second and third tri-mesters are recommended (4, 5). These recommendationsare mainly based on large studies conducted in divergentpopulations from America, Europe, China, and India,which can nowadays be considered multiethnic as a resultof increased migration (6–11). A number of studies haveshown that interindividual differences in thyroid hormonelevels may be explained, at least partially, by ethnic back-ground (12–14). Therefore, even within trimester-specificreference ranges, differences may exist as a result ofmultiethnicity.

Only a few studies have analyzed the effect of ethnicityon thyroid function test results during pregnancy. A smallstudy in 589 pregnant women demonstrated that AfricanAmerican women have lower TSH values than whitewomen (15). Subsequently, La’ulu and Roberts (16, 17)reported that reference range values for thyroid parame-ters may differ between Asian, white, black, and HispanicAmericans. Benhadi et al (18) demonstrated significantlylower mean TSH levels in pregnant Dutch women com-pared with those in Turkish, Moroccan, and Surinamesepregnant women. In contrast, a study by Pearce et al (19)showed that ethnicity was not a contributing factor toeither TSH, free T4 (FT4), or T4 in pregnancy (19), but thismay have been due to the number of groups compared anda relatively small sample size.

These ethnic differences between different pregnantpopulations underline the importance of calculating pop-ulation-specific reference ranges during pregnancy andsuggest that it may not be optimal to apply the same upperlimit for TSH during pregnancy for various populationsworldwide. Because most populations nowadays are con-sidered multiethnic, we investigated the consequences ofcalculating ethnicity-specific reference ranges (ESRR) forthe diagnosis of thyroid disease in a large, multiethnicpopulation of pregnant women from Rotterdam, TheNetherlands. To exclude an interfering role for iodine de-ficiency in specific ethnic groups, we also analyzed urinaryiodine levels in a subset of pregnant women.

Materials and Methods

DesignThis study was embedded in the Generation R Study, a pop-

ulation-based cohort from early fetal life onward in the multi-ethnic city of Rotterdam, The Netherlands, which has been de-scribed in detail previously (20). Written informed consent wasobtained from all adult participants.

Population for analysesData on TSH, FT4, and T4 were available for 4192 pregnant

women and data on thyroid peroxidase antibodies (TPOAbs)were available for 3928 Dutch, Moroccan, Turkish, and Suri-

namese pregnant women. Women having twin pregnancies (n �128), women having preexisting thyroid disease (n � 35),women using thyroid (interfering) medication (n � 32), andwomen undergoing fertility treatment (n � 53) were excluded.Data on ethnicity and ethnic origin were derived by question-naires. Ethnicity was determined by country of origin, which wasdefined according to the classification of Statistics Netherlands(20); ethnic origin was determined by common ancestry. Thefinal population comprised 3944 women who were included inone or more analyses.

Thyroid parametersMaternal serum samples were obtained in early pregnancy

(mean, 13.4 weeks; SD, 2.0). Plain tubes were centrifuged, andserum was stored at �80°C. TSH, FT4, and T4 were determinedin maternal serum samples using chemiluminescence assays (Vit-ros ECI; Ortho Clinical Diagnostics). The intra- and interassaycoefficients of variation were �5.4% for FT4 at a range of 14.3to 25.0 pmol/L and �6.4% for T4 at a range of 94 to 151 nmol/L.For TSH specifically, the intra- and interassay coefficients ofvariation were �4.1% at a range of 3.97 to 22.70 mU/L; per-formance characteristics and comparison with other assays havebeen described previously (21).

During pregnancy, profound changes in thyroid physiologyoccur (4, 5). The maternal supply of TH to the fetoplacental unitnecessitates increased TH production, requiring an intact thy-roid gland and an adequate availability of dietary iodine. Thisprocess is in part mediated by the pregnancy hormone humanchorionic gonadotrophin (hCG), which is a weak agonist of theTSH receptor and thus stimulates the maternal thyroid to pro-duce more thyroid hormone (22, 23). As a consequence, refer-ence ranges during pregnancy are different compared with thosein a nonpregnant state (4, 5). Therefore, reference ranges forTSH, FT4, and T4 were calculated for this specific population (7).Maternal TPOAbs were measured using the Phadia 250 immu-noassay (Phadia AB) and regarded as positive when values weregreater than 60 IU/mL (24).

Iodine measurementsUrinary iodine concentrations were determined in a random

subset of 793 women during early pregnancy (mean, 12.9 weeks;SD, 1.8). Urinary iodine was measured through the ceric-arsenitereaction after destruction by means of ammonium persulfate,which has been described previously (25).

CovariatesInformation on maternal age, smoking status, and socioeco-

nomic status (SES) was obtained by questionnaires during preg-nancy. Maternal smoking status was classified as no smoking,smoking until known pregnancy, and continued smoking duringpregnancy. SES was defined by educational level, net householdincome, and employment status. Weight and length were mea-sured at intake (the same time as blood sample collection) andwere used to calculate body mass index (BMI) (26).

Statistical analysesDescriptive characteristics were compared using ANOVA,

logistic regression analyses, and the Wilcoxon-Mann-Whitney Utest. Total and subgroup reference ranges for maternal TSH,FT4, and T4 were defined as the range between the 2.5th and97.5th percentiles. In addition, TPOAb-positive women were

doi: 10.1210/jc.2013-2005 jcem.endojournals.org 3679

excluded. To achieve a normal distribution, TSH was logarith-mically transformed. Mean TSH, FT4, and T4 levels and TPOAbpositivity were compared using ANOVA and logistic regressionand additionally adjusted for maternal age, gestational age atsampling, SES, smoking, parity, and BMI. The percentages ofwomen with TSH of �2.5 mU/L in the first trimester or �3.0mU/L in the second and third trimesters per ethnic group werecompared using logistic regression analyses. The prevalences of(subclinical) hyperthyroidism, (subclinical) hypothyroidism,and hypothyroxinemia in the 4 ethnic groups were calculatedusing both total population reference ranges (TPRR) and ESRR.

Hyperthyroidism was defined as a low (�2.5th percentile)TSH level with a high (�97.5th percentile) FT4 level, subclinicalhyperthyroidism as a low TSH level with a normal (2.5th–97.5thpercentile) FT4 level, hypothyroidism as a high TSH level with alow FT4 level, subclinical hypothyroidism as a high TSH levelwith a normal FT4 level, and hypothyroxinemia as a low FT4

level with a normal TSH level.For pregnant populations, the World Health Organization

(WHO) regards median urinary iodine levels of �150 �g/L asinsufficient, 150–249 �g/L as adequate, 250–499 �g/L as aboverequirements, and �500 �g/L as excessive (27). Median urinaryiodine levels were compared among the ethnic groups using theWilcoxon-Mann-Whitney U test. The percentage of women withurinary iodine levels of �150 and �500 �g/L were comparedusing �2 tests and logistic regression analyses.

Results

The study population consisted of 3944 women of whom70.1% were Dutch, 7.8% were Moroccan, 10.7% were

Turkish, and 11.4% were Surinamese. Descriptive char-acteristics of the ethnic groups studied are shown in Table1. Dutch women were older, had a lower gestational ageat blood sampling, had fewer pregnancies, had a higherSES, and had a lower BMI. The Turkish women had thehighest smoking prevalence.

Ethnic differences in serum TSH, (F)T4, and TPOAbpositivity

With regard to unadjusted thyroid parameters, TSHvalues were significantly higher in Dutch and Turkishwomen than in Moroccan and Surinamese women (1.41–1.39 vs 1.14–1.13 mU/L; P � .01). Although unadjustedFT4 levels were significantly lower in Moroccan and Turk-ish women (14.4–14.5 vs 14.9 pmol/L; P � .02), signifi-cance was lost after correction for gestational age at sam-pling and exclusion of TPOAb-positive women (Table 2).T4 levels in the Dutch women were lower than those in allother ethnic groups (140 vs 151–157 nmol/L; P � .01).Turkish women were more frequently TPOAb-positivethan the Dutch women (9.3% vs 5.8%; P � .01). Asshown in Table 2, differences in TSH and T4 betweenethnic groups remained significant after exclusion ofTPOAb-positive women and after additional adjustmentfor maternal age, gestational age at sampling, parity,smoking, SES, and BMI. Figure 1 shows the distribution ofserum TSH levels in the total population and in the dif-ferent ethnic subgroups separately.

Table 1. Descriptive Statistics

TotalPopulation Dutch Moroccan Turkish Surinamese P Valuea

Women included, n (%) 3944 (100) 2765 (70.1) 308 (7.8) 421 (10.7) 450 (11.4)Age, y, mean (SD) 30.0 (4.9) 31.1 (4.3) 28.0 (5.4)b 26.7 (4.6)b 27.6 (5.6)b �.01Gestational age at sampling, y, mean (SD) 13.4 (2.0) 13.2 (1.9) 14.3 (2.1)b 13.8 (2.1)b 13.6 (2.1)b �.01Parity, n (%)

0 2271 (57.7) 1681 (60.9) 119 (38.8)b 208 (49.4)b 263 (58.4) �.011 1195 (30.4) 826 (29.9) 104 (33.9) 131 (31.1) 134 (29.8) .54�1 469 (11.9) 251 (9.1) 84 (27.4)b 82 (19.5)b 52 (11.6)b �.01

Smoking during pregnancy, n (%)Yes 632 (17.5) 410 (16.2) 17 (6.0)b 129 (33.9)b 76 (18.1) �.01Stopped 336 (9.3) 246 (9.7) 5 (1.8)b 32 (8.4) 53 (12.6) �.01Nonsmokers 2645 (73.2) 1875 (74.1) 259 (92.2)b 220 (57.7)b 291 (69.3)b �.01

SES, n (%)Low 345 (8.9) 100 (3.6) 84 (29.0)b 121 (30.1)b 40 (9.0)b �.01Middle 1745 (44.9) 1028 (37.4) 166 (57.2)b 216 (53.7)b 335 (75.3)b �.01High 1798 (46.2) 1623 (59.0) 40 (13.8)b 65 (16.2)b 70 (15.7)b �.01

BMI, kg/m2, mean (SD) 24.5 (4.4) 24.1 (4.0) 26.1 (4.6)b 25.7 (5.0)b 24.8 (5.0)b �.01Median TSH, mU/L 1.35 1.41 1.14b 1.39 1.13b �.01Median FT4, pmol/L 14.9 14.9 14.4b 14.5b 14.9 .02Median T4, nmol/L 144 140 151b 157b 153b �.01TPOAb positivity, n (%) 224 (6.1) 151 (5.8) 14 (5.0) 36 (9.3)b 23 (5.6) .05

a P values for maternal age, gestational age at sampling, and BMI were calculated using ANOVA. P values for parity, smoking, SES, and TPOAbpositivity were calculated using logistic regression. P values for median thyroid hormone levels were calculated using the Wilcoxon-Mann-WhitneyU test.b Significant (P � .05) compared with the Dutch group.

3680 Korevaar et al Ethnic Differences in Thyroid Parameters During PregnancyJ Clin Endocrinol Metab, September 2013, 98(9):3678–3686

For completeness, women from Morocco and Surinamwere subsequently classified according to ethnic origin,because inhabitants of these countries belong to 2 or morelarge ethnic groups. Moroccan women were classified asBerber, Arabic, or unspecified origin, whereas Surinamesewomen were classified as Creoles, Hindustani, or otherorigin. Analyses showed that, in addition to ethnicity (de-fined by country of origin), thyroid parameters may alsodiffer according to ethnic origin (defined by common an-cestry). Besides common geography, cultural habits, andrecentheritage, subdivisionaccording toethnicoriginmayreflect genetic similarities more thoroughly (see Supple-mental Table 1 published on The Endocrine Society’sJournals Online web site at http://jcem.endojournals.org).

Ethnic differences in the risk of elevated TSHlevels

Current Endocrine Society and American Thyroid As-sociation guidelines recommend an upper limit of TSH of2.5 mU/L in the first trimester and of 3.0 mU/L in thesecond and third trimesters when population-based tri-mester-specific reference ranges are not available. Table 3displays the number of women with elevated trimester-specific TSH levels according to these cutoff values. Turk-ish women had a significantly higher frequency of elevatedTSH values in the second trimester than the Dutch women(13.6% vs 9.5%; P � .02), whereas Moroccan and Suri-namese women displayed a significantly lower frequencythan the Dutch women (5.0–5.8% vs 9.5%; P � .02). Thiseffect remained significant after the exclusion of TPOAb-positive women. Moroccan women had a borderline sig-nificantly lower frequency of elevated TSH levels (P �.05).

Diagnostic consequences of the use of ESRRSubsequently, we studied whether the diagnosis of

(subclinical) thyroid disease in these different ethnicgroups was influenced by the use of reference ranges basedon the total population (TPRR) or based on each ethnicgroup separately (ESRR). In total, of all 279 women whowere found to have abnormal thyroid function test resultswhen the TPRR was used, 51 women (18%) were reclas-sified when the ESRR was used; 44 changed to normalthyroid function test results and 7 changed to a differentdisease entity. Vice versa, of all 3665 women who hadnormal thyroid function test results using the TPRR, 45(1.2%) had abnormal thyroid function test results whenthen the ESRR was used. Table 4 shows the diagnosticchanges per disease entity for the total group.

Iodine status in ethnic subgroupsTo exclude the possibility that the differences among

different ethnic groups in our study were due to iodinedeficiency in specific populations, urinary iodine levelswere measured in a random selection of the total popula-tion. As is illustrated in Table 5, all ethnic groups wereiodine sufficient according to the WHO criteria (27), withmedian urinary iodine levels between 201 and 305 �g/L.These results remained similar after adjustment for uri-nary creatinine (data not shown). Median iodine levelswere significantly higher in Moroccan, Turkish, and Su-rinamese women (201 vs 235–305 �g/L), whereas Dutchwomen more often presented with urinary iodine levels of�150 �g/L and less frequently with urinary iodine levelsof �500 �g/L.

Table 2. Ethnicity-Specific Mean TSH, FT4, and T4 Levels and Reference Ranges During Pregnancy


AdjustedP Valueb

Adjusted mean TSH, mU/L 1.40 1.50 1.29c 1.48 1.33c �.01 �.01Reference range 0.06–4.51 0.12–4.72 0.004–3.99 0.04–4.50 0.002–3.85(TPOAb-positive excluded) (0.06–4.08) (0.11–4.18) (0.004–3.56) (0.03–4.26) (0.002–3.80)

Adjusted mean FT4, pmol/L 15.1 15.1 14.9 15.1 15.2 .30 .12Reference range 10.4–21.9 10.6–21.8 9.9–21.2 9.8–22.5 10.2–23.2(TPOAb-positive excluded) (10.6–21.9) (10.8–21.8) (9.9–21.0) (9.8–22.3) (10.3–23.9)

Adjusted mean T4, nmol/L 150 142 150c 156c 152c �.01 �.01Reference range 96–219 95–204 98–233 105–242 93–238(TPOAb-positive excluded) (96–219) (96–204) (97–231) (104–238) (93–246)

Mean values were calculated as the mean for the 2.5th to 97.5th percentiles of TSH, total T4, or FT4 after exclusion of TPOAb-positive women andafter correction for gestational age. Adjusted P values were additionally corrected for maternal age, parity, smoking, and SES. Reference rangeswere defined as the 2.5th to 97.5th percentiles of respective group after exclusion of women with twin pregnancies, women with preexistingthyroid disease, women using thyroid (interfering) medication, or women undergoing fertility treatment; in addition, TPOAb-positive women wereexcluded.a Adjusted for gestational age at sampling.b Adjusted for gestational age at sampling, maternal age, SES, smoking, parity, and BMI.c Significant (P � .05) compared with the Dutch group.


Discussion

Ethnic differences are currently not taken into account forthe diagnosis of thyroid disease during pregnancy. In thecurrent study, we demonstrate that ethnic differences,even within one geographical area, may influence the di-agnosis of thyroid disease. The use of the ESRR instead ofthe TPRR changed the diagnosis for 18% of women whowere initially found to have abnormal thyroid functiontest results.

Differences in TSH levels among pregnant women fromdifferent ethnic groups have been shown in a few otherstudies (15–18). Studies in relatively small populationsfrom different parts of the United States have shown ethnicTSH differences, without corresponding effects on FT4

(15–17, 19). A study among 589 pregnant women foundthat African American women had a median TSH value of1.1 mU/L compared with 1.5 mU/L in white women (15).A subsequent study among 2568 pregnant women in thefirst trimester of pregnancy found that black women hada median TSH value of 0.82 mU/L, whereas white womenhad a median TSH level of 1.02 mU/L (16). The sameauthors showed similar differences in median TSH be-tween black and white women during the second trimester(0.97 and 1.21 mU/L, respectively). Benhadi et al (18)found that Dutch women had a higher mean TSH valuethan Moroccan, Surinamese, and Turkish women (1.19 vs0.87 and 0.91 and 0.96 mU/L, respectively). Even thoughthe study of Benhadi et al was conducted in a similar pop-ulation, TSH values in our study were slightly higher over-all, which may be explained by different assays used todetermine TSH and FT4 levels. Additional adjustment forSES in our study combined with possible differences inpopulation iodine status, which was not assessed in thestudy by Benhadi et al, may also underlie these findings.Similar study differences may also explain why TSH levelsin our study are not different between Dutch and Turkishwomen, despite the larger sample size in our study.

Although no significant differences in FT4 levels wereobserved between the different ethnic groups, T4 levelswere ethnicity dependent. Data on ethnic T4 differencesare sparse. A previous study in a relatively small first tri-mester pregnancy population (n � 668) by Pearce et al (19)showed that ethnicity was not a factor significantly con-tributing to T4 levels (19). However, Aoki et al (28)showed that T4 levels were higher in Mexican Americanscompared with non-Hispanic black and non-Hispanicwhite Americans (n � 4392), but this was studied in apredominantly nonpregnant population. In the currentstudy, we found that pregnant Dutch women had signif-icantly lower T4 levels than all other ethnic groups. Thediscrepancy between FT4 and T4 levels might reflect ethnic

Figure 1. Histograms showing the distribution of normal rangematernal TSH for the total group and the separate ethnic groups.Normal ranges for maternal TSH were defined as the 2.5th–97.5thpercentiles of respective group after exclusion of twin pregnancies,pre-existing thyroid disease, thyroid (interfering) medication usage,fertility treatment, and TPOAb-positive women.


differences in binding proteins such as thyroid hormone–binding globulin, transthyretin, and albumin.

In our study, 224 (6.1%) women were TPOAb positive,which is similar to what has been shown previously inother international studies and in different pregnant pop-ulations in The Netherlands (18, 22, 29). Ethnic variety inTPOAb positivity has been shown in large American stud-ies among men and nonpregnant women (13), as well as inpregnant women (16, 17). However, other studies onpregnant populations failed to replicate these results (15,18, 19). Turkish women in our study had the highest prev-alence of TPOAb positivity. Interestingly, Turkish womenin our cohort were also more likely to smoke. Becausesmoking has been shown to reduce the chance of TPOAbpositivity (30), it may be that the reported increased riskof TPOAb positivity in Turkish women in this populationis evenanunderestimation.TPOAbpositivityduringpreg-nancy is associated with an increased risk of postpartumthyroiditis, miscarriage, and fetal death (31, 32). WhetherTurkish women in the Netherlands are more susceptible tothese pregnancy adversities remains to be investigated infuture studies.

To investigate whether part of the ethnic differencescould be explained by differences in iodine intake, we an-alyzed urinary iodine excretion in a random sample of thispopulation. Because iodine intake is highly variable withinpopulations, even iodine-sufficient populations such as

those of the United States of America and The Netherlandscan contain subgroups with iodine deficiency or excess.Nevertheless, all 4 ethnic groups were iodine sufficientaccording to the WHO criteria (27). Compared with theother groups, the Dutch group more frequently exhibiteda low urinary iodine value (�150 �g/L) and less frequentlya high urinary iodine value (�500 �g/L). However, be-cause all populations were iodine sufficient, it is unlikelythat these differences may have caused the differences inserum thyroid function tests. Furthermore, additional ad-justment for urinary iodine excretion in the subset of 793women that had these data available did not alter ethnicgroup differences or mean thyroid hormone levels.

In the absence of trimester-specific population-basedreference ranges, TSH limits of 2.5 mU/L in the first and3.0 mU/L in the second trimester are recommended astrimester-specific upper limits (4, 5). Even in TPOAb neg-ative women, a TSH level above these cutoffs has beenrelated to increased pregnancy loss (33), but ethnic dif-ferences in high TSH levels according to these limits havenot yet been investigated. Our results demonstrate ethnicdifferences in both the first and second trimesters, withTurkish women having a higher risk of an elevated TSHthan Dutch women, regardless of TPOAb status. We showthat the Dutch and Surinamese women less frequently hadelevated TSH levels, whereas the Moroccan and Turkishwomen more frequently had high TSH levels in the second

Table 4. Number of Pregnant Women With the Diagnosis of (Subclinical) Thyroid Disease Using the TPRR or ESRRin the Total Group

Diagnosis TPRR ESRR Change—Out Change—In

Hypothyroidism 12 (0.3) 9 (0.2) 4 (36) 0 (0)Subclinical hypothyroidism 86 (2.2) 88 (2.2) 11 (13) 11 (0.4)Hypothyroxinemia 85 (2.2) 88 (2.2) 17 (22) 17 (0.6)Hyperthyroidism 36 (0.9) 35 (0.9) 5 (15) 4 (0.1)Subclinical hyperthyroidism 60 (1.5) 60 (1.5) 14 (21) 13 (0.4)Total 279 (7.1) 280 (7.1) 51 (18) 45 (1.2)

Data are n (%). “Change—Out” is the number of pregnant women who were originally found to have abnormal thyroid function test resultsusing the TPRR, but who became euthyroid or were found to have a different disease entity when the ESRR was used. “Change—In” is thenumber of pregnant women who were euthyroid when the TPRR was used but were classified within the respective disease entity when the ESRRwas used. Disease entities were diagnosed according to the reference ranges including TPOAb-positive women as displayed in Table 2.

Table 3. Percentage of Women With a TSH Level of �2.5 mU/L in the First Trimester and �3.0 mU/L in the SecondTrimester


TSH �2.5 mU/L, first trimester 122 (14.8) 96 (15.5) 1 (2.7) 8 (10.8) 17 (17.9) .17TPOAb-positive women excluded 88 (12.0) 70 (12.6) 1 (3.0) 6 (9.5) 11 (13.6) .43

TSH �3.0 mU/L, second trimester 278 (9.2) 199 (9.5) 13 (5.0)b 46 (13.6)b 20 (5.8)b �.01TPOAb-positive women excluded 192 (7.1) 138 (7.3) 9 (3.8) 32 (11.0)b 13 (4.2)b �.01

Data are n (%).a P values were calculated using logistic regression.b Significant (P � .05) compared with the Dutch group.


compared with the first trimester. Because this findingcannot be attributed to large ethnic differences in TSHdistributions as is shown in Figure 1, the current studydoes not provide an explanation for this phenomenon. Wealso demonstrate that the use of the ESRR results in achange of diagnosis for 18% of women who are found tohave abnormal thyroid function test results during preg-nancy using a local, population-based TPRR. An equalnumber of women (n � 45) classified as euthyroid on thebasis of the TPRR were found to have an abnormal thyroidfunction result based on the ESRR.

In theory, ethnic differences in TSH during pregnancymay be explained by genetic differences in thyroid hor-mone pathway genes (34–37), because �65% of the in-terindividual variation in TSH levels has been estimated tobe due to genetic factors (38). Furthermore, ethnicity is awide concept that is most often socially defined by na-tionality and culture. Alternatively, environmental factorssuch as diet or racial disparity of maternal hCG levels maybe involved as well (39–41). Subtle ethnic differences inhCG have been shown in other contexts (39, 40) but didnot explain ethnic TSH differences in a study by Walker etal (15).

Ideally, each laboratory would calculate both trimesterand ethnicity-specific reference ranges for serum TSH. Be-cause ethnic differences within one population from onegeographical area already resulted in such a significantmisclassification of thyroid disease in our hospital, it islikely that the use of fixed trimester-specific cutoffs (ie, 2.5mU/L in the first trimester and 3.0 mU/L in the second andthird trimesters) throughout the world will result in aneven larger number of misclassified patients. It is thereforeimportant to incorporate at least regional trimester-spe-cific reference ranges, if no trimester-specific referenceranges are available in the laboratory.

To date, this is the largest and most detailed study eval-uating ethnic differences in thyroid parameters duringpregnancy. Moreover, no other study has yet investigatedthe diagnostic effects of the use of ESRR. A limitation ofthis study may be the size of some subgroups; in particular,the size of the Moroccan subgroup (n � 308) was rela-tively small. In addition, we were unable to evaluate dis-

ease prevalence per trimester, because most of the sampleswere obtained in the second trimester. However, ethnicgroupcomparisonsarenot likely tobeaffectedbecause the3 largest groups were equally distributed over the first andsecond trimesters. We were unable to fully exclude theeffects of thyroglobulin antibodies; however, such anti-bodies are less common than TPOAbs and in most casescoincide with TPOAb positivity (42). Finally, even thoughour data indicate that there are no differences in iodinestatus among the four subgroups, iodine and creatininedata were only available in a random sample of pregnantwomen.

In conclusion, we have shown that TPOAb status, TSHlevels, and T4 levels differ significantly according to eth-nicity in pregnant women living in an iodine-sufficientarea. The use of ESRR instead of a TPRR changed thediagnosis for 18% of women who were initially found tohave abnormal thyroid function test results. To diagnoseand treat pregnant women with (subclinical) thyroid dis-ease correctly, the establishment of reliable referenceranges is of paramount importance. It is likely that ethnicdifferences similar to the ones shown in this study arepresent in other populations, but there is currently notenough evidence to incorporate ESRR in daily practice.Therefore, further investigations on racial differences inthyroidhormoneparameters and theirdiagnostic andclin-ical consequences in different regions of the world arewarranted.

Acknowledgments

The contributions of H. van Toor and B. C.M. Dufour-van denGoorbergh, laboratory technicians, are highly appreciated. TheGeneration R study is conducted by the Erasmus Medical Center(Rotterdam) in close collaboration with the School of Law andFaculty of Social Sciences of the Erasmus University Rotterdam;the Municipal Health Service Rotterdam area, Rotterdam; theRotterdam Homecare Foundation, Rotterdam; and the StichtingTrombosedienst and Artsenlaboratorium Rijnmond, Rotter-dam. We gratefully acknowledge the contribution of childrenand parents, general practitioners, hospitals, midwives, andpharmacies in Rotterdam.

Table 5. Urinary Iodine Levels in the Four Ethnic Subgroups

Total Dutch Moroccan Turkish Surinamese P Value

n 793 545 76 90 82Median urinary iodine, �g/L,

interquartile rangeb224 (127–358) 201 (109–329) 305 (201–506)a 269 (178–368)a 235 (148–417)a �.01

Urinary iodine �150 �g/L, n (%) 239 (30.1) 193 (35.4) 11 (14.5)a 15 (16.7)a 20 (24.4)a �.01Urinary iodine �500 �g/L, n (%) 94 (11.9) 48 (8.8) 19 (25.0)a 14 (15.6)a 13 (15.9) �.01

a Significant (P � .05) compared with the Dutch group.b Shown as median (interquartile range).


Address all correspondence and requests for reprints to:Robin P. Peeters, MD, PhD, Room D 430, Department of In-ternal Medicine, Erasmus Medical Centre, Doctor Molewater-plein 50, 3015 GE, Rotterdam, The Netherlands, E-mail:[email protected].

This work was supported by The Netherlands Organizationfor Health Research and Development (VENI Grant 91696017to R.P.P.) and an Erasmus Medical Center Fellowship (toR.P.P.). The general design of the Generation R Study is madepossible by financial support from the Erasmus Medical Center,Rotterdam; the Erasmus University Rotterdam; The Nether-lands Organization for Health Research and Development; TheNetherlands Organisation for Scientific Research; the Ministryof Health, Welfare, and Sport; and the Ministry of Youth andFamilies.

Disclosure Summary: The authors have nothing to disclose.

References

1. Gärtner R. Thyroid diseases in pregnancy. Curr Opin Obstet Gy-necol. 2009;21:501–507.

2. Lazarus JH. Thyroid function in pregnancy. Br Med Bull. 2011;97:137–148.

3. Stagnaro-Green A. Maternal thyroid disease and preterm delivery.J Clin Endocrinol Metab. 2009;94:21–25.

4. De Groot L, Abalovich M, Alexander EK, et al. Management ofthyroid dysfunction during pregnancy and postpartum: an Endo-crine Society clinical practice guideline. J Clin Endocrinol Metab.2012;97:2543–2565.

5. Stagnaro-Green A, Abalovich M, Alexander E, et al. Guidelines ofthe American Thyroid Association for the diagnosis and manage-ment of thyroid disease during pregnancy and postpartum. Thyroid.2011;21:1081–1125.

6. Stricker R, Echenard M, Eberhart R, Chevailler MC, Perez V, QuinnFA. Evaluation of maternal thyroid function during pregnancy: theimportance of using gestational age-specific reference intervals. EurJ Endocrinol. 2007;157:509–514.

7. Soldin OP, Soldin D, Sastoque M. Gestation-specific thyroxine andthyroid stimulating hormone levels in the United States and world-wide. Ther Drug Monit. 2007;29:553–559.

8. Haddow JE, Knight GJ, Palomaki GE, McClain MR, Pulkkinen AJ.The reference range and within-person variability of thyroid stim-ulating hormone during the first and second trimesters of pregnancy.J Med Screen. 2004;11:170–174.

9. Bocos-Terraz JP, Izquierdo-Alvarez S, Bancalero-Flores JL, et al.Thyroid hormones according to gestational age in pregnant Spanishwomen. BMC Res Notes. 2009;2:237.

10. Panesar NS, Li CY, Rogers MS. Reference intervals for thyroid hor-mones in pregnant Chinese women. Ann Clin Biochem. 2001;38:329–332.

11. Marwaha RK, Chopra S, Gopalakrishnan S, et al. Establishment ofreference range for thyroid hormones in normal pregnant Indianwomen. BJOG. 2008;115:602–606.

12. Boucai L, Surks MI. Reference limits of serum TSH and free T4 aresignificantly influenced by race and age in an urban outpatient med-ical practice. Clin Endocrinol (Oxf). 2009;70:788–793.

13. Hollowell JG, Staehling NW, Flanders WD, et al. Serum TSH, T4,and thyroid antibodies in the United States population (1988 to1994): National Health and Nutrition Examination Survey(NHANES III). J Clin Endocrinol Metab. 2002;87:489–499.

14. Price A, Obel O, Cresswell J, Catch I, et al. Comparison of thyroidfunction in pregnant and non-pregnant Asian and western Cauca-sian women. Clin Chim Acta. 2001;308:91–98.

15. Walker JA, Illions EH, Huddleston JF, Smallridge RC. Racial com-

parisons of thyroid function and autoimmunity during pregnancyand the postpartum period. Obstet Gynecol. 2005;106:1365–1371.

16. La’ulu SL, Roberts WL. Ethnic differences in first-trimester thyroidreference intervals. Clin Chem. 2011;57:913–915.

17. La’ulu SL, Roberts WL. Second-trimester reference intervals for thy-roid tests: the role of ethnicity. Clin Chem. 2007;53:1658–1664.

18. Benhadi N, Wiersinga WM, Reitsma JB, Vrijkotte TG, van der WalMF, Bonsel GJ. Ethnic differences in TSH but not in free T4 con-centrations or TPO antibodies during pregnancy. Clin Endocrinol(Oxf). 2007;66:765–770.

19. Pearce EN, Oken E, Gillman MW, et al. Association of first-trimes-ter thyroid function test values with thyroperoxidase antibody sta-tus, smoking, and multivitamin use. Endocr Pract. 2008;14:33–39.

20. Jaddoe VW, van Duijn CM, Franco OH, et al. The Generation RStudy: design and cohort update 2012. Eur J Epidemiol. 2012;27:739–756.

21. Rawlins ML, Roberts WL. Performance characteristics of six third-generation assays for thyroid-stimulating hormone. Clin Chem.2004;50:2338–2344.

22. Poppe K, Velkeniers B, Glinoer D. The role of thyroid autoimmunityin fertility and pregnancy. Nat Clin Pract Endocrinol Metab. 2008;4:394–405.

23. van den Boogaard E, Vissenberg R, Land JA, et al. Significance of(sub)clinical thyroid dysfunction and thyroid autoimmunity beforeconception and in early pregnancy: a systematic review. Hum Re-prod Update. 2011;17:605–619.

24. Medici M, de Rijke YB, Peeters RP, et al. Maternal early pregnancyand newborn thyroid hormone parameters: the Generation R study.J Clin Endocrinol Metab. 2012;97:646–652.

25. van Mil NH, Tiemeier H, Bongers-Schokking JJ, et al. Low urinaryiodine excretion during early pregnancy is associated with altera-tions in executive functioning in children. J Nutr. 2012;142:2167–2174.

26. Jaddoe VW, van Duijn CM, van der Heijden AJ, et al. The Gener-ation R Study: design and cohort update 2010. Eur J Epidemiol.2010;25:823–841.

27. World Health Organization United Nations Children’s Fund Inter-national Council for the Control of Iodine Deficiency Disorders.2007 Assessment of Iodine Deficiency Disorders and MonitoringTheir Elimination. 3rd ed. Geneva, Switzerland: World Health Or-ganization; 2007.

28. Aoki Y, Belin RM, Clickner R, Jeffries R, Phillips L, Mahaffey KR.Serum TSH and total T4 in the United States population and theirassociation with participant characteristics: National Health andNutrition Examination Survey (NHANES 1999–2002). Thyroid.2007;17:1211–1223.

29. Stagnaro-Green A, Glinoer D. Thyroid autoimmunity and the riskof miscarriage. Best Pract Res Clin Endocrinol Metab. 2004;18:167–181.

30. Belin RM, Astor BC, Powe NR, Ladenson PW. Smoke exposure isassociated with a lower prevalence of serum thyroid autoantibodiesand thyrotropin concentration elevation and a higher prevalence ofmild thyrotropin concentration suppression in the third NationalHealth and Nutrition Examination Survey (NHANES III). J ClinEndocrinol Metab. 2004;89:6077–6086.

31. Thangaratinam S, Tan A, Knox E, Kilby MD, Franklyn J, Cooma-rasamy A. Association between thyroid autoantibodies and miscar-riage and preterm birth: meta-analysis of evidence. BMJ. 2011;342:d2616.

32. Muller AF, Drexhage HA, Berghout A. Postpartum thyroiditis andautoimmune thyroiditis in women of childbearing age: recent in-sights and consequences for antenatal and postnatal care. EndocrRev. 2001;22:605–630.

33. Negro R, Schwartz A, Gismondi R, Tinelli A, Mangieri T, Stagnaro-Green A. Increased pregnancy loss rate in thyroid antibody negativewomen with TSH levels between 2.5 and 5.0 in the first trimester ofpregnancy. J Clin Endocrinol Metab. 2010;95:E44–E48.

34. Medici M, van der Deure WM, Verbiest M, et al. A large-scale


mailto:[email protected]

association analysis of 68 thyroid hormone pathway genes withserum TSH and FT4 levels. Eur J Endocrinol. 2011;164:781–788.

35. Peeters RP, van der Deure WM, Visser TJ. Genetic variation inthyroid hormone pathway genes; polymorphisms in the TSH recep-tor and the iodothyronine deiodinases. Eur J Endocrinol. 2006;155:655–662.

36. Rawal R, Teumer A, Völzke H, et al. Meta-analysis of two genome-wide association studies identifies four genetic loci associated withthyroid function. Hum Mol Genet. 2012;21:3275–3282.

37. Porcu E, Medici M, Pistis G, et al. A meta-analysis of thyroid-relatedtraits reveals novel Loci and gender-specific differences in the reg-ulation of thyroid function. PLoS Genet. 2013;9:e1003266.

38. Hansen PS, Brix TH, Sørensen TI, Kyvik KO, Hegedüs L. Majorgenetic influence on the regulation of the pituitary-thyroid axis: a

study of healthy Danish twins. J Clin Endocrinol Metab. 2004;89:1181–1187.

39. O’Brien JE, Dvorin E, Drugan A, Johnson MP, Yaron Y, Evans MI.Race-ethnicity-specific variation in multiple-marker biochemicalscreening: alpha-fetoprotein, hCG, and estriol. Obstet Gynecol.1997;89:355–358.

40. Muller F, Bussières L, Pèlissier MC, et al. Do racial differences existin second-trimester maternal hCG levels? A study of 23,369 pa-tients. Prenat Diagn. 1994;14:633–636.

41. Brent GA, Braverman LE, Zoeller RT. Thyroid health and the en-vironment. Thyroid. 2007;17:807–809.

42. Unuane D, Velkeniers B, Anckaert E, et al. Thyroglobulin antibod-ies: any added value in the detection of thyroid autoimmunity inwomen consulting for fertility treatment [published online ahead ofprint February 13, 2013]. Thyroid. doi:10.1089/thy.2012.0562.

Members can search for endocrinology conferences, meetings and webinars on the Worldwide Events Calendar.

www.endo-society.org/calendar


ethnic differences in maternal thyroid parameters during pregnancy: the generation r study

Documents