Subclinical hypothyroidism and isolated hypothyroxinaemia in pregnancy have been associated with an increased risk of gestational diabetes. We aimed to ascertain if these women have a worse metabolic phenotype than euthyroid pregnant women.
Design, subjects and methods
We recruited 956 healthy Caucasian women with singleton, non-diabetic pregnancies from routine antenatal clinics. Detailed anthropometric measurements (including BMI and skinfold thickness) and fasting blood samples (for TSH, free thyroxine (FT4), free triiodothyronine (FT3), HbA1c, lipid profile, plasma glucose and insulin resistance (HOMA-IR) analysis) were obtained at 28 weeks gestation.
In comparison to euthyroid women (n=741), women with isolated hypothyroxinaemia (n=82) had significantly increased BMI (29.5 vs 27.5 kg/m2, P<0.001), sum of skinfolds (57.5 vs 51.3 mm, P=0.002), fasting plasma glucose (4.5 vs 4.3 mmol/l, P=0.01), triglycerides (2.3 vs 2.0 mmol/l, P<0.001) and HOMA-IR (2.0 vs 1.3, P=0.001). Metabolic parameters in women with subclinical hypothyroidism (n=133) were similar to those in euthyroid women. Maternal FT4 was negatively associated with BMI (r=−0.22), HbA1c (r=−0.14), triglycerides (r=−0.17), HOMA-IR (r=−0.15) but not total/HDL cholesterol ratio (r=−0.03). Maternal FT3:FT4 ratio was positively associated with BMI (r=0.4), HbA1c (r=0.21), triglycerides (r=0.2), HOMA–IR (r=0.33) and total/HDL cholesterol ratio (r=0.07). TSH was not associated with the metabolic parameters assessed.
Isolated hypothyroxinaemia, but not subclinical hypothyroidism, is associated with adverse metabolic phenotype in pregnancy, as is decreasing maternal FT4 and increasing FT3:FT4 ratio. These associations may be a reflection of changes in the thyroid hormone levels secondary to increase in BMI rather than changes in thyroid hormone levels affecting body weight and related metabolic parameters.
Recently, increasing numbers of studies have shown associations between mild maternal thyroid hormone insufficiency (including subclinical hypothyroidism and isolated hypothyroxinaemia) in pregnancy and impaired neuropsychological development of the offspring, as well as several obstetric complications, such as miscarriage, preterm delivery, gestational hypertension and pre-eclampsia (1, 2, 3). Several of these studies (4, 5, 6), although not all (7, 8) have also shown that subclinical hypothyroidism in pregnancy is associated with an increased risk of gestational diabetes. Isolated hypothyroxinaemia has also been shown to be associated with gestational diabetes (7), and a recent study found that lower free thyroxine (FT4) levels and higher free triiodothyronine (FT3) to FT4 ratios (suggesting higher peripheral deiodinase activity leading to conversion of FT4 to FT3) in pregnant women are associated with several adverse metabolic parameters relating to obesity, glycaemia, insulin resistance and lipid profile (9). It is possible that these adverse metabolic parameters partly explain the poor obstetric outcomes observed in pregnant women with mild maternal thyroid insufficiency (10, 11, 12).
As little is known on whether mild maternal thyroid hormone insufficiency in pregnancy is associated with adverse metabolic parameters, our study aimed to explore the relationship between maternal thyroid hormone levels and metabolic parameters in pregnancy and ascertain if pregnant women with subclinical hypothyroidism and isolated hypothyroxinaemia have worse metabolic phenotype than euthyroid pregnant women.
Subjects and methods
A total of 988healthy non-diabetic Caucasian women with singleton pregnancies were recruited into the Exeter Family Study of Childhood Health between 1999 and 2004 (13). These women had detailed anthropometric measurements, including height, weight and skinfold measures (triceps, biceps and subscapular) and fasting blood samples taken at 28 weeks gestation. After exclusion of women on thyroid-related medications (levothyroxine n=18, propylthiouracil n=3) and women with overt biochemical hypothyroidism (n=10) or hyperthyroidism (n=1) on the blood samples, data from 956 women were included in this study.
The study was approved by the North and East Devon Local Research Ethics Committee, and all participants gave informed written consent.
Analysis of thyroid function and biochemical metabolic parameters
Thyroid function tests (thyroid-stimulating hormone (TSH), FT4 and FT3) and thyroid peroxidase antibodies (TPO-Ab) were measured on the stored serum samples. Serum TSH, FT4 and FT3 were analysed using the electrochemiluminescent immunoassay, run on the Modular E170 Analyzer (Roche). Intra-assay coefficients of variations were as follows: TSH <5.3%, FT4 <5.3% and FT3 <5.1%. The manufacturer's population reference ranges were as follows: TSH 0.35–4.5 mIU/l, FT4 11–24 pmol/l, and FT3 3.9–6.8 pmol/l. High TSH in this study of women in the third trimester was defined as >3 mIU/l, the generally accepted upper limit of the reference range in the second and third trimesters in Caucasian populations (1, 2). Subclinical hypothyroidism was defined as high TSH (>3 mIU/l) with FT4 within the reference range (11–24 pmol/l). Isolated maternal hypothyroxinaemia was defined as FT4 below the 10th centile (<10.4 pmol/l) with TSH within the reference range. TPO-Abs were analysed using the competitive immunoassay (Roche), and a titre above 34 IU/ml was considered positive.
Fasting plasma glucose, serum insulin and lipid profile and glycosylated haemoglobin (HbA1c) were analysed in the fresh blood samples, as previously described (13). We analysed homeostasis model of assessment for insulin resistance (HOMA-IR) to estimate fasting insulin sensitivity using the following formula: fasting serum insulin (mIU/l)×fasting plasma glucose (mmol/l)/22.5.
We assessed variables for distribution and log-transformed where they were not normally distributed. We used the independent sample t-test to compare metabolic parameters in women with isolated hypothyroxinaemia or subclinical hypothyroidism and women with normal thyroid function, and in euthyroid women with and without TPO-Ab. We used the χ2 test to assess TPO-Ab-positive proportions in each group. We used Pearson's correlation to assess associations between variables, and multiple linear regression analysis to assess independent associations between variables. As smoking and TPO-Ab are potential confounders affecting thyroid function, we have also included these factors in the regression analysis (14, 15). We used independent samples t-test to compare metabolic parameters between women with lower and upper quartiles of FT4 and between women with lower and upper quartiles of FT3:FT4 ratio. We carried out statistical analyses using SPSS (version 22).
Participants had a mean (s.d.) age of 30.1 (5.1) years and BMI of 27.9 (4.6) kg/m2; 46% were primiparous and 14% smoked during pregnancy.
Metabolic parameters in pregnant women with subclinical hypothyroidism and isolated hypothyroxinaemia
A total of 133 (13.9%) women in our cohort had subclinical hypothyroidism and 82 (8.6%) had isolated hypothyroxinaemia.
There was no difference in metabolic parameters between women with subclinical hypothyroidism and euthyroid women (Table 1). Metabolic parameters in women with subclinical hypothyroidism who had TSH greater than 5 mIU/l (n=12) were also similar to those in euthyroid women (data not shown).
Comparison of thyroid function and metabolic parameters between euthyroid mothers and those with subclinical hypothyroidism and euthyroid mothers and those with isolated hypothyroxinaemia in pregnancy (data presented as mean±s.d. or ageometric mean (s.d. range) except for TPO-Ab positivity which is presented as percentage). The bold values are statistically significant results.
|Metabolic parameters||Euthyroid controls (n=741)||Subclinical hypothyroidism (n=133)||P valueb||Isolated hypothyroxinaemia (n=82)||P valueb|
|TSHa (mIU/l)||1.6 (0.9, 2.7)||3.7 (3.1, 4.5)||<0.001||1.8 (1.3, 2.5)||0.001|
|FT4 (pmol/l)||12.5 (1.4)||11.9 (1.5)||<0.001||9.7 (0.5)||<0.001|
|FT3 (pmol/l)||4.2 (0.5)||4.2 (0.5)||0.2||4.1 (0.4)||0.04|
|TPO-Ab positive (%)||5.0||13.7||0.001||6.0||0.6|
|BMI (kg/m2)a||27.5 (23.4–32.4)||27.5 (23.7–32.1)||0.7||29.5 (25.1–34.7)||<0.001|
|Sum of skinfolds (mm)a||51.3 (37.2–70.8)||50.1 (33.3–69.2)||0.9||57.5 (40.7–81.3)||0.002|
|HbA1c (mmol/mol)||28.7 (3.7)||28.5 (3.9)||0.7||29.9 (4.6)||0.05|
|Fasting plasma glucose (mmol/l)||4.3 (0.4)||4.3 (3.8)||0.8||4.5(0.4)||0.01|
|Triglycerides (mmol/l)a||2.0 (1.5–2.8)||2.0 (1.4–2.8)||0.7||2.3 (1.7–3.3)||<0.001|
|HOMA-IRa||1.3 (0.7–2.0)||1.0 (0.8–2.0)||0.4||2.0 (0.9–2.8)||0.001|
|Total/HDL cholesterol ratio||3.4 (0.9)||3.3 (0.9)||0.4||3.5 (1.0)||0.1|
Denotes log-transformed data.
P values refer to euthyroid controls vs subclinical hypothyroidism and euthyroid controls vs isolated hypothyroxinaemia.
When compared to euthyroid women, women with maternal isolated hypothyroxinaemia had significantly increased BMI (29.5 vs 27.5 kg/m2, P<0.001), sum of skinfolds (57.5 vs 51.3 mm, P=0.002), fasting plasma glucose (4.5 vs 4.3 mmol/l, P=0.01), triglycerides (2.3 vs 2.0 mmol/l, P<0.001) and HOMA-IR (2.0 vs 1.3, P=0.001), and borderline increased HbA1c (29.9 vs 28.7 mmol/mol, P=0.05) (Table 1).
Association between maternal thyroid hormone levels and metabolic parameters
In bivariate analyses, maternal FT4 was negatively associated with BMI (r=−0.22, P<0.001), sum of skinfolds (r=−0.22, P<0.001), HbA1c (r=−0.14, P<0.001), fasting plasma glucose (r=−0.13, P<0.001), triglycerides (r=−0.17, P<0.001), HOMA-IR (r=−0.15, P<0.001) but not Total/HDL cholesterol ratio (r=−0.03, P=0.3) (Table 2). Maternal FT3:FT4 ratio was positively associated with BMI (r=0.40), sum of skinfolds (r=0.37), HbA1c (r=0.21), fasting plasma glucose (r=0.24), triglycerides (r=0.2) and HOMA–IR (r=0.33) (P<0.001 for all) and Total/HDL cholesterol (r=0.07, P=0.04). Maternal TSH in pregnancy was not correlated with any of the maternal metabolic parameters assessed.
Correlations between maternal TSH, FT4, FT3:FT4 ratio and metabolic parameters at 28 weeks gestation. The bold values are statistically significant results.
r (P value)
r (P value)
r (P value)
|BMI (kg/m2)a||0.007 (0.8)||−0.22 (<0.001)||0.40(<0.001)|
|Sum of skinfolds (mm)a||0.013 (0.7)||−0.22 (<0.001)||0.37 (<0.001)|
|HbA1c (mmol/mol)||0.007 (0.8)||−0.14 (<0.001)||0.21 (<0.001)|
|Fasting plasma glucose (mmol/l)||−0.01 (0.8)||−0.13 (<0.001)||0.24 (<0.001)|
|Triglycerides (mmol/l)a||0.006 (0.9)||−0.17 (<0.001)||0.20 (<0.001)|
|HOMA–IRa||0.02 (0.5)||−0.15 (<0.001)||0.33 (<0.001)|
|Total/HDL cholesterol ratio||−0.04 (0.2)||−0.03 (0.3)||0.07 (0.04)|
Denotes log-transformed data.
Women with the lowest FT4 quartile and the highest FT3:FT4 ratio quartile had significantly higher BMI, sum of skinfolds, HbA1c, fasting plasma glucose, triglycerides and HOMA-IR as compared to women with the highest FT4 quartile and the lowest FT3:FT4 ratio quartile respectively (P≤0.001 for all) (Table 3). There was no difference in total/HDL cholesterol ratio between groups.
Comparisons between upper and lower FT4 and FT3:FT4 ratio quartiles, and maternal metabolic parameters (data presented as mean±s.d. or ageometric mean (s.d. range)). Bold values are statistically significant results.
|Metabolic parameters||FT4||P value||FT3:FT4 ratio||P value|
|Lower quartile||Upper quartile||Lower quartile||Upper quartile|
|BMI (kg/m2)a||28.8 (24.6–33.9)||26.9 (22.9–31.6)||<0.001||25.1 (21.9–28.8)||30.2 (26.7–35.5)||<0.001|
|Sum of skinfolds (mm)a||57.4 (41.7–79.4)||47.9 (34.7–66.1)||<0.001||44.7 (32.4–61.7)||63.1 (45.7–87.1)||<0.001|
|HbA1c (mmol/mol)||29.5 (4.1)||28.1 (3.8)||0.001||27.9 (3.6)||29.9 (4.4)||<0.001|
|Fasting plasma glucose (mmol/l)||4.4 (0.4)||4.3 (0.4)||<0.001||4.2 (0.3)||4.5 (0.4)||<0.001|
|Triglyceride (mmol/l)a||2.2 (1.6–3.1)||1.9 (1.4–2.5)||<0.001||1.9 (1.4–2.6)||2.2 (1.6–3.1)||<0.001|
|HOMA-IRa||2.0 (1.0–2.6)||1.0 (0.8–2.0)||<0.001||1.0 (0.6–1.6)||2.0 (1.0–2.5)||<0.001|
|Total/HDL cholesterol ratio||3.0 (1.8)||3.3 (0.7)||0.17||3.3 (0.9)||3.4 (1.0)||0.24|
In multivariable regression analyses (Table 4), FT4 was independently associated with BMI, HbA1c, triglycerides and smoking during pregnancy, but not HOMA-IR. FT3:FT4 ratio was independently associated with BMI, HbA1c, HOMA-IR, and borderline associated with smoking during pregnancy, but not triglycerides or total/HDL cholesterol.
Multivariable regression analysis to demonstrate the relationships between maternal FT4 and FT3:FT4 ratio with metabolic parameters. Partial R is the adjusted correlation coefficient. Bold values are statistically significant results.
|Associated metabolic parameters||Partial R||β||95% CI for β||t||P|
|BMI (kg/m2)a||−0.13||−1.55||−2.43, −0.68||−3.47||0.001|
|HbA1c (mmol/mol)||−0.09||−0.04||−0.07, −0.01||−2.54||0.01|
|Triglycerides (mmol/l)a||−0.10||−0.49||−0.84, −0.13||−2.71||0.007|
|Smoking during pregnancy||0.09||0.43||0.10, 0.77||2.52||0.01|
|TPO-Ab positive||−0.004||−0.02||−0.47, 0.41||−0.10||0.9|
|Maternal FT3:FT4 ratio|
|BMI (kg/m2)a||0.26||0.10||0.07, 0.12||7.18||<0.001|
|HbA1c (mmol/mol)||0.11||0.001||0.00, 0.002||2.92||0.004|
|Triglycerides (mmol/l)a||0.07||0.01||−0.001, 0.02||1.86||0.06|
|Total/HDL cholesterol||−0.07||−0.004||−0.009, 0.000||−3.52||0.06|
|Smoking during pregnancy||0.07||0. 01||0.00, 0.02||1.98||0.05|
|TPO-Ab positive||0.02||0.003||−0.01, 0.02||0.80||0.9|
Variable transformed using natural logs (allowing β coefficients to be interpreted in terms of percentage change).
Association between maternal TPO-Ab status and metabolic parameters
Amongst the euthyroid women, there was no difference in metabolic parameters between TPO-Ab-positive (n=37) and TPO-Ab (n=696)-negative women (data not shown).
In this study of healthy non-diabetic pregnant women, we found that women with isolated hypothyroxinaemia have worse metabolic parameters with increased obesity, glycaemia, triglycerides and insulin resistance compared to euthyroid women. In contrast, subclinical hypothyroidism in pregnancy was not associated with these adverse metabolic parameters. We also identified lower maternal serum FT4 and higher FT3:FT4 ratio (indicating an increased peripheral deiodinase activity leading to conversion of FT4 to FT3) associated with adverse metabolic parameters.
Thyroid hormones play key roles in regulating metabolic processes and energy homeostasis in the body (16). It is well known that overt thyroid dysfunction (hypothyroidism or hyperthyroidism) affects body weight, and recent studies have shown that even small variations in thyroid hormone levels within the reference range are associated with significant metabolic consequences and changes in body weight (16, 17). In the mouse model of type 2 diabetes, thyroid hormone has been shown to improve glycaemia and insulin sensitivity (18). Our finding of the inverse correlation between maternal serum FT4 and maternal BMI is consistent with previous studies in the pregnant and non-pregnant populations (19, 20, 21, 22, 23). In the general population, low FT4 as well as high FT3:FT4 ratio have also been shown to be associated with adverse metabolic parameters, including less favourable lipid profile, blood pressure and insulin resistance (24, 25, 26, 27, 28). It is thought that lower FT4 levels are compensated by a higher peripheral deiodinase activity resulting in higher conversion of FT4 to active thyroid hormone FT3 and higher FT3:FT4 ratio (20). In a recent study of 321 healthy pregnant women without a history of thyroid dysfunction, Bassols et al. (9) found that decreasing FT4 and increasing FT3:FT4 ratios are similarly associated with less favourable metabolic profile. In addition to confirming the associations in a larger cohort of pregnant women, our study also demonstrates the presence of worse metabolic parameters in women with maternal hypothyroxinaemia as compared to euthyroid women. In contrast, despite several studies (4, 5, 6), but not all (7, 8), showing association between subclinical hypothyroidism in pregnancy and gestational diabetes, we were unable to demonstrate that pregnant women with subclinical hypothyroidism have less favourable metabolic parameters than euthyroid pregnant women. The possible explanations for this discrepant observations include differences in study populations (for example, women with gestational diabetes were excluded in our study) and different TSH cut-off levels used to define subclinical hypothyroidism.
As it is a cross-sectional observational study, our study cannot ascertain that the observed less favourable metabolic parameters are caused by changes in the thyroid hormone levels. Indeed, it is possible that the associations between maternal thyroid function and metabolic parameters are mediated by obesity, as glycaemia, insulin resistance and dyslipidaemia are closely associated with obesity. Several lines of evidence support the hypothesis that changes in thyroid hormone levels are the consequence rather than the cause of changes in body weight. Increased subcutaneous fat has been shown to be associated with lower FT4 and higher TSH levels in euthyroid adults (29). A study in iodine-deficient pregnant women showed that obesity is associated with an increased risk of maternal hypothyroxinaemia (30). It is thought that obesity stimulates peripheral deiodinase activity as an adaptation process to increase energy expenditure resulting in an increased conversion of FT4 to FT3 and high FT3:FT4 ratio (22). More importantly, weight loss following dietary and life-style interventions have been shown to be associated with changes in thyroid hormone levels, including decreases in FT3 level and FT3:FT4 ratio (31, 32). Together, these observations suggest that pregnant women should be encouraged to avoid excess weight gain to prevent maternal hypothyroxinaemia.
We acknowledge a number of other limitations in this study. First, we carried out thyroid function tests on stored serum samples; however, it has been shown that serum TSH, FT4, FT3 and TPO-Ab concentrations remain stable after storage for several years (33). Second, women with gestational diabetes were excluded from the study during recruitment, therefore we were unable to analyse whether the observed association between isolated hypothyroxinaemia and adverse metabolic parameters extended to an increased risk of gestational diabetes. Finally, we were also unable to assess the potential impact of iodine deficiency on our results due to a lack of urinary iodine concentration data.
In conclusion, this study demonstrates that maternal isolated hypothyroxinaemia, but not subclinical hypothyroidism, is associated with adverse metabolic parameters in pregnancy. The association between low serum thyroxine level and adverse metabolic parameters may be a reflection of changes in the thyroid hormone levels secondary to increases in BMI rather than changes in thyroid hormone levels affecting body weight and related metabolic parameters.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported.
The Endocrine Research Fund funded the thyroid function tests. B A Knight, B M Shields and A T Hattersley are funded by the National Institute for Health Research (NIHR) as core members of the NIHR Exeter Clinical Research Facility. Work of B Vaidya is supported by the Research and Development Department of the Royal Devon and Exeter NHS Foundation Trust.
Stagnaro-GreenAAbalovichMAlexanderEAziziFMestmanJNegroRNixonAPearceENSoldinOPSullivanS. American Thyroid Association Taskforce on Thyroid Disease During Pregnancy & Postpartum. Guidelines of the American Thyroid Association for the diagnosis and management of thyroid disease during pregnancy and postpartum. Thyroid2011211081–1125. (doi:10.1089/thy.2011.0087).
De GrootLAbalovichMAlexanderEKAminoNBarbourLCobinRHEastmanCJLazarusJHLutonDMandelSJ. Management of thyroid dysfunction during pregnancy and postpartum: an Endocrine Society clinical practice guideline. Journal of Endocrinology and Metabolism2012972543–2565. (doi:10.1210/jc.2011-2803).
KarakostaPAlegakisDGeorgiouVRoumeliotakiTFthenouEVassilakiMBoumpasDCastanasEKogevinasMChatziL. Thyroid dysfunction and autoantibodies in early pregnancy are associated with increased risk of gestational diabetes and adverse birth outcomes. Journal of Endocrinology and Metabolism2012974464–4472. (doi:10.1210/jc.2012-2540).
BassolsJPrats-PuigASoriano-RodriguezPGarcia-GonzalezMMReidJMartinez-PascualMMateos-ComeronFde ZegherFIbanezLLopez-BermejoA. Lower free thyroxin associates with a less favorable metabolic phenotype in healthy pregnant women. Journal of Endocrinology and Metabolism2011963717–3723. (doi:10.1210/jc.2011-1784).
DennedyMCAvalosGO'ReillyMWO'SullivanEPGaffneyGDunneF. ATLANTIC-DIP: raised maternal body mass index (BMI) adversely affects maternal and fetal outcomes in glucose-tolerant women according to International Association of Diabetes and Pregnancy Study Groups (IADPSG) criteria. Journal of Endocrinology and Metabolism201297E608–E612. (doi:10.1210/jc.2011-2674).
MannistoTHartikainenALVaarasmakiMBloiguASurcelHMPoutaAJarvelinMRRuokonenASuvantoE. Smoking and early pregnancy thyroid hormone and anti-thyroid antibody levels in euthyroid mothers of the Northern Finland Birth Cohort 1986. Thyroid201222944–950. (doi:10.1089/thy.2011.0377).
HaddowJECraigWYPalomakiGENeveuxLMLambert-MesserlianGCanickJAMaloneFDD'AltonME. First & Second Trimester Risk Of Aneuploidy Faster Research C. Impact of adjusting for the reciprocal relationship between maternal weight and free thyroxine during early pregnancy. Thyroid201323225–230. (doi:10.1089/thy.2012.0440).
HaddowJECraigWYNeveuxLMHaddowHRPalomakiGELambert-MesserlianGMaloneFDD'AltonME. First & Second Trimester Risk of Aneuploidy Research C. Implications of High Free Thyroxine (FT4) concentrations in euthyroid pregnancies: the FaSTER trial. Journal of Clinical Endocrinology and Metabolism2014992038–2044. (doi:10.1210/jc.2014-1053).
KnudsenNLaurbergPRasmussenLBBulowIPerrildHOvesenLJorgensenT. Small differences in thyroid function may be important for body mass index and the occurrence of obesity in the population. Journal of Clinical Endocrinology and Metabolism2005904019–4024. (doi:10.1210/jc.2004-2225).
De PergolaGCiampolilloAPaolottiSTrerotoliPGiorginoR. Free triiodothyronine and thyroid stimulating hormone are directly associated with waist circumference, independently of insulin resistance, metabolic parameters and blood pressure in overweight and obese women. Clinical Endocrinology200767265–269. (doi:10.1111/j.1365-2265.2007.02874.x).
Garduno-Garcia JdeJAlvirde-GarciaULopez-CarrascoGPadilla MendozaMEMehtaRArellano-CamposOChozaRSauqueLGaray-SevillaMEMalacaraJMGomez-PerezFJAguilar-SalinasCA. TSH and free thyroxine concentrations are associated with differing metabolic markers in euthyroid subjects. European Journal of Endocrinology2010163273–278. (doi:10.1530/EJE-10-0312).
RoefGLRietzschelERVan DaeleCMTaesYEDe BuyzereMLGillebertTCKaufmanJM. Triiodothyronine and free thyroxine levels are differentially associated with metabolic profile and adiposity-related cardiovascular risk markers in euthyroid middle-aged subjects. Thyroid201424223–231. (doi:10.1089/thy.2013.0314).
MannistoTSurcelHMBloiguARuokonenAHartikainenALJarvelinMRPoutaAVaarasmakiMSuvanto-LuukkonenE. The effect of freezing, thawing, and short- and long-term storage on serum thyrotropin, thyroid hormones, and thyroid autoantibodies: implications for analyzing samples stored in serum banks. Clinica Chimica Acta2007531986–1987.