2017 Guidelines of the American Thyroid Association for the Diagnosis and Management of Thyroid Disease During Pregnancy and the Postpartum

Question 1: How do thyroid function tests change during pregnancy?

Normal pregnancy is associated with an increase in renal iodine excretion, an increase in thyroxine binding proteins, an increase in thyroid hormone production, and thyroid stimulatory effects of hCG. All of these factors influence thyroid function tests in the pregnant patient. The healthy thyroid adapts to these alterations through changes in thyroid hormone metabolism, iodine uptake, and the regulation of the hypothalamic-pituitary-thyroid axis (3,4). The thyroid function tests of healthy pregnant women, therefore, differ from those of healthy nonpregnant women. Furthermore, the reference ranges for the most widely applied tests, TSH and free thyroxine (FT4), may vary significantly in different populations.

Following conception, circulating thyroxine binding globulin (TBG) and total T4 (TT4) concentrations increase by week 7 of gestation and reach a peak by approximately week 16 of gestation (5). These concentrations then remain high until delivery. In the first trimester, maternal hCG directly stimulates the TSH receptor, increasing thyroid hormone production and resulting in a subsequent reduction in serum TSH concentration (4,6). Therefore, during pregnancy, women have lower serum TSH concentrations than before pregnancy, and a TSH below the nonpregnant lower limit of 0.4 mU/L is observed in as many as 15% of healthy women during the first trimester of pregnancy (7,8). In Japan, a suppressed TSH less than 0.6 mU/L is similarly frequently observed during the first trimester of pregnancy (9). The fraction of women with a suppressed TSH falls to about 10% in the second trimester, and 5% in the third trimester (4).

Measurement of FT4 concentration by automated immunoassays results in a significant and assay dependent reduction in the measured serum FT4 concentrations in the third trimester, even though direct measurement of free T4 by more precise methods does not show a similar degree of reduction (4,6,7). Automated immunoassays for FT4, which are employed in most clinical laboratories, are complicated in pregnant women by the increase in TBG and decrease in albumin concentrations (10,11). Other methods of direct measurement, such as measurement by equilibrium dialysis, ultrafiltration, or liquid chromatography/tandem mass spectrometry (LC/MS/MS) are less influenced by the pregnancy-associated changes in serum proteins but are significantly more expensive and less widely available. Thus, the automated immunoassays used for serum FT4 analysis are still widely used, but the important considerations discussed above must be noted. The use of population-based, trimester-specific reference ranges remains the best way to handle this issue.

Question 2: What is the normal reference range for serum TSH concentrations in each trimester of pregnancy?

A downward shift of the TSH reference range occurs during pregnancy, with a reduction in both the lower (decreased by about 0.1–0.2 mU/L) and the upper limit of maternal TSH (decreased by about 0.5–1.0 mU/L), relative to the typical nonpregnant TSH reference range. The largest decrease in serum TSH is observed during the first trimester because of elevated levels of serum hCG directly stimulating the TSH receptor and thereby increasing thyroid hormone production (Table 4). Thereafter, serum TSH and its reference range gradually rise in the second and third trimesters, but nonetheless remain lower than in nonpregnant women (12,13). Since hCG concentrations are higher in multiple pregnancies than in singleton pregnancies, the downward shift in the TSH reference interval is greater in twin pregnancies (11). In a study of 63 women with hCG concentrations >200,000 IU/L, TSH was suppressed (≤0.2 mU/L) in 67% of women and in 100% of women if hCG concentrations were >400,000 IU/L (14). Serum TSH reference range determinations should take into account iodine intake, TPO positivity, and according to some studies, body mass index (BMI).

Although the downward shift in TSH reference ranges is seen in essentially all populations, the extent of this reduction varies significantly between different racial and ethnic groups. Initial studies of pregnant women in the United States and Europe first led to recommendations for a TSH upper reference limit of 2.5 mU/L in the first trimester and 3.0 mU/L in the second and third trimesters (1,15). However, more recent studies in pregnant women in Asia, India, and the Netherlands, have demonstrated only a modest reduction in the upper reference limit (16 –20). A study of 4800 pregnant women in China recently showed that the downward shift in the TSH reference range occurred at weeks 7–12, but the upper reference limit was only reduced from 5.31 to 4.34 mU/L (17). Separate data from a recent prospective intervention trial in the United States support this finding (21). Analysis of the TSH and FT4 “set-point” in pregnant women showed that reductions in FT4 were observed only when the serum TSH was greater than 4.8 mU/L. Similar studies of pregnant women in India and Korea show a modest reduction in the first-trimester upper TSH limit of 0.5 1.0 mU/L. In some cases, this was not statistically different from the nonpregnant state (18,20). Thus, the current evidence supports only a slight downward shift in the upper reference range of TSH occurring in the latter first trimester of pregnancy, typically not seen prior to week 7 (17).

A reduction in the lower TSH reference range is observed during pregnancy in almost all studies. In a small percentage of women, TSH can be undetectable (<0.01 mU/L), and yet still represent a normal pregnancy. In addressing the clinical importance of a reduced serum TSH during pregnancy, it is important to note that subclinical hyperthyroidism has not been associated with adverse pregnancy outcomes. Therefore, a maternal TSH concentration that is low but detectable is likely not clinically significant (22). TSH ranges have been shown to vary slightly depending on different methods of analysis, although this variation is not clinically significant (23). One approach to reducing this variability is to use the Multiple of Medians calculation to compare values between assays. This calculation divides an individual value by the population median (24). The resulting value is not influenced by the differences between assays (24).

Significant geographic and ethnic diversity exist in TSH concentrations during pregnancy, as shown in Table 4. The task force recognizes the limited availability of trimester-specific reference ranges calculated for most ethnic and racial populations with adequate iodine intake who are free of thyroid autoantibodies. Nonetheless, to provide guidance to all patients and clinicians, the panel recommends use of the following trimester-specific ranges and cutoffs when local assessments are not available. In the first trimester, the lower reference range of TSH can be reduced by approximately 0.4 mU/L, while the upper reference range is reduced by approximately 0.5 mU/L. For the typical patient in early pregnancy, this corresponds to a TSH upper reference limit of 4.0 mU/L. This reference limit should generally be applied beginning with the late first trimester, weeks 7–12, with a gradual return towards the nonpregnant range in the second and third trimesters. For specific recommendations regarding the diagnosis and treatment of maternal hypothyroidism, see Section VII.

■ RECOMMENDATION 1

 When possible, population-based trimester-specific reference ranges for serum TSH should be defined through assessment of local population data representative of a health care provider's practice. Reference range determinations should only include pregnant women with no known thyroid disease, optimal iodine intake, and negative TPOAb status.

 Strong recommendation, moderate-quality evidence.

Question 3: What is the optimal method to assess serum T4 concentration during pregnancy?

Unbound T4 represents only about 0.03% of serum TT4 content. Importantly, only FT4 is available for tissue uptake, with the remainder of T4 bound to serum proteins, primarily TBG. Serum TT4 concentrations are measured in the nanomolar range, while FT4 concentrations are measured in the picomolar range. In part because of this, measuring FT4 in the presence of high concentrations of bound T4 has proved to be challenging. This challenge is especially true in conditions where binding-protein levels are altered, such as pregnancy.

Equilibrium dialysis and ultrafiltration are used for physical separation of serum FT4 from bound T4 prior to analysis of the dialysate or ultrafiltrate. While they are theoretically not influenced by changes in binding proteins and heterophilic antibodies, assays based on classical equilibrium dialysis or ultrafiltration are laborious, time-consuming, expensive, and not widely available.

As noted above, FT4 measurement performed by indirect analog immunoassays is used by the majority of clinical laboratories, largely because of its ability to be quickly performed on automated platforms. Unfortunately, this approach is prone to inaccuracy in the setting of pregnancy because of disruption of the original equilibrium—a process dependent upon dilution, temperature, buffer composition, affinity, and the concentration of the T4 antibody reagent and the T4-binding capacity within the serum sample (25). High protein concentrations in serum samples tend to result in higher FT4 values, whereas low protein concentrations are likely to yield lower FT4 values. In order to decrease nonspecific binding and neutralize the effect of nonesterified fatty acids on serum FT4, albumin is added in some assays. Albumin itself, however, binds T4 and when added in sufficient amounts, may disrupt the equilibrium. Nevertheless, the currently used FT4 immunoassays perform reasonably well in many circumstances, and most often accurately report both low FT4 levels in the setting of thyroid hormone deficiency and high FT4 levels in the setting of thyroid hormone excess (26,27).

Sera of pregnant women are characterized by higher concentrations of TBG and nonesterified fatty acids and by lower concentrations of albumin relative to the sera of nonpregnant women. In part because of this, many current FT4 analog immunoassays fail dilutional assessment (25,28). Because FT4 reference intervals in pregnancy vary widely between methods, interpretation of FT4 values requires method-specific as well as trimester-specific ranges (10,11,27). Whereas it is customary for manufacturers to suggest that laboratories establish their own reference range for such a test, this suggestion is frequently impractical for FT4 assessment because it is especially difficult to recruit subjects with specific conditions such as pregnancy from which to independently establish method- and trimester-specific reference ranges. Therefore, laboratories customarily adopt the pregnancy ranges provided by the test manufacturers. Typically, the characteristics of these reference pregnant cohorts are not disclosed, and differences in iodine intake and ethnicity may compromise the ability to generalize the manufacturer ranges across different populations. This problem adds to the complexity of accurate measurement of serum FT4 in the pregnant individual.

Current uncertainty around FT4 estimates in pregnancy has led some to question the wisdom of relying on any FT4 immunoassays during pregnancy (29,30). In contrast, measurement of TT4 and the calculated FT4 index do show the expected inverse relationship with serum TSH (29). This finding suggests that TT4 measurements may be superior to immunoassay measurement of FT4 measurements in pregnant women. However, reference values should take the 50% increase in TBG witnessed during pregnancy into account by calculating the FT4 index using a serum thyroid hormone uptake test (such as the thyroid hormone binding ratio). Changes in total serum T4 concentration through pregnancy among euthyroid women have been previously reported (5). Changes are predictable, with an increase in TT4 concentration from weeks 7–16 of gestation, ultimately reaching ∼50% above the prepregnancy level. This level is then sustained through pregnancy. Therefore, a clinically acceptable upper range determination can be calculated by shifting the nonpregnant limit 50% higher. However, this limit can only be used after week 16 of pregnancy. If a T4 measurement is required before that time (i.e., weeks 7–16 of pregnancy), a calculation can be made for the upper reference range based on increasing the nonpregnant upper reference limit by 5% per week, beginning with week 7. For example, at 11 weeks of gestation (4 weeks beyond week 7), the upper reference range for T4 is increased by 20% (4 weeks × 5%/week) (5).

As described above, free thyroid hormones can also be measured in the dialysate or ultrafiltrate using online solid phase extraction (LC/MS/MS). However, this approach is time consuming, costly, and often impractical. Using direct equilibrium dialysis and LC/MS/MS, the 95% FT4 reference intervals decrease gradually with advancing gestational age: from 1.08–1.82 ng/dL in week 14 to 0.86–1.53 ng/dL in week 20 (31). Serum FT4 by LC/MS/MS correlates very well with serum FT4 measured by classical equilibrium dialysis, but correlation with results from the FT4 immunoassay are less satisfactory (8). The use of isotope dilution-LC/MS/MS for measuring T4 in the dialysate from equilibrium dialysis of serum is helpful to obtain a gold-standard reference measurement procedure for serum FT4 (32). This assay technology, unfortunately, is currently not widely available due to high instrument and operating costs.

■ RECOMMENDATION 2

 The accuracy of serum FT4 measurement by the indirect analog immunoassays is influenced by pregnancy and also varies significantly by manufacturer. If measured in pregnant women, assay method-specific and trimester-specific pregnancy reference ranges should be applied.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 3

 In lieu of measuring FT4, TT4 measurement (with a pregnancy-adjusted reference range) is a highly reliable means of estimating hormone concentration during the last part of pregnancy. Accurate estimation of the FT4 concentrations can also be done by calculating a FT4 index.

 Strong recommendation, moderate-quality evidence.

Question 4: What is the impact of severe iodine deficiency on the mother, fetus, and child?

Maternal dietary iodine deficiency results in impaired maternal and fetal thyroid hormone synthesis. Low thyroid hormone values stimulate increased pituitary TSH production, and the increased TSH stimulates thyroid growth, resulting in maternal and fetal goiter (39). In areas of severe iodine deficiency, thyroid nodules can be present in as many as 30% of pregnant women (40). Severe iodine deficiency in pregnant women has been associated with increased rates of pregnancy loss, stillbirth, and increased perinatal and infant mortality (41).

Normal levels of thyroid hormone are essential for neuronal migration, myelination, and other structural changes of the fetal brain. Because thyroid hormones are needed throughout pregnancy, iodine deficiency affects both maternal and fetal thyroid hormone production, and insufficient iodine intake can lead to detrimental effects. Specifically, maternal and fetal iodine deficiency in pregnancy have adverse effects on the cognitive function of offspring (42 –45). Children whose mothers were severely iodine deficient during pregnancy may exhibit cretinism, characterized by profound intellectual impairment, deaf-mutism, and motor rigidity. Iodine deficiency is the leading cause of preventable intellectual deficits worldwide (42).

Universal salt iodization is the most cost-effective way of delivering iodine and improving maternal and infant heath (46).

Question 5: What is the impact of mild to moderate iodine deficiency on the mother, fetus, and child?

Groups of pregnant women whose median UICs are 50–150 μg/L are defined as mildly to moderately iodine deficient. Women with mild to moderate iodine deficiency during pregnancy are at increased risk for the development of goiter (39) and thyroid disorders (47). Low maternal UIC in pregnancy has been associated with reduced placental weight and neonatal head circumference (48). However, in areas with adequate dietary iodine intake, variations in maternal UICs have a limited influence on physical developmental outcomes (49). Mild to moderate maternal iodine deficiency has also been associated with attention deficit and hyperactivity disorders in children (50) as well as impaired cognitive outcomes (51 –53). In an iodine-deficient area iodized salt intake before pregnancy did improve maternal thyroid function; no difference in child neurodevelopment was noted (54), but improvement has been noted in other studies (55).

Question 6: What is the iodine status of pregnant and breastfeeding women in the United States?

Surveillance of urinary iodine values of the U.S. population has been carried out at intervals since 1971. Following a precipitous drop in urinary iodine values between 1971 and 1994, U.S. dietary iodine intake has stabilized (56 –61). The U.S. population overall remains iodine sufficient. However, U.S. women of reproductive age are the most likely group to have low urinary iodine values. According to the World Health Organization (WHO) guidelines, median urinary iodine values for pregnant women between 149 and 249 μg/L are consistent with optimal iodine intake (42). In the 2005–2010 National Health and Nutrition Examination Survey (NHANES) surveys, the median UIC for U.S. pregnant women was 129 μg/L, consistent with mild iodine deficiency (62). Current data regarding iodine sufficiency among lactating U.S. women are very limited. It is possible that a subset of pregnant and lactating U.S. women may have mildly to moderately inadequate dietary iodine intake resulting in insufficient amounts of iodine in the breast milk to meet infants' dietary requirements, but studies have been small and inconsistent (63 –65).

Question 7: What is the iodine status of pregnant and breastfeeding women worldwide?

Since 1990, the number of households worldwide using iodized salt has risen from less than 20% to more than 70% (66). Despite these advances, however, 30 countries remain iodine deficient, and iodine deficiency remains the leading cause of preventable intellectual deficits worldwide (44). Recent reports highlight the value of iodized salt in correcting iodine deficiency states in India (67,68), although remaining iodine deficiency was noted in one report (69). Only 6% of pregnant women had low UICs in rural Bangladesh (70), whereas 80% had low urine iodine concentrations in Pakistan (71). In Shanghai, China, iodine deficiency was noted in a subset of pregnant women (72), whereas in Shenyang city slight iodine excess was noted, with a consequent increase in subclinical hypothyroidism (73). Iodine status in Korea is more than adequate (median UIC 427 μg/L) (74). Iodine status in Japan is also more than adequate (median UIC 328 μg/L) (9). On the African continent, iodine status in pregnancy was inadequate in Niger (75) and was also poor in Ethiopia (76). Iodine nutrition in Iran was sufficient (77). In Brazil, median UIC was 138 μg/L, possibly due to reduced concentration of iodine in salt (78). In Europe many countries, including Belgium, the Czech Republic, Denmark, France, Latvia, Norway, Spain, and the United Kingdom, have recorded significant iodine deficiency in their pregnant populations (47,79 –85).

Question 8: Does iodine supplementation in pregnancy and lactation improve outcomes in severe iodine deficiency?

In areas of severe iodine deficiency, iodine supplementation of mothers prior to conception or in early pregnancy results in children with improved cognitive performance relative to children of mothers given a placebo (86 –88). The prevalence of cretinism and other severe neurological abnormalities is significantly reduced (89). Maternal iodine supplementation in severely iodine-deficient areas also decreases rates of stillbirth and neonatal and infant mortality (90,91). Oral administration of iodized oil can increase birth weight in addition to correcting iodine deficiency (92).

A recent randomized clinical trial demonstrated that in moderately to severely iodine-deficient areas without universal salt iodization, lactating women who receive one dose of 400 mg oral iodized oil after delivery can provide adequate iodine to their infants through breast milk for at least 6 months. Direct infant iodine supplementation was less effective at improving infant iodine status (93).

Question 9: Does iodine supplementation in pregnancy and lactation improve outcomes in mildly to moderately iodine-deficient women?

Eight controlled trials of iodine supplementation in mildly to moderately iodine-deficient pregnant European women have been published (94 –101), although doses and timing of iodine supplementation varied and only two trials examined effects on offspring development. Iodine supplementation of moderately deficient pregnant women appears to consistently decrease maternal and neonatal thyroid volumes and thyroglobulin (Tg) levels. Effects on maternal thyroid function have been mixed, with significant maternal TSH decreases with supplementation described in four (88,90,91,95) of the eight published trials, and increases in maternal T4 or FT4 noted in just two (90,95). A reduction in cord TSH also indicates improvement in gestational iodine status (102).

In two nonrandomized studies, neurodevelopmental outcomes were improved in children from mildly to moderately iodine-deficient areas whose mothers received iodine supplementation early in pregnancy (87,95). Another study failed to show neuropsychological improvement in 16-month-old children of mothers who received supplementation (103). The timing of supplementation is likely to be critical because the beneficial effects of iodine on offspring development appeared to be lost if supplementation is started after 10–20 weeks gestation. If iodine supplementation is started before pregnancy in iodine-deficient women better maternal thyroid function can be observed but, depending on dose and the timing of initiation, supplementation may not fully correct iodine deficiency in an already iodine-deficient population (104). A meta-analysis concluded that iodine supplementation improves some maternal thyroid indices and may benefit aspects of cognitive function in school age children, even in marginally iodine-deficient areas (105). Another review highlighted the lack of high-quality evidence in relation to these outcomes and suggested that randomized controlled trial (RCTs) may not be feasible where iodine supplementation is common (106).

Recently, there has been controversy regarding whether it is ethical to perform randomized clinical trials of iodine supplementation in pregnancy in regions that are mildly to moderately iodine deficient (107 –109). Caution in accepting the necessity of supplementation has been expressed, especially in areas where iodized salt is already in use (110).

No trials to date have specifically examined the effects of iodine supplementation in lactation in mildly to moderately iodine-deficient regions.

Question 10: What is the recommended daily iodine intake in women planning pregnancy, women who are pregnant, and women who are breastfeeding?

Iodine is an essential nutrient required for thyroid hormone production and is primarily derived from the diet and from vitamin/mineral preparations. The U.S. Institute of Medicine recommended dietary allowances to be used as goals for individual total daily iodine intake (dietary and supplement) are 150 μg/d for women planning a pregnancy, 220 μg/d for pregnant women, and 290 μg/d for women who are breastfeeding (111). The WHO recommends 250 μg/d for pregnant and lactating women (39). This level is supported by a study of more than 7000 pregnant Chinese women in whom it was found that subclinical hypothyroidism and hypothyroxinemia were least common at the urinary iodine range of 150–249 μg/L, but risk for both these abnormalities rose when the UIC was lower or higher than this range (73).

Dietary iodine sources vary regionally. Sources of iodine in the United States diet have been difficult to identify, in part because there are a wide variety of potential sources and food iodine content is not listed on packaging. Iodized salt remains the mainstay of iodine deficiency disorder elimination efforts worldwide. However, salt iodization has never been mandated in the United States and only approximately 50% of salt sold for household use in the United States is iodized (112). In the United States, dairy foods are another important source of dietary iodine because of iodine in cattle feed and the use of iodophor disinfectants by the dairy industry (113 –115). Commercially baked breads have been another major source of iodine in the United States because of the use of iodate bread conditioners (115). However, the use of iodate bread conditioners has decreased over the past several decades. Other sources of iodine in the United States diet are seafood, eggs, meat, and poultry (116). Foods of marine origin generally have high concentrations of iodine because marine animals concentrate iodine from seawater (96 –98), although the amount of iodine in different fish and shellfish species is quite variable.

In the United States, the dietary iodine intake of individuals cannot be reliably ascertained either by patient history or by any laboratory measure. Because of concerns that a subset of pregnant U.S. women may be mildly to moderately iodine deficient and an inability to identify individual women who may be at risk, the ATA has previously recommended 150 μg daily iodine supplementation for all North American women who are pregnant or breastfeeding (117). More recently the Endocrine Society, Teratology Society, and American Academy of Pediatrics have also advocated iodine supplementation for pregnant and lactating U.S. women (15,118,119). The goal is supplementation of dietary iodine intake, rather than its replacement. Special attention may need to be paid to those with dietary restrictions (e.g., lactose intolerant, gluten intolerant, low-carbohydrate, or vegan) because those individuals may have additional needs for supplementation (120).

Unfortunately, recommendations regarding iodine supplementation in the United States have not been widely adopted. In the NHANES 2001–2006 dataset, only 20% of pregnant women and 15% of lactating women reported ingesting iodine-containing supplements (121). When assessed in 2009, of the 223 types of prenatal multivitamins available in the United States, only 51% contained any iodine (122). However, this may recently have changed; in 2015 the Council for Responsible Nutrition, the U.S. supplement industry trade group, recommended that their members include 150 μg iodine in all prenatal multivitamin preparations. Iodine in U.S. prenatal multivitamins is typically derived either from potassium iodide or from kelp. The iodine content in prenatal multivitamin brands containing kelp may be inconsistent because of variability in kelp iodine content (123). Women consuming levothyroxine (LT4) regularly do not require supplemental iodine because the substrate is no longer needed for hormone formation.

The task force recommendations for iodine supplementation in the setting of lactation and breastfeeding are provided in Section XI.

■ RECOMMENDATION 5

 All pregnant women should ingest approximately 250 μg iodine daily. To achieve a total of 250 μg iodine ingestion daily, strategies may need to be varied based on country of origin.

 Strong recommendation, high-quality evidence.

■ RECOMMENDATION 6

 In most regions, including the United States, women who are planning pregnancy or currently pregnant, should supplement their diet with a daily oral supplement that contains 150 μg of iodine in the form of potassium iodide. This is optimally started 3 months in advance of planned pregnancy.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 7

 In low-resource countries and regions where neither salt iodization nor daily iodine supplements are feasible, a single annual dose of ∼400 mg of iodized oil for pregnant women and women of childbearing age can be used as a temporary measure to protect vulnerable populations. This should not be employed as a long-term strategy or in regions where other options are available.

 Weak recommendation, moderate-quality evidence.

■ RECOMMENDATION 8

 There is no need to initiate iodine supplementation in pregnant women who are being treated for hyperthyroidism or who are taking LT4.

 Weak recommendation, low-quality evidence.

Question 11: What is the safe upper limit for iodine consumption in pregnant and breastfeeding women?

Most people are tolerant of chronic excess dietary iodine intake due to a homeostatic mechanism known as the Wolff–Chaikoff effect (124,125). In response to a large iodine load, there is a transient inhibition of thyroid hormone synthesis. Following several days of continued exposure to high iodine levels, escape from the acute Wolff–Chaikoff effect is mediated by a decrease in the active transport of iodine into the thyroid gland, and thyroid hormone production resumes at normal levels (126).

Some individuals do not appropriately escape from the acute Wolff–Chaikoff effect, making them susceptible to hypothyroidism in the setting of high iodine intake. The fetus may be particularly susceptible, since the ability to escape from the acute Wolff–Chaikoff effect does not fully mature until about week 36 of gestation (127,128).

Tolerable upper intake levels for iodine have been established to determine the highest level of daily nutrient intake that is likely to be tolerated biologically and to pose no risk of adverse health effects for almost all individuals in the general population. The upper intake levels are based on total intake of a nutrient from food, water, and supplements and apply to chronic daily use. The U.S. Institute of Medicine has defined the tolerable upper limit for daily iodine intake as 1100 μg/d in all adults, including pregnant women (96) and the WHO has stated that daily iodine intake >500 μg may be excessive in pregnancy. Recent population data support the WHO threshold (73). The exception may be communities that have historically and consistently consumed greater than 500 μg daily without experiencing adverse effects (e.g., Japan).

Medications may be a source of excessive iodine intake for some individuals. Amiodarone, an antiarrhythmic agent (129), contains 75 mg iodine per 200 mg tablet. Iodinated intravenous radiographic contrast agents contain up to 380 mg/mL. Some topical antiseptics contain iodine, although systemic absorption is generally not clinically significant in adults except in patients with severe burns (130). Iodine-containing anti-asthmatic medications and expectorants are occasionally used. In addition, some dietary supplements such as kelp and some iodine preparations may contain very large amounts of iodine (several thousand times higher than the daily upper limit) and should not be taken. Ingestion of iodine and kelp supplements containing in excess of 500 μg/d is not recommended in pregnancy or lactation (131).

Concern exists that some populations may be exposed to excess iodine, possibly resulting in a high prevalence of thyroid dysfunction (132,133), an increased rate of hyperthyrotrophinemia (134), and an increased rate of hyperthyroid newborns (135). In addition, iodine-induced hypothyroidism has been reported in infants exposed to excess iodine from radiocontrast agents (136). It should be recognized that even low-dose iodine supplementation may trigger thyroid autoimmunity in a small proportion of women (73). The exception may be communities that have historically and consistently consumed greater then 500 μg daily without experiencing adverse effects. In one study, approximately one third of Japanese pregnant women demonstrated higher UIC beyond 500 μg/Cr, but there were no problems in pregnancy progress and fetal development (9).

■ RECOMMENDATION 9

 Excessive doses of iodine exposure during pregnancy should be avoided, except in preparation for the surgical treatment of GD. Clinicians should carefully weigh the risks and benefits when ordering medications or diagnostic tests that will result in high iodine exposure.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 10

 Sustained iodine intake from diet and dietary supplements exceeding 500 μg daily should be avoided during pregnancy due to concerns about the potential for fetal thyroid dysfunction.

 Strong recommendation, moderate-quality evidence.

Question 12: What is the prevalence of thyroid autoantibodies in pregnant women?

Anti-TPO or anti-Tg thyroid autoantibodies are present in 2% to 17% of unselected pregnant women (47,137 –144). The prevalence of antibodies varies with ethnicity. In U.S. populations, thyroid antibodies are most frequent in Caucasian and Asian women and least frequent in African Americans (139,145). Dietary iodine intake may also be associated with anti-thyroid Ab positivity during pregnancy. Shi and colleagues (73) recently demonstrated a U-shaped relationship between UICs and antibody positivity among pregnant women.

A recent study from Belgium in women seeking fertility treatment showed that both TPOAb and TgAb were present in 8% of women, while 5% demonstrated isolated Tg antibodies and 4% demonstrated isolated TPOAb concentrations (146). Those women with isolated TgAb positivity had a significantly higher serum TSH than women without thyroid autoimmunity. While the task force acknowledges that testing for thyroid autoimmunity using only TPOAb would likely miss a small proportion of women with isolated Tg antibodies, we note that the vast majority of studies investigating thyroid autoimmunity and clinical outcomes used only TPOAb measurements. For this reason, the task force recommends assessment of TPOAb when testing for the presence of thyroid autoimmunity.

Question 13: What is the natural history of antithyroid antibodies in pregnant women?

Question 14: How should euthyroid women who are thyroid antibody (Ab)-positive be monitored during pregnancy?

In women with thyroid autoimmunity, hypothyroidism may occur because of the stress of pregnancy because the ability of the thyroid to augment hormone production is compromised. In 1994, Glinoer et al. (147) performed a prospective study of 87 euthyroid (TSH ≤4 mU/L) women who were positive for TPOAb, TgAb, or both. Twenty percent of women in the study developed a serum TSH >4 mU/L during gestation despite normal prepregnancy TSH values. Antithyroid Ab titers were highest in the first trimester, although they decreased by about 60% over the course of gestation. Twelve years later, in a prospective study, Negro et al. (28) demonstrated similar results. The authors found that in TPOAb-positive euthyroid women, TSH levels increased as gestation progressed, from a mean of 1.7 mU/L (12th week) to 3.5 mU/L (term), with 19% of women having a supranormal TSH value at delivery. Because the risk of TSH elevation is increased in this population, increased surveillance of euthyroid thyroid Ab–positive women should occur.

TPO antibodies are able to cross the placenta. At the time of delivery, cord blood TPOAb levels strongly correlate with third-trimester maternal TPOAb concentrations (148). However, maternal passage of either TPOAb or TgAb is not associated with fetal thyroid dysfunction.

■ RECOMMENDATION 11

 Euthyroid pregnant women who are TPOAb or TgAb positive should have measurement of serum TSH concentration performed at time of pregnancy confirmation and every 4 weeks through midpregnancy.

 Strong recommendation, high-quality evidence.

Question 15: Should euthyroid women with thyroid autoimmunity be treated with selenium?

Some studies evaluating nonpregnant women have shown that selenium can diminish TPOAb concentrations (149 –152). However, this reduction has not been observed in all studies (153). Negro et al. (154) noted that euthyroid TPOAb-positive pregnant women randomized to treatment with 200 μg/d selenium not only had a significant decrease in the frequency of postpartum thyroid dysfunction (p < 0.01) but also had lower TPOAb concentrations during pregnancy compared to those in the untreated group. Importantly, this trial did not measure urinary iodine, a potential confounder because iodine status may influence the thyroidal effects of selenium. However, in another recent randomized clinical trial (155) performed in mildly iodine-deficient British pregnant women, treatment with 60 μg of selenium daily did not affect TPO concentrations or TPOAb positivity. Thus, conflicting data regarding selenium supplementation make any generalized recommendation unreliable, especially to regions with different intakes of iodine, selenium, or both. In addition, patients treated with selenium could be at higher risk for developing type 2 diabetes mellitus (156). For these reasons, the risk-to-benefit comparison does not presently support routine selenium supplementation of TPOAb-positive women during pregnancy.

■ RECOMMENDATION 12

 Selenium supplementation is not recommended for the treatment of TPOAb-positive women during pregnancy.

 Weak recommendation, moderate-quality evidence.

Question 16: Is there an association between thyroid antibodies and sporadic spontaneous pregnancy loss in euthyroid women?

Spontaneous pregnancy loss (miscarriage), occurs in 17%–31% of all gestations (157,158). A spontaneous pregnancy loss is usually defined as one occurring at less than 20 weeks of gestation. The individual risk varies according to clinical factors including maternal age, family history, environmental exposures, and medical comorbidities (159). Pregnancy losses are a significant emotional burden to patients and may also result in bleeding, infection, pain, and need for surgical intervention.

Endocrine disorders have been previously recognized as risk factors for spontaneous pregnancy loss. Patients with poorly controlled diabetes mellitus may have up to a 50% risk of loss (160). Thyroid dysfunction has similarly been associated with increased pregnancy loss (161). Stagnaro-Green and colleagues (162) first demonstrated an association between pregnancy loss and thyroid antibodies in a prospective observational study. Patients who were positive for thyroid antibodies (TPOAb, TgAb, or both) demonstrated a 2-fold increase in the risk for pregnancy loss (17% vs. 8.4%, p = 0.01). Since that time, numerous other studies have examined the association between maternal antithyroid Ab status and pregnancy loss risk, showing similar findings. In a recent meta-analysis of eight case–control studies, the pooled OR for pregnancy loss in women with thyroid autoimmunity versus women without antithyroid antibodies was 2.55 [95% CI 1.42–4.57]. Meta-analysis of 14 cohort studies showed a similar increased OR of 2.31 [95% CI 1.90–2.82] (163). However, Ab-positive women were noted to be slightly older and to have slightly higher serum TSH values. In support of these data, a separate meta-analysis similarly found an increase in loss rate among thyroid Ab–positive women (OR 3.90 for cohort studies [95% CI 2.48–6.12]; OR 1.80 for case control studies [95% CI 1.25–2.60]) (164).

Although a clear association has been demonstrated between thyroid antibodies and spontaneous pregnancy loss, it does not prove causality and the underlying mechanisms for such an association remain unclear. Three research groups have demonstrated one possible mechanism through increased fetal resorption in active immunization murine models (165 –167). Several other mechanistic hypotheses have been proposed, including Ab-mediated mild thyroid hypofunction, cross-reactivity of antithyroid antibodies with hCG receptors on the zona pellucida, the presence of concurrent non–organ-specific autoimmunity, and increased levels of endometrial cytokines in women with thyroid autoimmunity (168).

Question 17: Is there an association between thyroid antibodies and recurrent spontaneous pregnancy loss in euthyroid women?

Recurrent pregnancy loss is defined as either two consecutive spontaneous losses or three or more spontaneous losses, and it may occur in up to 1% of all women (169). Several causes have been reported, including parental chromosomal anomalies, immunologic derangements, uterine pathology, and endocrine dysfunction (170).

In a case–control study, Iravani and colleagues (171) reported that patients with primary recurrent pregnancy losses (three or more) had a higher incidence of TgAb and/or TPOAb positivity (OR 2.24 [95% CI 1.5–3.3]). Kutteh et al. (172) reported similar findings, with an increased rate of positivity for TgAb, TPOAb, or both in 700 women with recurrent pregnancy loss compared to 200 healthy controls (22.5% vs. 14.5%, p = 0.01). On the other hand, in a prospective observational study, Esplin and colleagues (173) demonstrated no difference in positivity for TgAb, TPOAb, or both between patients with recurrent pregnancy loss and healthy controls. Pratt and colleagues (174) reported a higher rate of subsequent pregnancy loss in patients with recurrent losses and thyroid Ab positivity. In a larger study with a similar population, Rushworth and colleagues (175) reported no significant difference in live birth rates between women with recurrent losses who were TgAb or thyroid microsomal Ab positive and those who were not. In a case–control study, Lata et al. (176) found that the prevalence of TPO antibodies was higher in women with recurrent pregnancy loss than in healthy pregnant controls without a history of recurrent loss (31% vs. 18%. p = 0.031). A meta-analysis of eight studies that included 460 Ab-positive patients and 1923 controls noted a significant association between thyroid Ab positivity and recurrent pregnancy loss (OR 2.3 [95% CI 1.5–3.5]) (177).

Thus, the data for an association between thyroid antibodies and recurrent pregnancy loss are less robust than for sporadic loss. This finding may be because recurrent pregnancy loss has many potential causes, and endocrine dysfunction may only account for 15%–20% of all such cases (170). Many of the studies described above did not control for other potential causes of recurrent losses. One intriguing study reported an apparent interaction of antiphospholipid antibodies and antithyroid antibodies in the risk for recurrent pregnancy loss (178). In support of this, Kim and colleagues (179) reported that women with recurrent pregnancy loss who were antithyroid Ab positive also demonstrated higher levels of anticardiolipin Ab and other non–organ-specific antibodies.

Question 18: Does treatment with LT4 or intravenous immunoglobulin therapy decrease the risk for pregnancy loss in euthyroid women with thyroid autoimmunity?

Negro and colleagues (28) reported a prospective, randomized interventional trial of LT4 in euthyroid patients who were TPOAb positive. The researchers reported a significantly decreased rate of pregnancy loss in the LT4-treated group (3.5% vs. 13.8%, p < 0.05). A limitation of the study is that the mean estimated gestational age at starting LT4 was 10 weeks, and all but one of the eight losses in the untreated group had occurred before 11 weeks. This finding raises uncertainty as to the impact of the intervention upon the endpoint.

However, in a separate nonrandomized, retrospective study, Lepoutre et al. (180) analyzed data from 65 TPOAb-positive pregnant women with serum TSH values of 1–3.5 mU/L at the first antenatal visit. Thirty-four of these women were treated with 50 μg LT4 daily starting at a mean 10 weeks gestation, while the others were not treated. None of the LT4-treated women miscarried, but 5 of 31 untreated women (16%) experienced pregnancy loss. Although limited in nature, these data support the findings of Negro et al. (28) and begin to suggest a potential benefit to this treatment approach. The underlying mechanism is, however, uncertain.

Three small nonrandomized case series have been published on the use of intravenous immunoglobulin (IVIG) therapy for the prevention of recurrent pregnancy loss in women with antithyroid antibodies (181 –183). The live birth rates ranged from 80% to 95%, and the one study with a control group (consisting of women who refused IVIG therapy) reported a highly significant improvement in live births in the IVIG-treated cohort (95% vs. 0% p = 0.001) (182). Comparison of a LT4 intervention to an IVIG intervention in one study demonstrated a higher rate of term delivery in the LT4-treated group (183). However, all three studies had serious design flaws including small sample sizes, heterogeneous patient populations, lack of or limited randomization, and differences in the timing of treatment initiation. Nonetheless, intervention trials with LT4 (and less so IVIG) in euthyroid TPOAb-positive women with recurrent abortion appear to show a possible decrease in miscarriage rates. Further randomized trials are needed to better understand the effectiveness of both LT4 and IVIG intervention. At present, however, the cost, complexity, and side effect profile associated with IVIG infusion must be noted and make its use undesirable given the questionable benefit. In contrast, LT4 administration in low dosage (25–50 μg/d) is safe. Therefore, its use among patients with recurrent pregnancy loss may be reasonably considered in the setting of early gestation, especially when no other known cause of prior pregnancy loss has been identified.

The task force makes note that two randomized clinical trials are currently ongoing. The Thyroid AntiBodies and LEvoThyroxine study (TABLET) trial in the United Kingdom is randomizing euthyroid TPOAb-positive women with a history of infertility or recurrent losses to either LT4 or placebo to assess effects on live birth rates. Separately, the T4Lifetrial in the Netherlands is examining the effects of LT4 treatment in euthyroid pregnant women with a history of recurrent loss. The primary outcome of this trial is the live birth rate.

■ RECOMMENDATION 13

 Intravenous immunoglobulin treatment of euthyroid women with a history of recurrent pregnancy loss is not recommended.

 Weak recommendation, low-quality evidence.

■ RECOMMENDATION 14

 Insufficient evidence exists to conclusively determine whether LT4 therapy decreases pregnancy loss risk in TPOAb-positive euthyroid women who are newly pregnant. However, administration of LT4 to TPOAb-positive euthyroid pregnant women with a prior history of loss may be considered given its potential benefits in comparison with its minimal risk. In such cases, 25–50 μg of LT4 is a typical starting dose.

 Weak recommendation, low-quality evidence.

Question 19: Is there an association between thyroid autoantibody positivity and premature delivery?

Preterm deliveries are defined as those occurring before 37 weeks gestation. In the United States, 11.4% of all births are preterm (184). Preterm birth remains one of the most prevalent and morbid perinatal complications. It is the leading cause of neonatal death and is associated with increased risks for acute respiratory, gastrointestinal, immunologic, central nervous system (CNS), hearing, and vision problems, as well as longer-term motor, cognitive, visual, hearing, behavioral, and growth impairment (185). A decade ago, the annual cost of preterm delivery to the U.S. health care system was estimated at $26 billion (185). Preterm birth has remained difficult to predict, prevent, and treat primarily because there are multiple potential causes and pathways that end in premature labor (186). Examples include infection, trauma, cervical insufficiency, premature rupture of membranes, and maternal medical conditions.

The relationship between thyroid autoantibodies and preterm delivery has been investigated with mixed results. Glinoer et al. (187) reported in a prospective cohort that euthyroid women positive for either TPOAb or TgAb had a significantly increased incidence of preterm birth (16% vs. 8%, p < 0.005). Ghafoor et al. (188) evaluated 1500 euthyroid women and found an increase in preterm delivery in TPOAb-positive women compared to women who were TPOAb negative (26.8% vs. 8.0%, p < 0.01). In contrast, Iijima et al. (189) did not find an increased risk for preterm birth in women positive for seven different autoantibodies and thyroid antibodies. This study had an unusually low rate of preterm birth in both study and control groups (3% vs. 3.1%). Haddow et al. (190) reported a significant increase in preterm premature rupture of the membranes, but not in preterm delivery, among women who were TPOAb and/or TgAb positive in the first trimester.

Importantly, three recent large prospective cohort studies failed to find significant associations between antithyroid antibody positivity and risk for premature delivery. However, in each of the studies, prematurity rates were nonsignificantly higher in TPOAb-positive women than in women who were antibody negative (19,190,191). By contrast, Karakosta et al. (192) found that euthyroid women (TSH <2.5 mIU/L) with positivity for TPOAb, TgAb, or both in early pregnancy were at increased risk for spontaneous preterm delivery (RR 1.7 [95% CI 1.1–2.8]). In a small cohort of 395 pregnant women, Kumru and colleagues (193) noted that spontaneous preterm delivery (<37 weeks) was more frequent in TPOAb-positive euthyroid (TSH <2.6 mU/L) women (OR 2.5 [95% CI 1.06–5.89]). Negro and colleagues found that untreated TPOAb-positive women had a substantially higher risk for preterm delivery than women without TPOAb (RR 12.18 [95% CI 7.93–18.7]) (28). A 2011 meta-analysis of seven studies, including about 23,000 participants, concluded that there was an association between thyroid autoimmunity and preterm delivery (OR = 1.6 [95% CI 1.44–1.94]) (194). In support, a meta-analysis of five cohort studies including 12,566 women similarly concluded that a positive association existed between the presence of thyroid Ab and preterm birth (OR 2.907 [95% CI 1.17–3.68]) (164). Finally, a third meta-analysis of 11 prospective cohorts including 35,467 participants determined that the relative risk for delivery at less than 37 weeks for women with positivity for TgAb, TPOAb, or both was 1.41 [95% CI 1.08–1.84] (195). Interestingly, in subgroup analyses, TPOAb but not TgAb positivity was associated with prematurity. Together, these data suggest that thyroid autoantibody positivity is associated with increased risk for preterm delivery.

Question 20: Does LT4 treatment of euthyroid women who are thyroid autoantibody positive reduce risk for premature delivery?

In contrast to association studies, interventional studies of LT4 therapy for the prevention of preterm delivery are sparse. Negro et al. (28) reported an increased risk of preterm delivery among euthyroid TPOAb-positive women compared to euthyroid TPOAb-negative women in the only prospective interventional trial to date (22.4% vs. 8.2%, p < 0.01). The TPOAb-positive subjects were randomized to either treatment with LT4 or no treatment, with the LT4 dose based on TSH level. The treated group had a significantly lower rate of preterm delivery than did the untreated group (7% vs. 22.4%, p < 0.05). This finding has not been replicated. The ongoing TABLET study in the United Kingdom, a randomized clinical trial examining effects of LT4 treatment of euthyroid women with a history of infertility or recurrent pregnancy loss, will examine preterm birth as a secondary outcome. Therefore, at present, there are insufficient data from which to draw any conclusion regarding the utility of LT4 administration for the purpose of reducing preterm delivery.

■ RECOMMENDATION 15

 Insufficient evidence exists to recommend for or against treating euthyroid pregnant women who are thyroid autoantibody positive with LT4 to prevent preterm delivery.

 No recommendation, insufficient evidence.

Question 21: Is thyroid autoimmunity in euthyroid pregnant women associated with adverse obstetric or child outcomes other than pregnancy loss and premature birth?

Antithyroid antibodies have been associated with perinatal death in some (191) but not all (140,196) studies. In two pregnancy cohorts an increased risk for placental abruption was observed in women with thyroid autoimmunity and without clinical hypothyroidism (140,190). Maternal thyroid autoimmunity has also been linked to postpartum depression (197,198) and to neonatal respiratory distress syndrome (196).

Several studies have examined associations between maternal thyroid autoimmunity and child development. In a nested case-control study, Li et al. (199) reported lower motor and intellectual development at age 25–30 months in the offspring of euthyroid women who were TPOAb positive compared to children of TPOAb-negative controls. Williams et al. (200) followed 97 full-term mothers and their children, assessing child neurocognition at age 5.5 years. Lower perceptual performance and motor scores were described in children of TgAb-positive mothers, and lower perceptual performance scores noted in children with TgAb-positive cord blood. However, no neurodevelopmental outcomes were associated with maternal or infant TPOAb status. Wasserman and colleagues (141) described higher rates of sensorineural hearing loss (22.7% vs. 4.3%, p = 0.004) at age 8 in children whose mothers were TPOAb positive in pregnancy compared to children of TPOAb-negative women. Although maternal thyroid function was not assessed in this study, the association remained significant after controlling for a known history of maternal hypothyroidism. In the same cohort, cognitive assessments were performed in children at ages 4 and 7 years (201). Maternal TPOAb positivity was associated with lower child intelligence quotient (IQ) at age 4, but effects were attenuated by age 7. The authors speculated that the lower IQ scores at age 4 might be mediated by sensorineural hearing loss. Ghassabian and colleagues (202) assessed a cohort of 3139 mother–child pairs in which child cognitive function was assessed at age 2.5 years and behavior assessed at age 3. Maternal TPOAb status did not predict child neurocognition, but TPOAb positivity was associated with externalizing problems in children (OR 1.64 [95% CI 1.17–2.29]), especially attention deficit/hyperactivity problems (OR 1.77 [95% CI 1.15–2.72]). This result remained significant after adjustment for maternal TSH. Finally, in a nested case–control study, Brown et al. (203) found that women with children on the autism spectrum were more likely to have had positive TPOAb during pregnancy than mothers of controls.

Question 22: Is overt thyroid dysfunction associated with infertility in women?

Irregular menses may occur in women with overt hyperthyroidism. Krassas and colleagues (207) reported that menstrual irregularities were present in 22% of hyperthyroid patients as compared to 8% in age- and weight-matched euthyroid controls. However, in a cross-sectional study, the prevalence of hyperthyroidism (both subclinical and overt) was similar in infertile women compared to fertile controls (208).

The risk of infertility in women with overt hypothyroidism is less well studied. In a study of 171 hypothyroid women with TSH concentrations >15 mU/L, 68% reported having irregular menses, far higher than the 12% rate of menstrual irregularities reported by euthyroid controls (209). In one cross-sectional study among 129 infertile women, 5% had serum TSH levels >4.5 mU/L (210). In a separate cross-sectional study among women age 18–50 years, the prevalence of infertility was 52.3% in women with GD and 47% among women with Hashimoto's thyroiditis; however, thyroid function was not reported in either group (211). Thus, despite imperfect data, the majority of evidence appears to support an association between overt thyroid dysfunction and an increased risk of infertility. Thyroid dysfunction is also reversible, and treatment is generally safe and may exert a positive effect on fertility. Therefore, it is reasonable to treat overt thyroid dysfunction in infertile women, with the goal of normalizing thyroid function.

Question 23: Is subclinical hypothyroidism associated with infertility in women?

Different definitions of subclinical hypothyroidism have been used in different studies examining this question, and results have been inconsistent. In a cross-sectional study of 704 women with infertility, the prevalence of TSH elevations was 2.3%, similar to background rates in the general population (212). In a prospective study, Poppe and colleagues (208) did not find increased rates of subclinical hypothyroidism among infertile women but did report slightly higher median serum TSH levels (1.3 vs. 1.1 mU/L) in the infertile women compared to controls. However, in a retrospective study, higher rates of subclinical hypothyroidism (13.9% vs. 3.9%) were reported in infertile women as compared to fertile controls (213). Among women with baseline TSH values 2.5–5 mU/L, thyrotropin releasing hormone (TRH) stimulation testing caused TSH increases to >30 mU/L in 46% of women with female-factor infertility versus 7% of women whose infertility was due to male factors (214). Among women presenting with infertility, TSH levels were highest among women with ovulatory dysfunction and unknown causes of infertility and lower among those women with tubal infertility and whose infertility was due to male factors (215). In an uncontrolled study, 94 infertile women with hypothyroidism (TSH >4.2 mU/L) were treated with LT4 (25–150 μg/d) and 72 of them spontaneously conceived within 1 year (216). In a retrospective study, Yoshioka et al. (217) reported that 84.1% of infertile women with subclinical hypothyroidism (TSH >3.0 mU/L) successfully conceived and their infertility duration was shortened after LT4 therapy, suggesting that LT4 therapy may enhance fertility in patients with subclinical hypothyroidism. Importantly, whether or not LT4 treatment increases the likelihood of conception in subclinically hypothyroid women not undergoing ART has not been studied in controlled trials. Thus, insufficient data exist for recommending for or against routine LT4 therapy in subclinically hypothyroid, thyroid autoantibody–negative infertile women who are attempting conception but not undergoing ART.

Question 24: Is thyroid autoimmunity linked to infertility in women?

Limited evidence suggests that women with female-factor infertility are more likely to be TPOAb positive than age-matched women who are not infertile, even if euthyroid (208). In addition, the prevalence of antithyroid antibodies may be higher in women with polycystic ovarian syndrome (PCOS) than in age-matched controls (218). Antithyroid antibodies are detectable in the ovarian follicles of women with thyroid autoimmunity and correlate with serum antibody levels (219), although whether such antibodies interfere with the fertilization potential of maturing oocytes is unknown. Among infertile women with PCOS, the presence of antithyroid antibodies has been associated with a decreased likelihood of developing ovarian follicles in response to treatment with clomiphene citrate (220).

■ RECOMMENDATION 16

 Evaluation of serum TSH concentration is recommended for all women seeking care for infertility.

 Weak recommendation, moderate-quality evidence.

■ RECOMMENDATION 17

 LT4 treatment is recommended for infertile women with overt hypothyroidism who desire pregnancy.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 18

 Insufficient evidence exist to determine if LT4 therapy improves fertility in subclinically hypothyroid, thyroid autoantibody–negative women who are attempting natural conception (not undergoing ART). However, administration of LT4 may be considered in this setting given its ability to prevent progression to more significant hypothyroidism once pregnancy is achieved. Furthermore, low dose LT4 therapy (25–50 μg/d) carries minimal risk.

 Weak recommendation, low-quality evidence.

■ RECOMMENDATION 19

 Insufficient evidence exists to determine if LT4 therapy improves fertility in nonpregnant, thyroid autoantibody–positive euthyroid women who are attempting natural conception (not undergoing ART). Therefore, no recommendation can be made for LT4 therapy in this setting.

 No recommendation, insufficient evidence.

Question 25: Is maternal subclinical hypothyroidism associated with worse ART outcomes?

Question 26: Does treatment of subclinically hypothyroid women improve ART outcomes?

In a retrospective cohort study among women undergoing intrauterine insemination (IUI), the use of LT4 to treat subclinical hypothyroidism (TSH >2.5 mU/L) was associated with higher pregnancy rates, and the use of LT4 to treat overt hypothyroidism with lower pregnancy rates, despite the fact that all LT4-treated patients had a serum TSH <2.5 at the time of IUI (221). In a separate prospective cohort study of 1477 women undergoing IUI, preconception serum TSH in the range of 0.4–4.99 mU/L was not associated with IUI parameters, pregnancy rates, or live birth rates per cycle (222).

The majority of evidence suggests that IVF outcomes do not differ between women with serum TSH <2.5 mU/L and those with very mild TSH elevations, defined as a TSH between 2.5 and 5 mU/L. A retrospective cohort of 195 IVF cycles that resulted in deliveries found that gestational age at birth was lower and mean birth weight was lower among women with preconception TSH >2.5 mU/L (223). In another retrospective cohort of 164 women undergoing IVF, clinical pregnancy rates were higher (22% vs. 9%, p = 0.045) in women with TSH ≤2.5 versus those with TSH >2.5 mU/L (224). However, in a retrospective cohort of 1055 women undergoing their first IVF cycle, in age-adjusted analyses there was no difference in pregnancy, pregnancy loss, or delivery rates among women with serum TSH <2.5 mU/L versus those with TSH <4.5 mU/L (225). Another recent retrospective study examined IVF outcomes in 627 women without known thyroid disease and concluded that the presence of prepregnancy subclinical hypothyroidism (serum TSH >4.5 mU/L) did not affect pregnancy rates, live birth rates, or miscarriage rates (226). A case–control study among women undergoing first IVF cycles found no difference in embryo quality among women with serum TSH 0.45–2.5 mU/L versus 2.5–4.5 mU/L (227). Another retrospective cohort similarly reported no associations between serum TSH in the range of 0.5–4.5 mU/L and IVF outcomes (228). Finally, in a retrospective study which aimed to control for embryo quality by limiting the cohort to 1599 euploid blastocyst transfer cycles, variations in serum TSH ≤2.5 mU/L were not associated with differences in implantation, pregnancy loss, or live birth rates (229).

However, data generally demonstrate that treatment of more significant elevations in TSH concentrations (although still classified as subclinical hypothyroidism) appears beneficial. A randomized clinical trial was conducted in women aged 20–40 years with subclinical hypothyroidism (serum TSH >4.5 mU/L with normal FT4) who were undergoing IVF (230). A total of 64 women were randomized to treatment with LT4 (50 μg/d starting at the time of initiation of ovarian stimulation, and titrated to maintain TSH <2.5 mU/L in the first trimester in those women who did achieve pregnancy) versus placebo. The treated women had higher rates of clinical pregnancy, lower rates of miscarriage, and higher delivery rates. Another trial randomized 64 infertile women with subclinical hypothyroidism (TSH >4.2 mU/L with normal FT4) to treatment with 50 μg/d LT4 versus placebo and similarly reported higher pregnancy rates, lower pregnancy loss rates, and higher live birth rates in the treated women than in the control group (231). A recent meta-analysis pooled results of these trials with a third study (232) examining the effects of LT4 treatment for TPOAb-positive euthyroid women undergoing ART and concluded that although LT4 treatment did not have any effect on clinical pregnancy rates (pooled relative risk 1.75 [95% CI 0.90–3.38]), it did result in a higher delivery rate (pooled relative risk 2.76 [95% CI 1.20–6.44]) (233).

Together, these data suggest that subclinical hypothyroidism likely affects ART in a dose-related fashion, such that impact worsens as TSH concentrations rise. It is also well known that TSH concentrations are variable and may differ by several milli-units per liter from week to week, despite no change in treatment (234). It therefore seems prudent to recommend treatment of subclinically hypothyroid women seeking pregnancy with ART for any TSH elevation >2.5 mU/L.

■ RECOMMENDATION 20

 Subclinically hypothyroid women undergoing IVF or intracytoplasmic sperm injection (ICSI) should be treated with LT4. The goal of treatment is to achieve a TSH concentration <2.5 mU/L.

 Strong recommendation, moderate-quality evidence.

Question 27: Is treated maternal hypothyroidism associated with worse ART outcomes compared to healthy controls?

In a retrospective cohort study, among 240 women aged ≤37 years undergoing first IVF retrieval cycles, 21 euthyroid (serum TSH 0.35–4.0 mU/L) women with treated hypothyroidism had lower pregnancy rates than the women without thyroid disease (235). However, a subsequent case–control study including 137 LT4-treated hypothyroid women (mean serum TSH 1.6 ± 0.7 mU/L) and 274 age-matched euthyroid controls determined that the presence of adequately treated hypothyroidism was not associated with inferior rates of pregnancy, implantation, or live birth following IVF/ICSI (236). Thus, no conclusion can be drawn from these data.

Question 28: Is maternal antithyroid Ab positivity associated with poorer outcomes following ART?

Studies examining ART outcomes in thyroid autoantibody-positive and -negative women have enrolled heterogeneous populations with differing underlying etiologies for infertility. Some, but not all studies have been limited to women who were euthyroid. Different ART/IVF protocols have also been employed in different places and over time. A meta-analysis of prospective cohort studies suggests that pregnancy rates following IVF do not differ between Ab-positive and Ab-negative women, but as discussed previously (see Section V), that risk of pregnancy loss is higher in women with thyroid autoantibodies positivity (237). However, a recent retrospective study examined IVF outcomes in 627 women without known thyroid disease and concluded that the presence of prepregnancy thyroid autoimmunity did not affect pregnancy rates, live birth rates, or pregnancy loss rates (226). Similarly, two recent retrospective cohorts reported no differences in pregnancy, pregnancy loss, or live birth rates in thyroid Ab-positive versus thyroid Ab–negative euthyroid women undergoing IVF with ICSI (238,239). Separately, a prospective cohort study found no differences in outcomes of IVF with ICSI in TPOAb-positive and/or TgAb-positive versus euthyroid women with Ab negativity (240). By contrast, a recent retrospective IVF cohort found that fertilization, implantation, and pregnancy rates were lower in 90 antithyroid Ab-positive compared to 676 Ab-negative women, but thyroid function was not reported in either group (241).

Question 29: Does treatment of antithyroid Ab–positive euthyroid women improve ART outcomes?

Based on a single small randomized clinical trial and one retrospective cohort, LT4 treatment for thyroid Ab-positive women without thyroid dysfunction undergoing IVF does not appear to improve outcomes (232,242). Two small trials suggest the potential for improved pregnancy rates in thyroid Ab-positive infertile women who are treated with glucocorticoids prior to ART (243,244). In the trial by Litwicka et al. (243), 60 euthyroid (TSH <2.5 mU/L) TPOAb- and/or TgAb-positive women undergoing IVF were randomized to prednisolone 5 mg/d starting from the day of oocyte retrieval and continuing through the first trimester. There were higher overall pregnancy (60% vs. 30%, p = 0.02) and live birth (46.6% vs. 20%, p = 0.05) rates in the treated versus untreated women. Turi and colleagues (244) randomized 48 TPOAb-positive infertile women to prednisone versus placebo for 4 weeks before IUI. The pregnancy rate was 33.3% in treated women compared to 8.4% in the placebo group (p = 0.03). Pregnancy loss rates were not significantly different between the two groups. Although these small trials appear promising, the risks of corticosteroid use in early pregnancy are not well understood (245).

■ RECOMMENDATION 21

 Insufficient evidence exists to determine whether LT4 therapy improves the success of pregnancy following ART in TPOAb-positive euthyroid women. However, administration of LT4 to TPOAb-positive euthyroid women undergoing ART may be considered given its potential benefits in comparison to its minimal risk. In such cases, 25–50 μg of LT4 is a typical starting dose.

 Weak recommendation, low-quality evidence.

■ RECOMMENDATION 22

 Glucocorticoid therapy is not recommended for thyroid autoantibody–positive euthyroid women undergoing ART.

 Weak recommendation, moderate-quality evidence.

Question 30: Does ovarian hyperstimulation alter thyroid function?

Although multiple ART protocols are currently in use, protocols for IVF or IVF with ICSI typically begin by inducing controlled ovarian hyperstimulation through the use of gonadotrophins, gonadotrophin-releasing hormone analogues, or gonadotrophin-releasing hormone antagonists in combination with gonadotrophins. Follicular development is monitored by ultrasound, and when leading follicles are large enough, hCG is administered to produce fully mature oocytes. Controlled ovarian hyperstimulation causes serum estradiol to rapidly rise to supraphysiologic levels (4000–6000 ng/L), comparable to those detected in late pregnancy (246). These hormonal manipulations may alter thyroid function. The presumed mechanism for this effect relates to the rise in TBG associated with high estrogen levels, which reduce free thyroid hormone concentrations and in turn feed back to cause serum TSH elevations. In addition, the administered hCG can directly stimulate thyroidal TSH receptors, causing increases in thyroid hormone and subsequent decreases in TSH. A systematic review (247) found inconsistent effects of ovarian stimulation on serum thyroid hormones. For example, during and up to 1 month after ovarian stimulation serum TSH was increased in three of five studies, while FT4 was increased in two studies, decreased in one study, and unchanged in another. In some women (248 –250), serum TSH levels increased to ≥2.5 mU/L following ovarian stimulation. Gracia and colleagues (251) reported increases in both serum TSH and FT4 during ovarian stimulation that peaked 1 week after hCG administration, with 44% of the women with baseline TSH ≤2.5 mU/L having increases to >2.5 mU/L. Reinblatt et al. (252) reported that serum TSH increased in parallel with serum estradiol during the course of ovarian stimulation, with more marked TSH increases seen among TPOAb-positive women.

Ovarian hyperstimulation syndrome (OHSS) is a complication of controlled ovarian hyperstimulation in which increased vascular permeability results in fluid shifts from intravascular to third space compartments. Mild OHSS, manifested by mild ascites and abdominal symptoms, occurs in up to 20%–33% of all IVF cycles. Severe OHSS, which occurs in 0.1%–2% of IVF cycles, may be associated with thrombosis and respiratory distress, and is occasionally fatal (253). A recent case report described a patient who had subclinical hypothyroidism prior to IVF that was adequately treated with LT4. The patient rapidly developed overt hypothyroidism in the setting of severe OHSS (254). However, Poppe et al. (246) prospectively studied thyroid function in women undergoing controlled ovarian hyperstimulation and found no differences in serum TSH or FT4 levels between 25 women who developed OHSS and 52 women who did not.

Among women with adequately treated hypothyroidism (i.e., receiving LT4 therapy), ovarian stimulation has been reported to induce hypothyroidism by the time of egg retrieval (255). In one cohort study, increases in serum TSH during ovarian stimulation were more marked in LT4-treated women than in women without underlying hypothyroidism 1 week after hCG administration. However, TSH levels no longer differed between the groups by the time of pregnancy testing (251). In a prospective cohort study in 72 treated hypothyroid women with serum TSH <2.5 mU/L prior to ovarian stimulation, serum TSH rose sequentially from 1.7 ± 0.7 mU/L at baseline to 2.9 ± 1.3 mU/L at the time of hCG administration, and to 3.2 ± 1.7 mU/L 16 days after hCG, with serum hCG >2.5 mU/L in 46 (64%) of subjects by the time hCG was administered (256). A retrospective cohort study demonstrated that among treated hypothyroid women who undergo IVF, 84% required LT4 dose increases, most within the first 5–7 weeks of gestation (257). However, a small prospective study found that treated hypothyroid patients who conceived with the help of gonadotrophin therapy did not require larger LT4 dose increases than treated hypothyroid women who conceived spontaneously (258).

■ RECOMMENDATION 23

 When possible, thyroid function testing should be performed either before or 1–2 weeks after controlled ovarian hyperstimulation because results obtained during the course of controlled ovarian stimulation may be difficult to interpret.

 Weak recommendation, moderate-quality evidence.

■ RECOMMENDATION 24

 In women who achieve pregnancy following controlled ovarian hyperstimulation, TSH elevations should be treated according to the recommendations outlined in Section VII. In nonpregnant women with mild TSH elevations following controlled ovarian stimulation, serum TSH measurements should be repeated in 2–4 weeks because levels may normalize.

 Weak recommendation, moderate-quality evidence.

Question 31: What is the definition of hypothyroidism in pregnancy?

Elevations in serum TSH concentrations during pregnancy should ideally be defined using pregnancy- and population-specific reference ranges. It is important to note that detection of an increased TSH concentration is not always synonymous with decreased FT4 concentrations. Frequently, elevated maternal TSH is detected when FT4 concentrations are normal. Conversely, low FT4 concentrations can be detected despite normal TSH concentrations. The latter situation is referred to as isolated hypothyroxinemia. Excepting the very rare scenarios noted above, serum TSH measurement remains the principal determinant of maternal thyroid status at the present time and should be used to guide treatment decisions and goals. Because substantial differences exist in the upper reference limit for TSH between different populations (Table 4), each practitioner and hospital should ideally seek to determine their own trimester-specific reference ranges, obtained from analysis of healthy, TPOAb-negative, and iodine-sufficient women. However, the task force recognizes that this goal is frequently not feasible. Therefore, this guideline also proposes a stepwise approach to evaluation, providing a practical guide for clinicians.

■ RECOMMENDATION 25

 In the setting of pregnancy, maternal hypothyroidism is defined as a TSH concentration elevated beyond the upper limit of the pregnancy-specific reference range.

 Strong recommendation, high-quality evidence.

■ RECOMMENDATION 26

 The pregnancy-specific TSH reference range should be defined as follows:

○ When available, population- and trimester-specific reference ranges for serum TSH during pregnancy should be defined by a provider's institute or laboratory and should represent the typical population for whom care is provided. Reference ranges should be defined in healthy TPOAb-negative pregnant women with optimal iodine intake and without thyroid illness.

Strong recommendation, high-quality evidence.

Question 32: How is isolated hypothyroxinemia defined in pregnancy?

Isolated hypothyroxinemia is typically defined as a FT4 concentration in the lower 2.5th–5th percentile of a given population in conjunction with a normal maternal TSH concentration.

Question 33: What adverse outcomes are associated with overt hypothyroidism during pregnancy?

Overt maternal hypothyroidism has consistently been shown to be associated with an increased risk of adverse pregnancy complications (275) as well as detrimental effects upon fetal neurocognitive development (276). Specific adverse outcomes associated with overt maternal hypothyroidism include increased risks of premature birth, low birth weight, pregnancy loss, and lower offspring IQ. Abalovich et al. (277) demonstrated that women with overt hypothyroidism carry an estimated 60% risk of fetal loss when not adequately treated. Separately, Leung et al. (278) demonstrated a 22% risk of gestational hypertension in pregnant women with overt maternal hypothyroidism. Finally, Allan et al. (260) similarly described an increased risk of fetal death among pregnant women with overt disease. Together, these data demonstrate a clear association between overt maternal hypothyroidism and risk to the maternal–fetal unit.

Question 34: What adverse outcomes are associated with subclinical hypothyroidism during pregnancy?

Subclinical hypothyroidism is variably associated with an increased risk of adverse pregnancy outcomes in most, but not all studies, partly because separate studies use differing cutoffs to define an elevated TSH concentration. Also, many studies do not account for TPOAb status. Table 5 provides an overview of all available observational and prospective studies (inclusive of at least 400 subjects) investigating the effect of maternal subclinical hypothyroidism upon pregnancy (19,192,193,259,279 –285,287 –289,291,294,295,306,314,318). Separately, Table 6 describes the additive adverse impact of TPOAb status upon maternal hypothyroidism (19,192,279,284,287 –289,291,295,314).

For the purposes of discussion, the investigations that have studied the association between elevated maternal TSH concentration and adverse clinical endpoints can be broadly grouped into three categories below based upon adverse endpoints. These include adverse effects on pregnancy outcome (i.e., pregnancy loss), adverse perinatal outcomes (i.e., premature delivery, hypertensive disorders), and adverse neurocognitive outcomes (IQ) in offspring.

Question 35: What adverse outcomes are associated with isolated hypothyroxinemia in pregnancy?

Pop and colleagues (312) initially reported a decrease in psychomotor test scores among offspring born to women with FT4 indices in the lowest 10th percentile, despite having normal serum TSH concentrations. Similarly, Li et al. (199) observed a reduction in IQ among offspring of mothers suffering from either hypothyroidism or isolated hypothyroxinemia during the first trimester. In recent years, additional prospective, nonrandomized studies have similarly reported adverse child outcomes in children born to mothers with isolated hypothyroxinemia (19,192,193,259,279,280 –284,287 –289,291,294,295,306,313,314,318) (see Tables 5 and 6). Adverse outcomes include lower IQ, language delay, worsened motor function, smaller head circumference, and an increased risk of autism. These data are derived from different populations across the world (China, Belgium, the Netherlands, Spain) with known differences in iodine status. In contrast to those studies investigating the association of elevated TSH concentrations with adverse pregnancy outcome, however, very few studies have investigated isolated hypothyroxinemia and adverse pregnancy outcomes (see Table 7), excepting birth weight (314 –316) and premature delivery (19). Available data suggest an association with higher birth weight and higher risk of premature delivery. Interestingly, many large-scale studies demonstrate that the populations of women with elevated TSH concentrations are generally exclusive from those identified with low FT4 concentrations. For example, in the CATS study, approximately the same proportion of screened mothers were identified in the hypothyroxinemic and subclinically hypothyroid groups, with little overlap.

Overall, available evidence appears to show an association between hypothyroxinemia and cognitive development of the offspring, with uncertain effects on prematurity (314 –316) and low birth weight (19). However, no studies exist in which LT4 administration has been shown to ameliorate such harmful effects. In the CATS trial (303), 499 women were identified with low FT4 concentrations and 242 randomized to treatment with LT4. This intervention failed to demonstrate any improvement in neurocognitive outcomes in the offspring at 3 years of age. However, women in this study received a fairly high dose (150 μg) of LT4 treatment and 10% needed a dose reduction because of biochemical or clinical signs of overtreatment. This is interesting because a recent study by Korevaar et al. (317) showed that both low and high FT4 concentrations may be associated with a decrease in child IQ and reduction of cerebral gray matter volume as assessed by magnetic resonance imaging.

Question 36: Should women with overt hypothyroidism be treated in pregnancy?

As mentioned above, numerous retrospective and case–control studies confirm the detrimental effects of overt hypothyroidism on both pregnancy and fetal health (318). A recent retrospective study of more than 1000 pregnant women on chronic LT4 replacement, showed that the risk of pregnancy loss increased proportionally to the degree of TSH elevation, with no increased risk associated with TSH normalization (318). Although no prospective, randomized investigation of LT4 intervention to improve obstetric outcomes or child development has occurred in pregnant women with overt hypothyroidism, such an investigation would be unethical to complete. Nonetheless, available data confirm the benefits of treating severe hypothyroidism during pregnancy.

■ RECOMMENDATION 27

 Treatment of overt hypothyroidism is recommended during pregnancy.

 Strong recommendation, moderate-quality evidence.

Question 37: Should women with subclinical hypothyroidism be treated in pregnancy?

Many prospective and retrospective studies have demonstrated an increased risk of pregnancy complications associated with mildly elevated maternal TSH concentrations, especially in TPOAb-positive women. However, only a small number of studies have investigated the impact of LT4 treatment on pregnancy complications in such women.

As noted above, a single RCT has demonstrated a potential benefit of LT4 intervention at ∼9 weeks gestation. Importantly, this study documented a reduction in the adverse pregnancy composite outcome only in TPOAb-positive women with mild hypothyroidism (defined as a TSH >2.5 mU/L) (286). It should again be noted that the majority of women with subclinical hypothyroidism detected in this investigation were TPOAb negative, for whom no intervention or treatment was provided. This study also used a composite endpoint that included subjective endpoints such as cesarean section rates and postdelivery admission to the neonatal intensive care unit. Importantly, the authors' conclusion for their primary endpoint stated that universal screening for elevated TSH concentration in a broad population of pregnant women did not improve outcomes compared to a high risk screening strategy. Outcomes of universal screening compared to no screening were not assessed.

A separate RCT demonstrated a decrease in preterm delivery and pregnancy loss in euthyroid (defined as TSH <4.2 mU/L) TPOAb-positive women who were treated with LT4 beginning in the first trimester of pregnancy (28). However, the majority of pregnancy losses in the control group occurred before the average start of LT4 therapy.

Many studies have stratified the risk imparted by hypothyroidism according to TPOAb status and consistently show that this risk is higher in TPOAb-positive women (288). These data also suggest that the adverse impact associated with maternal TSH levels is apparent at lower TSH elevations in women known to be TPOAb positive compared to women who are TPOAb negative. Furthermore, two studies suggest a reduction in pregnancy loss when TPOAb-positive women are treated with LT4, even when biochemically euthyroid (28,180). Intervention trials have not been performed in TPOAb-negative women.

However, despite the limitations of available interventional trials of LT4 therapy in this subclinically hypothyroid group, the data taken in aggregate appear to suggest a benefit of treatment, especially as it applies to reducing miscarriage in TPOAb-positive women. Therefore, it seems reasonable to recommend or consider LT4 treatment for specific subgroups of pregnant women with subclinical hypothyroidism. The strength of such recommendations, however, should differ depending on TPOAb status, as will the strength of evidence supporting treatment for each subgroup. This recommendation also necessitates that any pregnant women with an elevated TSH concentration must also be evaluated for TPOAb status. In making the recommendation, the task force acknowledges the very low risk inherent in initiating low-dose LT4 treatment. A dose of only 50 μg/d is typically required for effective treatment of subclinically hypothyroid women.

■ RECOMMENDATION 28

 Pregnant women with TSH concentrations >2.5 mU/L should be evaluated for TPOAb status.

■ RECOMMENDATION 29

 Subclinical hypothyroidism in pregnancy should be approached as follows:

(a) LT4 therapy is recommended for

Question 38: Should women with isolated hypothyroxinemia be treated with LT4 in pregnancy?

Although several studies have reported adverse outcomes in children born to mothers with isolated hypothyroxinemia, no interventional data have yet been published that demonstrate beneficial effects of LT4 therapy. One observational study analyzing women at 12, 24, and 32 weeks of pregnancy demonstrated delayed infant neurodevelopment in women with persistent hypothyroxinemia. However, when FT4 concentrations increased during pregnancy, infant development was not improved (312). Nevertheless, at present there are only two randomized, prospective, intervention trials in which women with a low FT4 were treated with LT4, at 13 and 17 weeks gestation, respectively (21,306). Both investigations failed to show any beneficial effect on cognitive development following LT4 administration, though a major limitation of the studies was the late timing of the intervention, after completion of the first trimester. Nevertheless, given the existing interventional data, treatment of isolated hypothyroxinemia cannot be recommended at this time.

■ RECOMMENDATION 30

 Isolated hypothyroxinemia should not be routinely treated in pregnancy.

 Weak recommendation, low-quality evidence.

Question 39: What is the optimal method of treating hypothyroidism in pregnant women?

Delivery of T4 is crucial for the developing fetal brain (313). The ratio of T4 to T3 in desiccated thyroid preparations is 4.2:1, which is significantly lower than the 14:1 ratio of secretion by the human thyroid gland. This relative excess of T3 leads to supraphysiologic maternal levels of T3 and relatively low levels of T4 (319). Patients using either desiccated thyroid or a treatment regimen combining T3 and T4 (320) are likely at risk for having insufficient transfer of maternal T4 to the fetal brain. It is notable that the majority of fetal T3 present in the CNS during pregnancy is derived from maternal T4 actively transported into this space. The fetal CNS is relatively impermeable to T3, which therefore argues against use of exogenous T3 during pregnancy. For these reasons, the task force feels that any T3-containing preparation should be avoided for the treatment of maternal hypothyroidism during pregnancy.

■ RECOMMENDATION 31

 The recommended treatment of maternal hypothyroidism is administration of oral LT4. Other thyroid preparations such as T3 or desiccated thyroid should not be used in pregnancy.

 Strong recommendation, low-quality evidence.

Question 40: What is the biochemical goal when treating hypothyroidism in pregnant women?

■ RECOMMENDATION 32

 In parallel to the treatment of hypothyroidism in a general population, it is reasonable to target a TSH in the lower half of the trimester-specific reference range. When this is not available, it is reasonable to target maternal TSH concentrations below 2.5 mU/L.

 Weak recommendation, moderate-quality evidence.

Question 41: Are there circumstances in which euthyroid women are at risk for hypothyroidism once pregnant?

In 1994, Glinoer et al. (147) performed a prospective study in 87 TPOAb-positive euthyroid women evaluated before and during early pregnancy. Twenty percent of women in the study developed a TSH >4 mU/L during gestation despite a normal TSH and no requirement for LT4 prenatally. This occurred despite the expected decrease in TPOAb titers during pregnancy (147). Negro et al. (28) demonstrated similar results in a prospective study. The authors found that in TPOAb-positive euthyroid women, TSH levels increased progressively as gestation progressed, from a mean of 1.7 mU/L (12th week) to 3.5 mU/L (term), with 19% of women having a supranormal TSH concentration at delivery. These findings confirm that an increased requirement for thyroid hormone occurs during gestation. In women who are TPOAb positive, both overt and subclinical hypothyroidism may occur because of a lack of ability of the thyroid to augment production when needed during pregnancy. Similarly, patients who undergo hemithyroidectomy or receive radioactive iodine and are euthyroid before pregnancy are at risk for developing elevated serum TSH levels during gestation.

In summary, euthyroid patients who are antithyroid Ab positive, post-hemithyroidectomy, or treated with radioactive iodine have an increased propensity for the development of hypothyroidism in gestation and should be monitored regularly.

Question 42: How should women with hypothyroidism or at risk for hypothyroidism be monitored through pregnancy?

In women at risk for hypothyroidism (TPOAb- or TgAb-positive, post-hemithyroidectomy, and/or post-radioactive iodine) increased surveillance is recommended. Based on findings extrapolated from investigations of treated hypothyroid women from early pregnancy onwards (321), it is reasonable to evaluate these women for TSH elevation approximately every 4 weeks during pregnancy. Serial testing is preferably continued through midpregnancy because the increased T4 demand continues throughout the first half of gestation.

A study by Yassa and colleagues (322) investigated the optimal timing of subsequent assessment of thyroid function following dose modification though this was in patients consuming LT4 prenatally and may not be generalizable to patients not taking LT4 but at risk for hypothyroidism.

■ RECOMMENDATION 33

 Women with overt and subclinical hypothyroidism (treated or untreated) or those at risk for hypothyroidism (e.g., patients who are euthyroid but TPOAb or TgAb positive, post-hemithyroidectomy, or treated with radioactive iodine) should be monitored with a serum TSH measurement approximately every 4 weeks until midgestation and at least once near 30 weeks gestation.

 Strong recommendation, high-quality evidence.

Question 43: How do treated hypothyroid women (receiving LT4) differ from other patients during pregnancy? What changes can be anticipated in such patients during gestation?

The major physiologic thyroid changes during pregnancy have been thoroughly studied. Total body T4 requirements are not static throughout gestation. Rather, data demonstrate that the total body T4 pool must increase by approximately 40%–50% to maintain a euthyroid state (321,323). In a healthy woman who becomes pregnant, the pregnancy hormone hCG plays a major role as a stimulus of maternal thyroid hormone production, especially throughout the first half of pregnancy. Together with pituitary TSH, placental hCG stimulates endogenous thyroid hormone production when an intact thyroid is present, and it helps to maintain a euthyroid state during gestation.

In women with known hypothyroidism, serum hCG and TSH cannot stimulate adequate T4 production. If exogenous LT4 is not adjusted, the increased demand of pregnancy will outstrip supply, and maternal hypothyroidism will occur. Clinical studies have confirmed that the increased requirement for thyroxine (or exogenous LT4) occurs as early as 4–6 weeks of pregnancy (321). Such requirements gradually increase through 16–20 weeks of pregnancy and plateau thereafter until the time of delivery. These data provide the basis for recommending adjustments of LT4 dosage when affected women become pregnant and also for the timing of follow-up intervals for TSH in treated patients.

Question 44: What proportion of treated hypothyroid women (receiving LT4 prenatally) require changes in their LT4 dose during pregnancy?

Between 50% and 85% of LT4-treated hypothyroid women need to increase exogenous LT4 dosing during pregnancy (277,321,323). The incremental increase largely depends on the underlying etiology of the hypothyroidism. There is a greater likelihood that dose increases will be required in those patients without functional thyroid tissue (e.g., due to radioablation or surgery) in comparison with patients with Hashimoto's thyroiditis (324,325). The preconception level of TSH as well as other factors can also influence the rapidity and extent of LT4 augmentation necessary to maintain a euthyroid state during pregnancy. For example, variation and changes in maternal estrogen levels during pregnancy correlate with variations in the gestational requirements for LT4 (321).

The LT4 adjustment should be made as soon as possible after pregnancy is confirmed to reduce the probability of hypothyroidism. Normalization of TSH concentrations throughout gestation is the goal. For women receiving LT4 preconception, a prospective, randomized study has provided evidence that supports a single dose-adjustment strategy rather than a stepwise approach for LT4 dosage adjustment postconception (322). For euthyroid women receiving once-daily dosing of LT4 (regardless of the dosage), a recommendation to increase by two additional tablets weekly (nine tablets per week instead of seven tablets per week, giving a 29% increase) can effectively mimic gestational physiology and thus prevent maternal hypothyroidism during the first trimester (322). Another option is to increase the dosage of daily LT4 by approximately 25%–30%. Dosage augmentation should occur as soon as possible when a missed menstruation or suspected pregnancy occurs, and this should be discussed with every patient in the prepregnancy setting. Confirmatory biochemical testing should also occur simultaneously.

Question 45: How should preconception LT4 be adjusted in treated hypothyroid women (receiving LT4) planning pregnancy?

The difficulties inherent to achieving rapid, postconceptional TSH normalization have also focused attention upon preconception TSH modulation. Different cutoff values for preconception TSH, ranging from <1.2 to <2.5 mU/L have been advocated. In one study, only 17% of women with TSH <1.2 mU/L had to increase LT4 dose later during pregnancy (326). Given this, it is recommended that all treated hypothyroid women (currently receiving LT4) optimize thyroid parameters preconception. A maternal serum TSH concentration <2.5 mU/L is a reasonable goal for such women. Even lower preconception TSH values (<1.5 mU/L) could reduce the risk of TSH elevation during the first trimester (322), but a lower treatment target may not improve outcomes because the LT4 dose can be immediately increased upon a positive pregnancy test. Furthermore, the process of achieving a TSH concentration at the lower end of the reference range could induce subnormal TSH concentrations in some patients. Though generally safe for any developing fetus, potential such effects upon conception and/or successful implantation are unknown.

■ RECOMMENDATION 34

 Treated hypothyroid women of reproductive age should be counseled regarding the likelihood of increased demand for LT4 during pregnancy. Such women should also be counseled to contact their caregiver immediately upon a confirmed or suspected pregnancy.

 Strong recommendation, high-quality evidence.

■ RECOMMENDATION 35

 In hypothyroid women treated with LT4 who are planning pregnancy, serum TSH should be evaluated preconception, and LT4 dose adjusted to achieve a TSH value between the lower reference limit and 2.5 mU/L.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 36

 Hypothyroid patients receiving LT4 treatment with a suspected or confirmed pregnancy (e.g., positive home pregnancy test) should independently increase their dose of LT4 by ∼20%–30% and urgently notify their caregiver for prompt testing and further evaluation. One means of accomplishing this is to administer two additional tablets weekly of the patient's current daily LT4 dosage.

 Strong recommendation, high-quality evidence.

Question 46: How should LT4 be adjusted postpartum?

The increased LT4 dose requirements during gestation are a function of pregnancy itself. Therefore, following delivery, maternal LT4 dosing should be reduced to prepregnancy levels, and a serum TSH assessed 6 weeks thereafter. However, a study demonstrated that more than 50% of women with Hashimoto's thyroiditis required an increase in the pregestational thyroid hormone dose in the postpartum period, presumably due to an exacerbation of autoimmune thyroid dysfunction post partum (327). In women started on LT4 during pregnancy for thyroid autoimmunity in the absence of TSH elevation, the LT4 can be stopped at delivery, with serum TSH assessment at 6 weeks post partum.

■ RECOMMENDATION 37

 Following delivery, LT4 should be reduced to the patient's preconception dose. Additional thyroid function testing should be performed at approximately 6 weeks post partum.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 38

 Some women in whom LT4 is initiated during pregnancy may not require LT4 post partum. Such women are candidates for discontinuing LT4, especially when the LT4 dose is ≤50 μg/d. The decision to discontinue LT4, if desired, should be made by the patient and their caregiver. If LT4 is discontinued, serum TSH should be evaluated in approximately 6 weeks.

 Weak recommendation, moderate-quality evidence.

Question 47: What is the outcome and long-term prognosis when maternal hypothyroidism is effectively treated through gestation?

Although untreated (or incompletely treated) hypothyroidism can adversely affect pregnancy, no data suggest that women with adequately treated subclinical or overt hypothyroidism have an increased risk of any obstetrical complication. Consequently, there is no indication for any additional obstetric testing or surveillance in pregnancies of women with either subclinical or overt hypothyroidism who are being monitored and treated appropriately.

Question 48: Except for measurement of maternal thyroid function, should additional maternal or fetal testing occur in treated hypothyroid women during pregnancy?

■ RECOMMENDATION 39

 In the care of women with adequately treated hypothyroidism, no other maternal or fetal testing (such as serial fetal ultrasounds, antenatal testing, and/or umbilical blood sampling) is recommended beyond measurement of maternal thyroid function unless needed due to other circumstances of pregnancy. An exception to this is women with GD effectively treated with ¹³¹I ablation or surgical resection, who require TSH receptor antibody (TRAb) monitoring.

 Strong recommendation, moderate-quality evidence.

Question 49: What are the causes of thyrotoxicosis in pregnancy?

Thyrotoxicosis is the clinical syndrome of hypermetabolism and hyperactivity that results when a person is exposed to supraphysiological amounts of thyroid hormones. The most common cause of thyrotoxicosis is hyperfunction of the thyroid gland (hyperthyroidism), and the most common cause of hyperthyroidism in women of childbearing age is autoimmune GD occurring before pregnancy in 0.4%–1.0 % of women and in approximately 0.2% during pregnancy (328).

Less common non-autoimmune causes of hyperthyroidism in pregnancy include toxic multinodular goiter and toxic adenoma. Subacute painful or painless thyroiditis with passive release of thyroid hormones from a damaged thyroid gland are less common causes of thyrotoxicosis in pregnancy, and a number of other conditions such as a TSH-secreting pituitary adenoma (329), struma ovarii (330), functional thyroid cancer metastases, or germline TSH receptor mutations (331) are very rare. A special cause of thyrotoxicosis is overtreatment with or factitious intake of thyroid hormone.

More frequent than GD as the cause of thyroid function tests demonstrating hyperthyroxinemia is “gestational transient thyrotoxicosis,” which is limited to the first half of pregnancy. This condition, characterized by elevated FT4 and suppressed serum TSH, is diagnosed in about 1%–3% of pregnancies. This frequency depends on the geographic area and is secondary to elevated hCG levels (332,333). Often it is associated with hyperemesis gravidarum, defined as severe nausea and vomiting in early pregnancy with more than 5% weight loss, dehydration, and ketonuria. Hyperemesis gravidarum occurs in 3–10 per 1000 pregnancies (334,335). Other conditions associated with hCG-induced thyrotoxicosis include multiple gestation, hydatidiform mole, and choriocarcinoma (336,337). Most cases present with marked elevations of serum hCG (14). A TSH receptor mutation leading to functional hypersensitivity to hCG also has been recognized as a rare cause of pregnancy-associated hyperthyroidism (338).

Question 50: What is the appropriate initial evaluation of a suppressed serum TSH concentration during the first trimester of pregnancy?

Serum TSH may decrease in the first trimester of normal pregnancy as a physiological response to the stimulating effect of hCG upon the TSH receptor. A peak hCG level typically occurs between 7 and 11 weeks gestation (339). In particular, a serum TSH below 0.1 mU/L (in some cases even undetectable) may be present in approximately 5% of women by week 11 of pregnancy (270). Any subnormal serum TSH value should be evaluated in conjunction with serum TT4 (or FT4) and T3 values. The biochemical diagnosis of overt hyperthyroidism is confirmed in the presence of a suppressed or undetectable serum TSH and inappropriately elevated serum TT4/FT4, or T3 (Section III).

Question 51: How can gestational transient thyrotoxicosis be differentiated from Graves' hyperthyroidism in pregnancy?

Diagnosing the cause of the disease is essential in any patient with thyrotoxicosis. In early pregnancy, the differential diagnosis in the majority of cases is between Graves' hyperthyroidism and gestational transient thyrotoxicosis (332,333). In both situations, common clinical manifestations include palpitations, anxiety, tremor, and heat intolerance. A careful history and physical examination is of utmost importance in establishing the etiology. The findings of no prior history of thyroid disease, no stigmata of GD (goiter, orbitopathy), a self-limited mild disorder, and symptoms of emesis favor the diagnosis of gestational transient thyrotoxicosis.

If other causes for thyrotoxicosis are suspected, measurement of TRAb is indicated. If this is negative or thyroid nodules are suspected based on clinical examination, a thyroid ultrasound should be performed to evaluate nodularity. Serum hCG is higher on average in gestational transient thyrotoxicosis than in patients with GD, but overlap is considerable and the clinical usefulness of such measurement is limited (340). No study has demonstrated usefulness of thyroid ultrasonography for differentiating between gestational transient thyrotoxicosis and GD. In the presence of a nodular goiter, a serum TT3 determination is helpful in assessing the possibility of the “T3 toxicosis” syndrome. TT3 determination may also be of benefit in diagnosing T3 thyrotoxicosis caused by GD. In general, serum T3 tends to be disproportionately elevated more than T4 in cases of thyrotoxicosis caused by direct thyroid hyperactivity. In comparison, T4 tends to be disproportionately elevated beyond T3 when thyrotoxicosis is caused by destructive processes such as thyroiditis (341).

■ RECOMMENDATION 40

 When a suppressed serum TSH is detected in the first trimester (TSH less than the reference range), a medical history, physical examination, and measurement of maternal serum FT4 or TT4 concentrations should be performed. Measurement of TRAb and maternal TT3 may prove helpful in clarifying the etiology of thyrotoxicosis.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 41

 Radionuclide scintigraphy or radioiodine uptake determination should not be performed in pregnancy.

 Strong recommendation, high-quality evidence.

Question 52: What is the appropriate management of gestational transient thyrotoxicosis?

The management of women with gestational transient thyrotoxicosis depends on the severity of symptoms. In women with hyperemesis gravidarum, control of vomiting and treatment of dehydration with intravenous fluids is the customary treatment. Women with severe hyperemesis gravidarum need frequent medical visits for management of dehydration and electrolyte abnormalities. In some cases, hospitalization is required. Antithyroid drugs (ATDs) are not indicated because the serum T4 returns to normal by 14–18 weeks gestation and ATD use in early pregnancy increases risk of birth defects. Importantly, obstetrical outcome was not improved in isolated cases in which gestational transient thyrotoxicosis was treated with ATDs (342). However, there are no studies reported in the literature comparing ATD therapy versus supportive therapy. In situations in which symptomatic therapy is indicated, small amounts of β-blockers given over a limited time period may be useful, and close follow-up with repeat investigation for the cause of disease should be performed.

■ RECOMMENDATION 42

 The appropriate management of abnormal maternal thyroid tests attributable to gestational transient thyrotoxicosis and/or hyperemesis gravidarum includes supportive therapy, management of dehydration, and hospitalization if needed. ATDs are not recommended, though β-blockers may be considered.

 Strong recommendation, moderate-quality evidence.

Question 53: How should women with GD seeking future pregnancy be counseled?

The planning of therapy in relation to possible future pregnancy should be discussed with all women of childbearing age who develop thyrotoxicosis. In general, pregnancy should be postponed until a stable euthyroid state is reached. As a guide, two sets of thyroid function test within the reference range, at least 1 month apart, and with no change in therapy between tests, can be used to define a stable euthyroid state. The use of contraception until the disease is controlled is strongly recommended.

A hyperthyroid patient who desires future pregnancy may be offered ablative therapy using ¹³¹I, thyroid surgery, or medical therapy. Each therapeutic option carries advantages and disadvantages, as detailed below (see Table 8). This is further discussed in the ATA guidelines for the management of thyrotoxicosis (343).

Question 54: What is the management of patients with Graves' hyperthyroidism during pregnancy?

Several studies have shown that obstetric and medical complications are directly related to control of maternal hyperthyroidism, and the duration of the euthyroid state throughout pregnancy (339,342,344,349). Poor control of thyrotoxicosis is associated with pregnancy loss, pregnancy-induced hypertension, prematurity, low birth weight, intrauterine growth restriction, stillbirth, thyroid storm, and maternal congestive heart failure (350). Moreover, some studies suggest fetal exposure to excessive levels of maternal thyroid hormone may program the offspring to develop diseases such as seizure disorders and neurobehavioral disorders in later life (351).

Thionamide ATDs (MMI, carbimazole [CM], and PTU) are the mainstays of treatment for hyperthyroidism during pregnancy (352,353). They reduce iodine organification and coupling of monoiodotyrosine and diiodotyrosine, therefore inhibiting thyroid hormone synthesis. Because the block is not absolute and the thyroid contains a depot of thyroid hormone bound to Tg, the normalization of thyroid function tests takes place gradually over weeks. The initial dose of ATD depends on the severity of the symptoms and the degree of hyperthyroxinemia. In general, initial doses of ATDs during pregnancy are as follows: MMI, 5–30 mg/d (typical dose in average patient 10–20 mg); CM, 10–40 mg/d; and PTU, 100–600 mg/d (typical PTU dose in average patient 200–400 mg/d). The equivalent potency of MMI to PTU is approximately 1:20 (e.g., 5 mg MMI = 100 mg of PTU) (354 –356). Ten milligrams of CM is rapidly metabolized to approximately 6 mg of MMI (328). Because the half-life of PTU is shorter than that of MMI, PTU dosing should generally be split into two or three daily doses. In comparison, MMI can generally be given in one daily dose. In rare cases of severe hyperthyroidism, twice or three-times daily dosing may be of benefit (357,358).

Importantly, side effects occur in 3%–5% of patients taking thionamide drugs, the majority of which are allergic reactions such as skin rash (352), whereas the severe side effects of agranulocytosis (0.15%) (359,360) and liver failure (<0.1%) (92) are rare. Most side effects develop within the first months following initiation (359) or re-initiation (361) of therapy. In 2010 the U.S. Food and Drug Administration (FDA) called attention to the risk of hepatotoxicity in patients exposed to PTU because PTU had been found to be third on the list of drugs leading to liver transplantation in the United States (362,363). An advisory committee recommended limiting the use of PTU to the first trimester of pregnancy (364). Exceptions to this are patients with MMI allergy or those with thyroid storm. Monitoring hepatic enzymes during administration of PTU may be considered. However, no prospective data exist that have demonstrated that the monitoring of liver enzymes is effective in preventing fulminant PTU-induced hepatotoxicity.

However, the greatest risk surrounding the use of ATDs in pregnancy is related to their potential teratogenic effects (365). As early as 1972, exposure to MMI had been associated with aplasia cutis (366). Subsequently, several other types of congenital malformations have been associated with MMI use. A “syndrome of methimazole/carbimazole embryopathy” was described, which also includes dysmorphic facies (367, 368). Apart from aplasia cutis, defects with a statistically significant association with the use of MMI include choanal or esophageal atresia; various types of abdominal wall defects including umbilicocele; and eye, urinary system, and ventricular septal defects (369,370,371). Recent studies have shown that these complications are more common than previously thought, and they affect 2%–4% of children who have been exposed to MMI in early pregnancy, especially during gestational weeks 6–10 (369,370,372).

PTU was previously considered a safe medication for use during gestation (364,367). Recently, however, a Danish study revealed that 2%–3% of children exposed to PTU developed birth defects associated with this therapy (370). The defects were primarily face and neck cysts (often considered to be minor birth defects) and urinary tract abnormalities (in males). Prior to the recent investigation, such abnormalities were not commonly associated with PTU exposure, likely because they were diagnosed later in life when complications ensued. Importantly, however, most affected patients received surgery for the abnormality (373). Thus, PTU-associated birth defects appear less severe than MMI-associated birth defects but occur with similar incidence.

Beta-adrenergic blocking agents, such as propranolol 10–40 mg every 6–8 hours may be used for controlling hypermetabolic symptoms until patients have become euthyroid on ATD therapy. The dose should be reduced as clinically indicated. In the vast majority of cases the drug can be discontinued in 2–6 weeks. Long-term treatment with β-blockers has been associated with intrauterine growth restriction, fetal bradycardia, and neonatal hypoglycemia (374). One study suggested a higher rate of spontaneous pregnancy loss when both drugs were taken together, as compared with patients receiving only MMI (375). However, it was not clear that this difference was attributable to the medication as opposed to the underlying condition. Beta-blocking drugs may be used as preparation for thyroidectomy.

Question 55: Should antithyroid medication be withdrawn or modified in early pregnancy?

Many patients receiving ATD therapy for GD gradually enter remission of the autoimmune abnormality when made euthyroid. Though approximately half of patients eventually experience a relapse of hyperthyroidism when the antithyroid medication is withdrawn after 1–2 years of therapy, only a small fraction of patients who have become TRAb negative during therapy will become hyperthyroid within the first months. In support of this, in a prospective Norwegian study of 218 GD patients treated with ATD for 12 months, only 5% of TRAb-negative patients became hyperthyroid within 8 weeks after ATD withdrawal (376).

Thus, one option when pregnancy is diagnosed in a woman receiving ATD therapy for GD and who, based on clinical and biochemical findings appears to be in remission, is to withdraw ATD medication and perform repeated thyroid function testing during the first trimester of pregnancy. If ATD therapy is needed during the first trimester, PTU is preferred over MMI because the risk for severe birth defects is lower. Cessation of medication has to be recommended early in gestation, before the major teratogenic periods (gestational weeks 6–10) (372,373,376,377).

Gestational weeks are calculated from the first day of the last normal menstrual period; it is typically at week 5 when the next normal menstruation does not appear in a pregnant woman. Sensitive pregnancy tests are widely available and should detect pregnancy by this time. Optimally, women receiving ATD should test for pregnancy within the first days of missing or unusually weak menstruation. If the test is positive, the woman should contact the caregiver responsible for thyroid therapy to receive instruction regarding withdrawing or modifying ATD therapy and to discuss thyroid function testing.

The risk of rapid relapse of hyperthyroidism after medication withdrawal in early pregnancy varies among patients. The risk is high in patients who have been treated for a short period (<6 months), who have suppressed or low serum TSH while on medication prepregnancy, who require >5–10 mg of MMI per day to stay euthyroid, who have active orbitopathy or large goiter, and those who have high levels of TRAb (378). Because this assessment is based upon clinical judgment, no specific single parameter may be used to assign risk. However, if the risk is considered high, medication should not be withdrawn, and PTU should be administered as the drug of choice.

Even if conventional ATDs are effective in achieving a euthyroid state, the risk of birth defects has raised the question of alternative types of drug therapy for hyperthyroidism. Iodine in pharmacological doses was widely used to treat hyperthyroidism before the thionamide drugs (and radioiodine) became available. It was (and is) especially effective as a preparation for subsequent thyroid surgery, as introduced by Plummer in 1923 (379). However, escape from the effect during prolonged therapy was not uncommon, especially in patients with severe hyperthyroidism (380), and iodine therapy is now mostly used to reduce thyroid blood flow before surgery and as part of the combination of therapies given to patients with thyrotoxic crises.

An exception to this narrow indication for the treatment of hyperthyroidism is Japan, where there is considerable experience with high-dose iodine therapy in pregnancy (381,382). The doses of iodine used have varied between 5 and 75 mg/d, adjusted according to thyroid function tests. In a recent retrospective nonrandomized study, substituting MMI with potassium iodine in early pregnancy reduced the risk of birth defects. However, a proportion of such pregnant women became hyperthyroid following the change in treatment (382). Important to note, Japan is a high iodine intake country, and the Japanese results may not be reproducible in other countries. Cases of iodine-induced congenital hypothyroidism have been reported in children of U.S. women treated with high-dose iodine during pregnancy (383), and the use of iodine-containing disinfectants during labor is a well-established cause of elevated TSH during screening for congenital hypothyroidism (384). Thus, at present, such therapy cannot be recommended outside Japan until more evidence on safety and efficacy is available.

Theoretically, other possible treatments can also be considered. Cholestyramine is not absorbed from the gut (and thus not transferred to the fetus) and may moderately reduce circulating thyroid hormones by binding the hormones during their enterohepatic circulation (385 –387). Cholestyramine has been used in pregnancy to treat obstetric cholestasis (388). Gastrointestinal discomfort with nausea, vomiting, and diarrhea may be a problem (388), and there is a risk of binding in the gut of vitamins and other substances important in pregnancy. Perchlorate competitively inhibits the active transport of iodide into the thyroid (and into breast milk), and it has been used to treat hyperthyroidism (389,390). Limited evidence suggests it has no teratogenic effects as long as normal maternal thyroid function is maintained (391). In summary, more data are needed before the use of any such drugs can be recommended. Lithium has thyroid-inhibiting effects somewhat similar to pharmacological doses of iodine (392), but it may be teratogenic (393) and therefore should not be used to treat hyperthyroidism in pregnancy.

■ RECOMMENDATION 45

 Women taking MMI or PTU should be instructed to confirm potential pregnancy as soon as possible. If the pregnancy test is positive, pregnant women should contact their caregiver immediately.

 Strong recommendation, high-quality evidence.

■ RECOMMENDATION 46

■ RECOMMENDATION 47

 In pregnant women with a high risk of developing thyrotoxicosis if antithyroid drugs were to be discontinued, continued antithyroid medication may be necessary. Factors predicting high clinical risk include being currently hyperthyroid, or requirement of >5–10 mg/d MMI or >100–200 mg/d PTU to maintain a euthyroid state. In such cases,

Question 56: What are the principles of thyroid testing and ATD administration when treating Graves' hyperthyroidism during pregnancy?

Thyroid stimulating antibodies, ATDs, and most maternal thyroid hormones can effectively cross the placenta barrier. When the fetal thyroid is functional, it can respond to TRAb antibodies, causing excess fetal production of thyroid hormone. Furthermore, if the mother has an intact thyroid and is hyperthyroid from GD, the fetus will also be exposed to the hyperthyroxinemia produced by the mother's thyroid during gestation. Typically, fetal hyperthyroidism due to cross-placental passage of TRAb develops at or after week 20 of pregnancy. However, one case has been published in which excessive maternal TRAb production led to fetal hyperthyroidism by week 18 (394).

As mentioned, MMI, PTU, and CM also effectively cross the placenta, and therefore ATD therapy for maternal hyperthyroidism also modulates fetal thyroid function. Importantly, all ATDs tend to be more potent in the fetus than in the mother. Thus, when the mother is made euthyroid, the fetus is often overtreated (395). Therefore, in order to avoid a deleterious fetal impact, the aim of treatment is to maintain maternal TT4/FT4 values at, or just above the pregnancy-specific upper limit of normal. As a general rule, the smallest possible dose of ATDs should be used whenever possible.

In the setting of hyperthyroidism during pregnancy, maternal TT4/FT4 and TSH (and in cases of severe hyperthyroidism, also serum T3) should be measured approximately every 2–4 weeks following initiation of therapy, and every 4–6 weeks after achieving the target value (396 –398). When trimester-specific FT4 values are not available, use of the reference range for nonpregnant patients is recommended. Separately, a TT4 measurement with reference value 1.5 times the nonpregnancy range may be used in second and third trimesters. Overtreatment should be avoided because of the possibility of inducing fetal goiter and or fetal hypothyroidism (399).

In the first trimester of pregnancy some women with GD experience an exacerbation of symptoms (400), which is parallel to the moderate increase in incidence of GD in early pregnancy (401). By the third trimester the incidence of GD becomes very low (401) corresponding to the general decrease in thyroid autoimmunity with a decrease in TRAb.

Discontinuation of all ATD therapy is feasible in 20%–30% of patients in the last trimester of gestation (402). Maternal serum TSH well within the reference range is a sign that the ATD dose has to be reduced to avoid fetal overtreatment. If this is not done, fetal hypothyroidism and goiter may develop from overtreatment with ATDs. Disappearance of maternal TRAb in late pregnancy indicates a high likelihood of successful ATD withdrawal.

■ RECOMMENDATION 48

Because the fetal thyroid responds more strongly to ATD therapy than the maternal thyroid, mothers on an ATD in the second half of pregnancy, who by nonpregnancy standards would be considered euthyroid, should have the ATD dose reduced to protect the fetus.

In occasional patients, GD stays very active and serum TT3 may remain elevated even if TT4/FT4 becomes normal or even low (403). An increase in ATD dose to normalize maternal serum TT3 will cause elevated serum TSH in the infants at birth (398), and a balance in ATD dosing with careful clinical evaluation of the fetus and the mother is needed.

Worsening of disease activity with a need for an increase in ATD dose or relapse of previously remitted disease often occurs after delivery (400,404). Women should be informed about this risk, and appropriate monitoring performed. A single Japanese study has suggested that relapse may be prevented by low-dose ATD during the postpartum period (405), but more studies on this are needed.

The combination of LT4 and an ATD (block-replace therapy) has in general been shown not to improve GD remission rates (406), and it results in a larger dose of ATD required in order to maintain the FT4 within the target range. The placenta is readily permeable to the ATD but not to the LT4 given to the mother, and the fetal thyroid is relatively more sensitive to the effect of ATDs than the maternal thyroid. Therefore, block-replacement therapy given to the mother in the second half of pregnancy will generally lead to fetal goiter and hypothyroidism (403). The only indication for such combination therapy during pregnancy is in the treatment of isolated fetal hyperthyroidism caused by maternal TRAb production in a mother who previously received ablative therapy for GD (407). The ATD will pass the placenta and treat the fetal hyperthyroidism, whereas the LT4 is necessary to keep the mother euthyroid.

■ RECOMMENDATION 49

 A combination regimen of LT4 and an ATD should not be used in pregnancy, except in the rare situation of isolated fetal hyperthyroidism.

 Strong recommendation, high-quality evidence.

Question 57: What are the indications and timing for thyroidectomy in the management of GD during pregnancy?

Thyroidectomy should be considered in cases of allergies/contraindications to both ATDs, in the patient who is not compliant with drug therapy, and in women in whom euthyroidism cannot be achieved even on large doses of ATDs. If surgery is indicated, the second trimester is the optimal time. Thyroidectomy is often followed by a gradual, but not immediate disappearance of TRAb, and withdrawal of ATD in the mother after thyroidectomy may lead to isolated fetal hyperthyroidism (403). High serum TRAb values before surgery indicate a risk for isolated fetal hyperthyroidism, and after maternal ATD withdrawal a program of careful fetal monitoring and possible therapy should be planned (403). In hyperthyroid patients not tolerating or nonresponsive to ATD therapy, preparation for surgery with β-blocking agents and a short course of potassium iodide solution (50–100 mg/d) are recommended (381). Potassium iodide preparation before surgery is also recommended in patients with other signs of active disease to reduce bleeding during surgery.

■ RECOMMENDATION 50

 Thyroidectomy in pregnancy may be indicated for unique scenarios. If required, the optimal time for thyroidectomy is in the second trimester of pregnancy. If maternal TRAb concentration is high (>3 times the upper reference for the assay) the fetus should be carefully monitored for development of fetal hyperthyroidism throughout pregnancy, even if the mother is euthyroid post thyroidectomy.

 Strong recommendation, high-quality evidence.

Question 58: How should pregnant patients with GD be prepared for urgent nonthyroid surgery?

■ RECOMMENDATION 51

 We concur with the American College of Obstetricians and Gynecologists' Committee on Obstetric Practice consensus guidelines (written in 2011 and revised in 2015) (408) which state the following: “1) A pregnant woman should never be denied indicated surgery, regardless of trimester. 2) Elective surgery should be postponed until after delivery. 3) If possible, nonurgent surgery should be performed in the second trimester when preterm contractions and spontaneous abortion are least likely.”

 In the setting of a patient with GD undergoing urgent, nonthyroid surgery, if the patient is well controlled on ATD, no other preparation is needed. Beta-blockade should also be utilized if needed.

 Strong recommendation, moderate-quality evidence.

Question 59: What is the value of TRAb measurement in the evaluation of a pregnant woman with Graves' hyperthyroidism?

Fetal risks in women with previous or current Graves' hyperthyroidism include (a) fetal hyperthyroidism, (b) neonatal hyperthyroidism, (c) fetal hypothyroidism, (d) neonatal hypothyroidism, and (e) central hypothyroidism. The above potential complications depend on several factors: (a) poor control of hyperthyroidism throughout pregnancy may induce transient central hypothyroidism (409,410); (b) excessive amounts of ATDs may be responsible for fetal and neonatal hypothyroidism, even if the mother is biochemically euthyroid (399,403); and (c) high levels of thyroid stimulating antibodies in the second half of pregnancy may induce fetal and neonatal hyperthyroidism (411,412,413,414). TRAb is measurable in around 95% of patients with active Graves' hyperthyroidism, and levels may remain high following ablation therapy, even more so after radioiodine treatment than surgical removal (415).

Indications for ordering a TRAb test in pregnant women with GD include (a) mothers with untreated or ATD-treated hyperthyroidism in pregnancy, (b) a previous history of GD with past treatment with radioiodine or total thyroidectomy, (c) a previous history of delivering an infant with hyperthyroidism, or (d) a known history of thyroidectomy for the treatment of hyperthyroidism in pregnancy (416). In the majority of patients, maternal TRAb concentrations decrease with the progression of pregnancy; however, as in nonpregnant patients, the course of GD is variable.

The incidence of fetal and neonatal hyperthyroidism is between 1% and 5% in all women with active or a past history of Graves' hyperthyroidism and is associated with increased fetal/neonatal morbidity and mortality if unrecognized and untreated (417). In a recent French study (418), follow-up of 47 newborns of mothers harboring measurable levels of TRAb in pregnancy showed that nine developed transient neonatal hyperthyroidism and five of these (9% overall) had to be treated with ATD for a median duration of 60 days. A maternal TRAb serum concentration >5 IU/L (approximately 3 times the upper limit of normal for the assay) in the second and third trimester predicted neonatal hyperthyroidism with 100% sensitivity and 43% specificity. A similar TRAb level risk cutoff was found in a separate French study (419).

A determination of serum TRAb in early pregnancy is helpful in detecting pregnancies at risk. A value >5 IU/L or 3 times the upper limit of normal in a mother who previously received ablative therapy for GD is an indication for establishing close follow-up of the fetus in collaboration with a maternal–fetal medicine physician. Cases of overlooked isolated fetal hyperthyroidism leading to fetal loss in such women have repeatedly been published.

A determination of serum TRAb in late pregnancy in a mother who is still in need of ATD therapy to remain euthyroid, is helpful for detecting risk for neonatal hyperthyroidism. A value >5 IU/L or 3 times the upper limit of normal in the mother indicates that the fetal thyroid may be strongly stimulated by TRAb passing through the placenta. After birth, any ATD from the mother is much more rapidly cleared in the neonate than are the TRAb, and the neonate may become hyperthyroid.

If TRAb becomes undetectable in a pregnant woman taking ATD, it may be feasible to reduce or withdraw the ATD to protect the fetus against hypothyroidism and goiter.

In women who are in remission and euthyroid after a previous course of ATD therapy for GD, measurement of TRAb in pregnancy is not required (416).

■ RECOMMENDATION 52

Question 60: Under what circumstances should additional fetal ultrasound monitoring for growth, heart rate, and goiter be performed in women with Graves' hyperthyroidism in pregnancy?

Serial ultrasound examinations may be performed for the assessment of gestational age, fetal viability, amniotic fluid volume, fetal anatomy, and detection of malformations. Fetal well-being may be compromised in the presence of elevated TRAb, uncontrolled hyperthyroidism, and pre-eclampsia (349,420 –422). Signs of potential fetal hyperthyroidism that may be detected by ultrasonography include fetal tachycardia (heart rate >170 bpm, persistent for over 10 minutes), intrauterine growth restriction, presence of fetal goiter (the earliest sonographic sign of fetal thyroid dysfunction), accelerated bone maturation, signs of congestive heart failure, and fetal hydrops (411,421 –423). A team approach to the management of these patients is required, including an experienced obstetrician or maternal–fetal medicine specialist, neonatologist, and anesthesiologist. In most cases, the diagnosis of fetal hyperthyroidism should be made on clinical grounds based on maternal history, interpretation of serum TRAb levels, and fetal ultrasonography (396,411,422,423).

■ RECOMMENDATION 53

 Fetal surveillance should be performed in women who have uncontrolled hyperthyroidism in the second half of pregnancy, and in women with high TRAb levels detected at any time during pregnancy (greater than 3 times the upper limit of normal). A consultation with an experienced obstetrician or maternal–fetal medicine specialist is recommended. Monitoring may include ultrasound to assess heart rate, growth, amniotic fluid volume, and the presence of fetal goiter.

 Strong recommendation, moderate-quality evidence.

Question 61: When should umbilical blood sampling be considered in women with GD in pregnancy?

Umbilical cord blood sampling (cordocentesis) is associated with both fetal mortality and morbidity (424,425). It has been utilized when a mother is TRAb positive and treated with ATDs, a fetal goiter is present, and the thyroid status of the fetus is unclear (403,411,426). The presence of elevated TRAb alone is not an indication for cordocentesis (427).

■ RECOMMENDATION 54

 Cordocentesis should be used in rare circumstances and performed in an appropriate setting. It may occasionally be of use when fetal goiter is detected in women taking ATDs to help determine whether the fetus is hyperthyroid or hypothyroid.

 Weak recommendation, low-quality evidence.

Recently, biological assays able to distinguish between stimulating and blocking TRAb have become commercially available (428). The use of such assays for characterization of TRAb biological activity may theoretically make cordocentesis unnecessary, though the complexity of differential clearance of stimulatory and blocking antibodies should be considered (see Section X, Fetal and Neonatal Considerations).

Question 62: How should hyperthyroidism caused by autonomous thyroid nodules be handled in pregnancy?

One or more autonomous thyroid nodules is a common cause of hyperthyroidism in populations with current or previous mild to moderate iodine deficiency (429). However, even in such areas this type of hyperthyroidism is quite rare under the age of 40 years (430). Hyperthyroidism caused by autonomous nodules tends to develop more insidiously and be less severe than in GD.

In pregnancy, a major difference from GD is that no TRAb is produced by the mother, and consequently the fetal thyroid is not stimulated as it is in GD. Therefore, ATD therapy to make the mother euthyroid would significantly increase the risk of hypothyroidism and goiter in the fetus. Given the limited data on this subject, if ATDs are required in this setting, careful monitoring of the fetus should occur. Furthermore, the dose of ATD should be kept low, and maternal surgical therapy with removal of autonomous nodule(s) considered if signs of fetal hypothyroidism develop. Generally, if possible, ablative therapy should be considered before conception for hyperthyroidism caused by thyroid autonomy in women seeking future pregnancy.

If only low-grade autonomous thyroid hormone production with subclinical hyperthyroidism is present in a nonpregnant woman, the physiological increase in thyroid hormone needs may ameliorate any hyperthyroidism in pregnancy. On the other hand, the early pregnancy high hCG level may theoretically activate nonaffected normal thyroid tissue and increase thyroid secretion in early pregnancy. Furthermore, hormone production in autonomous thyroid nodules will depend on the available amount of iodine. Therefore, women with such disease should probably refrain from taking iodine-containing supplements in pregnancy.

■ RECOMMENDATION 55

 If ATD therapy is given for hyperthyroidism caused by autonomous nodules, the fetus should be carefully monitored for goiter and signs of hypothyroidism during the second half of pregnancy. A low dose of ATD should be administered with the goal of maternal FT4 or TT4 concentration at the upper limit or moderately above the reference range.

 Strong recommendation, low-quality evidence.

Question 63: What are the etiologies of thyrotoxicosis in the postpartum period?

The most common cause of thyrotoxicosis in the postpartum period is postpartum thyroiditis (PPT), with a thyrotoxicosis prevalence of 4% (431,432) (see Section XII). Many cases are mild and of short duration and spontaneously revert to euthyroidism. More severe cases in need of a short course of β-blockers are seen in patients with high levels of TPOAb, and they are often followed by a period of hypothyroidism. The majority of women return to euthyroidism by 1 year post partum (433,434).

The postpartum increase in thyroid autoimmunity is also associated with a 3- to 4-fold increase in the incidence of new GD (401), in addition to the increase in the risk of relapse of GD in remission after previous ATD therapy. In one study, the overall relapse rate of GD following a pregnancy was 84%, as compared to a relapse rate of 56% in women who did not become pregnant (404).

Question 64: How should Graves' hyperthyroidism be treated in lactating women?

See Section XI.

Question 65: What is the prevalence of thyroid nodules during pregnancy?

Three studies have evaluated the prevalence of thyroid nodules during pregnancy, the impact of pregnancy on nodular size, and the proportion of women who have new nodules detected during pregnancy. All three studies were performed in areas with mild to moderate iodine deficiency (Brussels, China, and Germany), with the majority using ultrasound examination of the gland. The prevalence of thyroid nodules varied between 3% and 21% (187,435,436) and increased with increasing parity (9.4% without a prior pregnancy, 20.7% with one prior pregnancy, 20.7% with two prior pregnancies, and 33.9% with three or more prior pregnancies) (435). In the Belgian study, 60% of nodules doubled in size during pregnancy, yet remained between 5 and 12 mm (187). The maximum diameter of the dominant nodule did not increase during pregnancy in the Chinese study (mean diameter was 5.1 mm, 5.1 mm, and 5.5 mm, in the first, second, and third trimester, respectively). However, an increase in nodular volume was reported during pregnancy, with a return to first-trimester volumes by the third postpartum month (436). The studies in Belgium and Germany reported that 11%–20% of women with a nodule detected in the first trimester of pregnancy developed a second nodule through the course of pregnancy (187,436). Increasing age is associated with an increase in the proportion of pregnant women who have thyroid nodules (187,436).

Question 66: What is the prevalence of thyroid cancer in women with thyroid nodules discovered during pregnancy?

Data regarding the prevalence of thyroid cancer derive from three retrospective studies performed at three tertiary referral centers (Mayo Clinic, George Washington University Hospital, and Mount Sinai Hospital-Toronto), one prospective study, and a retrospective study of the California Cancer Registry. The research performed at the referral centers revealed a 15% (6/40, Mayo Clinic) (437), 12% (7/57, George Washington Hospital) (438), and 43% (7/16, Mount Sinai Hospital-Toronto) (439) rate of thyroid malignancy. All three studies are limited by two major methodological flaws. The first, and most problematic, is selection bias. The population studies consisted of women referred for diagnosis and treatment at major referral centers. As such, they are not representative of the population of pregnant women with thyroid nodules detected during pregnancy. Instead, they represent a select group of women referred to a tertiary medical center because of physician concern for thyroid malignancy. Consequently, they likely represent an overrepresentation of the prevalence of thyroid malignancy during pregnancy. Secondly, each study was retrospective in nature and therefore neither the accuracy of the diagnosis nor the completeness of case identification within the database could be verified. The lone prospective study investigated the prevalence of thyroid cancer during pregnancy in 212 Chinese women. The study found a 15.3% (34/221) rate of thyroid nodules and a 0% rate of malignancy. Interpreting these data is hampered by the limited number of women enrolled in the study (436). The final study consisted of a population-based retrospective analysis of all obstetrical deliveries in California from 1991 through 1999 identified by cross-referencing maternal/neonatal hospital discharges in California and the California Cancer Registry (n = 4,846,505 women). A prevalence of thyroid cancer in pregnancy of 14.4/100,000 was reported, with papillary cancer being the most frequent pathological type (440). Timing of diagnosis of the thyroid malignancy was as follows: 3.3/100,000 cases diagnosed before delivery, 0.3/100,000 at delivery, and 10.8/100,000 within 1 year post partum.

Question 67: What is the optimal diagnostic strategy for thyroid nodules detected during pregnancy?

Question 68: Does pregnancy impact the prognosis of newly diagnosed thyroid carcinoma?

Six studies have compared the diagnostic features and prognosis of women diagnosed with differentiated thyroid cancer (DTC) either during gestation or within the first year post partum to nonpregnant women (464 –466,472 –474). None of the studies were RCTs, all were retrospective, and the size of many of the studies was limited. Four studies either predate or do not use the contemporary tools for recurrence detection, including sensitive serum Tg assays and neck sonography (464 –466,473). Using large database evaluation (464 –466), the analyzed outcomes were either overall survival or disease-free survival, neither of which differed between the two groups. This outcome is not unexpected because the vast majority (>99%) of both the pregnant and nonpregnant populations had stage 1 disease. In addition, all studies consistently found that timing of surgery, either during pregnancy or deferred until postpartum, did not affect survival.

Recently, using both sonography and sensitive Tg assays, two studies from referral centers in Italy reported higher persistence or recurrence rates in women diagnosed with thyroid cancer during pregnancy or within 12–24 months of delivery compared to both nulliparous women and those diagnosed after a longer interval post partum (472,474). DTC persistence or recurrence was defined as (a) either stimulated or suppressed serum Tg >2 ng/mL, (b) increasing levels of Tg autoantibodies, (c) sonography detection of metastatic lymph nodes, or (d) uptake outside the thyroid bed on scintigraphy. Subjects were followed for a median of about 6 years after definitive therapy. Vannucchi et al. (472) observed a much higher than expected persistence/recurrence rate (60%) in the pregnant/early postpartum group than otherwise anticipated for Stage 1 disease, compared with 4% in women diagnosed more than 12 months after delivery and 13% in nulliparous women. The majority of the recurrences (67%) were identified by Tg or TgAb criteria without identification of structural disease, so it is unclear if Tg/TgAb detection is associated with other confounding factors such as potentially less-aggressive surgery if done during gestation. Messuti et al. (474) reported a recurrence rate of 10% that was more typical for early stage disease, but this was statistically higher than the 1.3% for women diagnosed 2 years after delivery and 4.3% for nulliparous women. No details for recurrence classification, biochemical versus structural, were provided. Importantly, delaying surgery until after delivery did not further affect recurrence rates among women diagnosed with DTC during gestation in either study. Although several potential explanations exist for the poorer prognosis of pregnant/early postpartum women versus the comparator groups, findings are not consistent in the two studies. Metastatic cervical adenopathy at diagnosis was more frequent in pregnant/postpartum women compared to the other groups in one study (63% vs. ∼40%) (472) but not in the other (474). Vannucchi et al. (472) documented more frequent presence of estrogen receptor α in the tumors of women diagnosed during pregnancy/early post partum as compared to the other two groups, which may indicate that the poorer prognosis is related to estrogen-mediated growth and stimulation of the MAPK pathway. However, this was not confirmed in the study by Messuti et al. (474). The discrepancy may be due to methodologic issues because a less-dilute estrogen receptor Ab was employed in the study with the positive findings. Hence, there is a recent observation of higher persistence/recurrence rates for women diagnosed with DTC during or immediately after pregnancy, without elucidation of pathophysiologic mechanisms. Further studies may help to clarify causality. Importantly, survival is not compromised.

The impact of pregnancy on women with medullary or anaplastic carcinoma is unknown.

Question 69: What are the perioperative risks to mother and fetus of thyroid surgery during pregnancy?

Surgery is the treatment of choice for DTC. As deferring surgery until postpartum has not been associated with a worse prognosis, it is imperative to assess maternal and neonatal complications before recommending an operation during pregnancy. Between 1986 and 2008, nine studies evaluated the impact of thyroidectomy during pregnancy on a total of 113 patients (study size ranged from 1 to 96) (437,441,442,444,446,475 –478). The majority, but not all, of the operations were performed in the second trimester. There were no maternal or fetal complications in any of the studies. A 2009 population-based study of nonfederal hospitals in the United States compared 201 pregnant women who underwent thyroid and parathyroid surgery during pregnancy with 31,155 similarly treated nonpregnant women. One hundred sixty-five operations were thyroid related and 46% of the women had thyroid cancer. Pregnant patients had a higher rate of endocrine and general complications, longer lengths of stay, and higher hospital costs. The fetal and maternal complication rates were 5.5% and 4.5%, respectively (479). Interpretation of the results of this study is difficult because there were substantial baseline differences between the two groups. Pregnant women were more likely to have either urgent or emergent admissions and had a higher percentage of government insurance. However, surgical volume was also an independent predictor of poor outcomes, so that if surgery is performed during pregnancy, an experienced surgeon is preferred. Recently, a study from Japan focused on 45 patients with DTC, of whom 24 had thyroidectomy during pregnancy (19 during the second trimester) and 21 had surgery after delivery. The groups were compared with regard to disease and other factors. There was no difference in recurrence rates in the mothers and there were no pregnancy losses or birth defects. The authors concluded that although thyroidectomy can be performed safely in the second trimester, surgery after delivery is recommended for most patients with nonaggressive DTC (480). Thus, in instances in which surgery during pregnancy is indicated or desired, it should be performed in the second trimester in order to minimize complications to both the mother and fetus (altered organogenesis and spontaneous abortion in the first trimester, preterm labor and delivery in the third trimester) (481), preferably by an experienced thyroid surgeon. The risk of post-thyroidectomy maternal hypothyroidism and hypoparathyroidism should also be considered.

Question 70: How should cytologically benign thyroid nodules be managed during pregnancy?

Although pregnancy is a risk factor for modest progression of nodular thyroid disease, there is no evidence demonstrating that LT4 is effective in decreasing the size or arresting the growth of thyroid nodules during pregnancy. Hence, LT4 suppressive therapy for thyroid nodules is not recommended during pregnancy. Nodules that were benign on FNA but show rapid growth or ultrasound changes suspicious for malignancy should be evaluated with a repeat FNA and be considered for surgical intervention. In the absence of rapid growth, nodules with biopsies that are benign do not require surgery during pregnancy (482).

■ RECOMMENDATION 60

 Pregnant women with cytologically benign thyroid nodules do not require special surveillance strategies during pregnancy and should be managed according to the 2015 ATA Management Guidelines for Adult Patients with Thyroid Nodules and Differentiated Thyroid Cancer.

 Strong recommendation, moderate-quality evidence.

Question 71: How should cytologically indeterminate nodules be managed during pregnancy?

There have been no prospective studies evaluating the outcome and prognosis of women with an FNA that is interpreted as either atypia of undetermined significance/follicular lesion of undetermined significance (AUS/FLUS), suspicious for follicular neoplasm (SFN), or suspicious for malignancy (SUSP). The reported malignancy rates associated with these cytologic diagnoses range from 6% to 48% for AUS/FLUS, 14% to 34% for SFN, and 53% to 87% for SUSP (483). Although molecular testing for cytologically indeterminate nodules is now being considered, no validation studies address application of these tests in pregnant women. It is theoretically possible that thyroid gestational stimulation may alter a nodule's gene expression and change diagnostic performance of molecular tests based upon RNA expression, whereas testing based upon either single base-pair DNA mutations or translocations would be less likely to be affected.

Since prognosis for DTC diagnosed during pregnancy is not adversely affected by performing surgery post partum, it is reasonable to defer surgery until following delivery. Because the majority of these women will have benign nodules, LT4 therapy during pregnancy is not recommended.

■ RECOMMENDATION 61

 Pregnant women with cytologically indeterminate (AUS/FLUS, SFN, or SUSP) nodules, in the absence of cytologically malignant lymph nodes or other signs of metastatic disease, do not routinely require surgery while pregnant.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 62

 During pregnancy, if there is clinical suspicion of an aggressive behavior in cytologically indeterminate nodules, surgery may be considered.

 Weak recommendation, low-quality evidence.

■ RECOMMENDATION 63

 Molecular testing is not recommended for evaluation of cytologically indeterminate nodules during pregnancy.

 Strong recommendation, low-quality evidence.

Question 72: How should newly diagnosed thyroid carcinoma be managed during pregnancy?

The 2015 ATA guidelines (443) recommend that a nodule with cytology indicating papillary thyroid carcinoma discovered early in pregnancy should be monitored sonographically and, if either it grows substantially by 24 weeks gestation (50% in volume and 20% in diameter in two dimensions), or if metastatic cervical lymph nodes are present, surgery should be considered in the second trimester. However, if it remains stable by midgestation, or if it is diagnosed in the second half of pregnancy, surgery may be performed after delivery. Surgery in the second trimester is an option if the DTC is advanced stage at diagnosis or if the cytology indicates medullary or anaplastic carcinoma.

If surgery is not performed, the utility of thyroid hormone therapy targeted to lower serum TSH levels to improve the prognosis of DTC diagnosed during gestation is not known. Because higher serum TSH levels may be correlated with a more advanced stage of cancer at surgery (484), if the patient's serum TSH is >2 mU/L, it may be reasonable to initiate thyroid hormone therapy to maintain the TSH between 0.3 and 2.0 mU/L for the remainder of gestation.

■ RECOMMENDATION 64

 PTC detected in early pregnancy should be monitored sonographically. If it grows substantially before 24–26 weeks gestation, or if cytologically malignant cervical lymph nodes are present, surgery should be considered during pregnancy. However, if the disease remains stable by midgestation, or if it is diagnosed in the second half of pregnancy, surgery may be deferred until after delivery.

 Weak recommendation, low-quality evidence.

■ RECOMMENDATION 65

 The impact of pregnancy on women with newly diagnosed medullary carcinoma or anaplastic cancer is unknown. However, a delay in treatment is likely to adversely affect outcome. Therefore, surgery should be strongly considered, following assessment of all clinical factors.

 Strong recommendation, low-quality evidence.

Question 73: What are the TSH goals for pregnant women with previously treated thyroid cancer receiving LT4 therapy?

Based on studies which have demonstrated a lack of maternal or neonatal complications from subclinical hyperthyroidism, it is reasonable to assume that the preconception degree of TSH suppression can be safely maintained throughout pregnancy. The appropriate level of TSH suppression depends upon preconception risk of residual or recurrent disease. According to the 2009 and 2015 ATA management guidelines for DTC (443,482), and the ETA consensus (485), the serum TSH should be maintained indefinitely below 0.1 mU/L in patients with persistent structural disease. Target TSH values are first based upon initial risk of recurrence determined postoperatively (Table 12 of the 2015 ATA guidelines) and then are modified based upon classification of the patient's response to therapy as defined in Table 13 of the 2015 ATA guidelines. Hence, for a patient at initial high risk for recurrence, TSH suppression at or below 0.1 mU/L is recommended. If the patient then demonstrates an excellent response to therapy at one year with an undetectable suppressed serum Tg and negative imaging, the TSH target may rise to the lower half of the reference range.

The main challenge in caring for women with previously treated DTC is maintaining the TSH level within the preconception range. In a recent report (325), thyroid cancer patients were reported to require smaller LT4 dose increases than patients who had undergone thyroid ablation for benign thyroid disorders or patients with primary hypothyroidism. However, this outcome was because the nonpregnant serum level TSH was <0.1 mU/L in 83% of the thyroid cancer patients prior to pregnancy and this target was not achieved during gestation. It is likely that if the prepregnancy TSH target level is maintained throughout gestation, the LT4 dosage increase will be similar to the ∼49% required in ablated hypothyroidism. Patients require careful monitoring of thyroid function tests in order to avoid hypothyroidism.

Thyroid function should be evaluated as soon as pregnancy is confirmed. The adequacy of LT4 treatment should be checked 4 weeks after any LT4 dose change. The same laboratory should be utilized to monitor TSH and Tg levels before, during, and after pregnancy.

■ RECOMMENDATION 66

 Pregnant women with thyroid cancer should be managed at the same TSH goal as determined preconception. TSH should be monitored approximately every 4 weeks until 16–20 weeks of gestation, and at least once between 26 and 32 weeks of gestation.

 Strong recommendation, moderate-quality evidence.

Question 74: What is the effect of therapeutic radioactive iodine treatment on subsequent pregnancies?

Following surgery for DTC, many patients will receive an ablative dose of radioactive iodine. The possible deleterious effect of radiation on gonadal function and the outcome of subsequent pregnancies has been evaluated by Sawka et al. (486) and Garsi et al. (487); the latter collected 2673 pregnancies, 483 of which occurred after radioiodine treatment. Neither study found an increased risk of infertility, pregnancy loss, stillbirths, neonatal mortality, congenital malformations, preterm births, low birth weight, or death during the first year of life or cancers in offspring.

Radioiodine treatment may lead to suboptimal thyroid hormonal control during the month following administration. The potential adverse effect upon the pregnancy is described in Section VII. It therefore seems reasonable to wait a minimum of 6 months to ensure that thyroid hormonal control is stable before conceiving following radioiodine ablative therapy.

Iodine 131 may affect spermatogenesis. In one study following men after ¹³¹I therapy, there was a dose-dependent increase in follicle-stimulating hormone levels and a reduction in normokinetic sperm (488). Therefore, it seems prudent for a man who has received ¹³¹I to wait 120 days (the lifespan of sperm) after ¹³¹I therapy before attempting to conceive.

■ RECOMMENDATION 67

 Pregnancy should be deferred for 6 months after a woman has received therapeutic radioactive iodine (¹³¹I) treatment.

 Strong recommendation, low-quality evidence.

Question 75: What is the effect of tyrosine kinase inhibitors on pregnancy?

Several tyrosine kinase inhibitor (TKI) medications are now FDA approved for the therapy of metastatic differentiated and medullary thyroid cancer. These include sorafenib, lenvatinib, and cabozantinib. All three drugs have demonstrated both teratogenicity and embryo toxicity in animals when administered at doses below the recommended human dose (489). There are no human studies. The FDA recommends that women be advised of the potential risk to the fetus with TKI medications. However, specific advisories vary by medication. Furthermore, the use of any TKI must be guided by an assessment of risks to benefits that is also affected by the stage of disease and recommended drug indications. For sorafenib, the FDA prescribing information counsels to avoid pregnancy while taking the drug. For lenvatinib, contraception is explicitly recommended. For cabozantinib, no additional warnings are listed (489).

Question 76: Does pregnancy increase the risk of DTC recurrence?

Five studies have evaluated the impact of pregnancy after a woman has been treated for DTC. Two were published five decades ago, predating contemporary tools for thyroid cancer surveillance (490,491). Still, Rosvoll et al. (490) demonstrated pregnancy did not accelerate tumor growth in the 22 women who had stable or slowly progressive metastatic disease. Hill et al. (491) found no difference in clinically detected thyroid cancer recurrence rates in 70 women who had one or more pregnancies following initial diagnosis of DTC and 109 women who had no subsequent pregnancies. A more recent study from Memorial Sloan Kettering reported on 36 women who became pregnant between 1997 and 2006, a median of 4.3 years after initial treatment for DTC (492). Although all had serum Tg testing, only 16 (44%) had ultrasound imaging before and after delivery. Disease progression occurred in one of the two women who had metastatic cervical lymph nodes on baseline imaging (the lymph node grew from 2.3 cm to 2.7 cm). The mean suppressed Tg after delivery was not significantly different than the prepartum value. However, eight women had postpartum Tg values more than 20% higher than before pregnancy (three with known disease, five with no clinical evidence of disease). No evidence of recurrence was detected in the early postpartum period in women with negative prepregnancy neck ultrasound and a suppressed serum Tg <1 ng/mL. Two more recent studies (493,494) routinely used neck sonography and sensitive Tg assays to evaluate cancer progression during pregnancy. No recurrence was observed in 64 Brazilian women (48 papillary cancer and 16 follicular cancer) who had suppressed Tg levels <2 ng/mL and negative neck sonography prepregnancy. However, follow-up was limited to only 6 months after delivery (493). Using a definition of disease-free as a suppressed Tg <0.9 ng/mL and negative neck ultrasound, Hirsch and colleagues (494) followed 63 women who had given birth after receiving treatment for DTC (all papillary thyroid cancer) for a mean of 4.8 years. None of the 50 women categorized as disease-free had progression of disease. However, 6 of the 13 women (46%) with persistent disease prepregnancy experienced progression. Metastatic cervical lymph nodes grew in 50% of those with known prepregnancy nodal disease, and new nodal metastases appeared in two of the seven women with negative baseline sonography.

Thus, pregnancy does not pose a risk for tumor recurrence in women without structural or biochemical disease present prior to the pregnancy. Therefore, women with an excellent response to therapy as defined by the 2015 ATA guidelines (443) do not require additional monitoring during gestation. However, pregnancy may represent a stimulus to thyroid cancer growth in patients with known structural (ATA 2015 structural incomplete response to therapy) or biochemical (ATA 2015 biochemical incomplete response to therapy) disease present at the time of conception and requires monitoring.

Question 77: What type of monitoring should be performed during pregnancy in a patient who has already been treated for DTC prior to pregnancy?

■ RECOMMENDATION 68

 Ultrasound and Tg monitoring during pregnancy is not required in women with a history of previously treated DTC with undetectable serum Tg levels (in the absence of Tg autoantibodies) classified as having no biochemical or structural evidence of disease prior to pregnancy.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 69

 Ultrasound and Tg monitoring should be performed during pregnancy in women diagnosed with well-differentiated thyroid cancer and a biochemically or structurally incomplete response to therapy, or in patients known to have active recurrent or residual disease.

 Strong recommendation, moderate-quality evidence.

Question 78: What type of monitoring should be performed during pregnancy in a patient who is under active surveillance for papillary thyroid microcarcinoma?

Although FNA of subcentimeter nodules in the absence of suspicious cervical adenopathy is not recommended by the 2015 ATA guidelines (443), there are times when this has been performed with subsequent diagnosis of a papillary thyroid microcarcinoma (PTMC). In some situations, women with this diagnosis may have chosen active surveillance. Observation with active surveillance has been demonstrated to be a viable option for patients cytologically diagnosed with PTMC (size <1 cm) in the absence of suspicious lymph nodes or extrathyroidal invasion by sonographic imaging (495). Although tumor growth >3 mm and appearance of novel metastatic lymph nodes occurred respectively overall in 8% and 3.8% of a cohort of 1235 Japanese patients, progression was more likely in patients younger than 40 years of age. Therefore, the same group published their observations of nine women with PTMC who became pregnant during active surveillance and compared outcomes to 27 age-matched nonpregnant women. During pregnancy, growth ≥3 mm occurred more frequently in pregnant (44%) than nonpregnant (11%) women (p = 0.0497). There was no correlation between growth and gestational TSH or Tg levels, and novel metastatic cervical lymph nodes were not observed. Three of the four patients who experienced tumor growth had surgery post partum, but the tumor decreased in size post partum in the fourth patient who is still under active surveillance. Furthermore, the PTMC enlarged by less than 3 mm in the other five pregnant women, but then decreased in three postpartum (496). This supports the hypothesis that pregnancy is either permissive for or promotes growth of papillary thyroid cancer. While this may have no impact on overall health, it seems reasonable to monitor such women with serial sonographic evaluations.

■ RECOMMENDATION 70

 Ultrasound monitoring of the maternal thyroid should be performed each trimester during pregnancy in women diagnosed with PTMC who are under active surveillance.

 Weak recommendation, low-quality evidence.

Question 79: What special considerations should be followed for women with medullary cancer due to germline RET mutations?

Hereditary medullary thyroid cancer (MTC) may occur as a result of activating germline mutations of the RET oncogene. Over 100 mutations, duplications, insertions, or deletions involving RET have been identified in patients with hereditary MEN2a (including familial MTC) or MEN2b syndromes. The newly Revised American Thyroid Association Guidelines for the Management of Medullary Thyroid Carcinoma have graded the most common mutations according to risk of aggressive MTC and likelihood of concomitant pheochromocytoma and hyperparathyroidism (497). For a woman with a given RET mutation with or without clinical MTC, who is either pregnant or contemplating conception, the associated risks of MTC aggressiveness combined with her own history may inform her about whether to seek genetic counseling. Both prenatal and pre-implantation genetic testing are available, and the reader is referred to the excellent discussion in the 2015 ATA guidelines that culminates in Recommendation 12: “The duty to warn of genetic risk extends to both preconception and prenatal contexts. Genetic counseling about the options of pre-implantation or prenatal diagnostic testing should be considered for all RET mutation carriers of childbearing age, particularly those with MEN2B. Parents who do not wish to have prenatal RET gene mutation testing should be offered genetic counseling and informed of the availability of genetic testing of their child to detect a mutated RET allele. This is particularly important for mutations associated with the onset of MTC before 5 years of age.” In addition, testing to exclude a pheochromocytoma should be done prior to pregnancy for all women with MEN2. No studies address the effect of pregnancy on MTC progression in patients with biochemical or structural evidence of residual or metastatic disease.

Question 80: What is the relationship between maternal and fetal thyroid hormone status?

Under normal circumstances the fetal hypothalamic-pituitary-thyroid system develops relatively independently of maternal influence because of the presence of the placenta, which regulates the passage of many substances, including T4, to the fetus. Circulating fetal thyroid hormone levels therefore largely reflect the stage of fetal maturation.

Thyroid hormone receptors are present in the fetal brain at low concentrations up to week 10 of gestation. Thyroid hormone receptor concentrations then rapidly increase 10-fold through week 16 of pregnancy (498).

Serum TT4 and FT4 are first measurable in low levels in fetal serum at 12–14 weeks of development (499 –501). Based on values obtained by fetal cord sampling in normal pregnant women, it has been estimated that the mean fetal serum TT4 is 2 μg/dL (26 nmol/L) at 12 weeks, and a relatively larger proportion is in the free form. Beginning in midgestation, the fetal TT4 concentration begins to increase and typically reaches values comparable to nonpregnant females (10 μg/dL [138 nmol/L]) by 36 weeks. This rise in TT4 follows an increase in the serum TBG concentration due primarily to the stimulatory effects of maternal estrogen on the fetal liver. In addition, owing to up-regulation of the TSH receptor, there is an increase in the fetal FT4 concentration from a mean of approximately 0.1 ng/dL (1.3 pmol/L) to 2 ng/dL (25.7 pmol/L), beginning early in the third trimester (502).

In contrast to T4, the fetal circulating levels of its active metabolite T3 remain significantly lower in fetal life than postnatally, whereas the inactive metabolites reverse T3 and T3 sulfate are elevated. The fetal serum T3 concentration is approximately 6 ng/dL (0.09 nmol/L) at 12 weeks, rising to 45 ng/dL (0.68 nmol/L) at 36 weeks. Despite the low levels of circulating T3, brain T3 levels are 60%–80% those of the adult by fetal age 20–26 weeks (503). This reflects the importance of local conversion of T4 to T3 in the brain itself. During the second trimester, expression of type 2 deiodinase, which converts T4 to T3, increases in the fetal cerebral cortex in parallel to the T3 concentrations (504). T3 sulfate may also serve as an alternate source of T3 for the pituitary (505).

The serum concentration of TSH rises modestly from ∼4 mU/L at 12 weeks to 8 mU/L at term and is always higher than the corresponding maternal levels (500,506). The reason for this is not completely understood but may be a consequence of incomplete maturation of the hypothalamic-pituitary-thyroid axis and/or the high TRH levels in the fetus.

Question 81: What is the role of placental transfer?

Question 82: When maternal thyroid abnormalities are detected, how and when should information be provided to the neonatologist or pediatrician?

Question 83: What information should be provided to the neonatologist or pediatrician?

Most infants born to women with known thyroid illness are healthy. Furthermore, all newborn infants in the United States are screened for thyroid dysfunction as part of universal screening mandates. Nonetheless, maternal thyroid illness, abnormal maternal thyroid function, the presence of maternal TSH receptor antibodies, and/or the use of antithyroid medications during gestation can each contribute to adverse effects in the newborn. Therefore, knowledge of these facts is important and should be documented in the newborn infant's medical record. In many cases, direct communication with the neonatologist or pediatrician is recommended, especially in the setting of severe maternal hyperthyroidism or maternal use of ATDs at any time throughout pregnancy. The etiology, timing, severity, and treatment of maternal thyroid disease should all be conveyed. In women with GD (even if previously ablated with ¹³¹I or athyreotic following surgery) concentrations of circulating TRAb should be assessed at initial thyroid function testing in early pregnancy, and if TRAb is positive, again at weeks 18–22 (see Section VIII). Such information should be documented and conveyed to the pediatrician because maternally derived thyroid antibodies as well as antithyroid medication can transfer to the fetus and affect newborn thyroid health.

■ RECOMMENDATION 71

 A history of maternal thyroid illness, use of antithyroid medications (PTU, MMI) during gestation, or measurements of abnormal maternal thyroid function or TRAb during gestation should be communicated to the newborn's neonatologist or pediatrician.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 72

 The severity of maternal and fetal thyroid illness should guide the timing of communication. Severe, progressive, or complex thyroid illness during pregnancy mandates communication with the neonatologist or pediatrician before birth and consideration of consultation with a pediatric endocrinologist. Most other illness is optimally communicated shortly after birth.

 Strong recommendation, moderate-quality evidence.

Question 84: Should all neonates be screened for thyroid dysfunction?

Congenital hypothyroidism is one of the most frequent treatable causes of intellectual impairment, and newborn screening for the detection of congenital hypothyroidism is performed routinely in all 50 states of the United States, as well as in Canada, Europe, Israel, Japan, Australia, and New Zealand (521). Screening programs are also being introduced in many developing countries, accounting overall for approximately one third of the world's population (521). Blood spot specimens eluted from filter paper are utilized. The optimal timing of measurement is at 2–5 days of age in order to avoid confounding by the physiologic surge in neonatal TSH, which occurs shortly after birth. Some programs employ primary T4 screening with TSH measurement when specimens demonstrate a value below a specified cutoff. Others use a primary TSH determination with reflex T4 strategy. The introduction of newborn screening for congenital hypothyroidism has led to the virtual elimination of intellectual impairment due to hypothyroidism so long as early and adequate postnatal LT4 treatment is initiated immediately (522,523).

■ RECOMMENDATION 73

 All newborns should be screened for hypothyroidism by blood spot analysis typically 2–5 days after birth.

 Strong recommendation, high-quality evidence.

Question 85: How is neonatal hypothyroidism treated?

In women receiving ATDs at the time of delivery, transplacental transfer of the medication can potentially induce neonatal hypothyroidism. In such cases, neonatal metabolism removes the remaining MMI or PTU from the newborn circulation. This results in the return of normal thyroid function, typically within 3–5 days. For neonates with congenital hypothyroidism, LT4 must be administered. The recommended starting dose for full-term infants is 10–15 μg/kg/d, and depends on the severity of the initial hypothyroidism (522,523). In premature hypothyroid infants a lower dose is generally utilized. For optimal cognitive outcomes, therapy should be initiated within 2 weeks of life. Both inadequate and excess thyroid hormone replacement may be harmful, so close follow-up is important, particularly during the first 3 years of life. During this time, brain development is thyroid hormone dependent. Age-dependent normative values for TSH and T4 should be used to guide therapy. Until further information becomes available, infants with severe congenital hypothyroidism (i.e., initial serum TSH concentration >100 mU/L) should remain on a single formulation of LT4 without substitution (524). Treatment should be guided by a pediatric endocrinologist or experienced pediatrician.

The presence of a goiter in a newborn should prompt referral to a pediatric endocrinologist. An ultrasound examination should be performed to evaluate the patency of the trachea. A goiter may reflect a hypothyroid or hyperthyroid state in response to maternal thyroid illness or overtreatment with antithyroid medications. Regardless, close observation and urgent treatment may be required to achieve a euthyroid state and avoid airway compromise in the newborn.

For a more detailed discussion of this and other aspects of neonatal hypothyroidism, the reader is referred to several recent excellent reviews and guidelines (522,523,525).

Question 86: How is neonatal hyperthyroidism treated?

Most neonatal hyperthyroidism is caused by maternal transfer of TRAb to the fetus. Typically, neonatal GD does not present until the end of the first week of life when maternal ATD, but not the TRAb, have been cleared from the neonatal circulation. This may be delayed in babies born to mothers with a mixture of stimulating and blocking antibodies. Thus, results on newborn screening may be paradoxically normal. A high degree of suspicion for infants at risk is important.

The conventional treatment for neonatal hyperthyroidism is PTU 5–10 mg/kg administered in divided doses. However recent evidence of rare hepatotoxicity due to this agent has led to a switch to MMI therapy (0.5–1 mg/d) by many pediatric endocrinologists. Propranolol (2 mg/kg) may be added if the hyperthyroidism is severe. Close follow-up of the affected newborn is important, with downward adjustment in the dose of ATD required as the hyperthyroidism resolves. The usual duration of neonatal GD is 1–3 months, but it depends on antibody potency. Separate from neonatal GD (which is self-limited), neonatal hyperthyroidism may rarely be caused by a gain-of-function mutation in the TSH receptor, or by McCune–Albright syndrome. In such cases, newborn hyperthyroidism may be permanent (526).

Question 87: Does maternal thyroid hormone status impact lactation?

Question 88: In the breastfeeding mother, should maternal hypothyroidism be treated to improve lactation?

Question 89: In the breastfeeding mother, should maternal hyperthyroidism be treated to improve lactation?

Abnormal maternal thyroid hormone concentrations (both hypothyroidism and hyperthyroidism) can impact milk letdown and the ability to successfully breastfeed. While there are no controlled or randomized human trials, several case series and cohort analyses have been reported. One study reported 26 human pregnancies complicated by hypothyroidism in New Delhi, India. Deficient lactation was reported in 19.2% of subjects (527). Other case reports describe impaired lactation, which improved upon successful treatment of hypothyroidism (528). Interventional trials in nonhuman mammals corroborate these findings. Hapon and colleagues (529) investigated the effects of PTU-induced hypothyroidism in rats at days 10–12 of lactation. After 30 minutes of suckling, milk excretion was reduced compared to euthyroid controls. The authors suggested mammary function was maintained while a reduced milk ejection reflex was likely responsible. A separate investigation of Brahman cows demonstrated reduced milk production in PTU-induced hypothyroid cows compared to euthyroid controls. Following cessation of PTU and normalization of thyroid status, lactation in these animals returned to baseline (530). Together, these data strongly suggest maternal hypothyroidism can adversely affect lactation. However, with limited human data and no controlled or randomized intervention studies, it is unclear what proportion of hypothyroid mothers would be affected. Furthermore, it is possible that many mothers experiencing only mild to moderate hypothyroidism could experience only minimal difficulties with breastfeeding and milk production. Importantly, the available data confirm that maternal hypothyroidism itself is not harmful to the breastfeeding infant per se, so long as adequate nutrition (milk) is provided. Therefore, it appears most reasonable to recommend evaluation of thyroid function in women who demonstrate poor lactation, with the goal of effectively returning any hypothyroid mother to a euthyroid state.

Maternal hyperthyroidism may also impair lactation, although the data are less clear. No controlled human trial or large cohort analysis has been performed. Two rat studies demonstrated adverse effects of hyperthyroidism on lactation. In one study, moderate to severe hyperthyroidism was induced in pregnant rats during and after pregnancy. Complete lactation failure was noted (531). A second study in which moderate hyperthyroidism was induced demonstrated continued ability of rats to lactate, but impairment of milk ejection. Ultimately, this resulted in death of the offspring due to malnutrition (532). Such limited data, obtained from only nonhuman populations, make it impossible to draw firm conclusions. Therefore, no recommendation can be made to treat maternal hyperthyroidism on the grounds that it will improve lactation.

■ RECOMMENDATION 74

 As maternal hypothyroidism can adversely impact lactation, women experiencing poor lactation without other identified causes should have TSH measured to assess for thyroid dysfunction.

 Weak recommendation, low-quality evidence.

■ RECOMMENDATION 75

 Given its adverse impact upon milk production and letdown, subclinical and overt hypothyroidism should be treated in lactating women seeking to breastfeed.

 Weak recommendation, low-quality evidence.

■ RECOMMENDATION 76

 The impact of maternal hyperthyroidism upon lactation is not well understood. Therefore, no recommendation to treat maternal hyperthyroidism on the grounds of improving lactation can be made at this time.

 No recommendation, insufficient evidence.

Question 90: Is maternal thyroid hormone (thyroxine or LT4) transferred to the newborn via breast milk?

A small amount of maternal thyroid hormone is present in the breast milk of lactating women, and also detected in cow, goat, and sheep milk. A study investigated T4 and T3 in the breast milk of mothers who delivered very premature infants. Thyroxine concentration in breast milk ranged from 0.17 to 1.83 μg/L (mean 0.83 μg/L) (533). Assuming an average consumption of ∼750 mL/d of breast milk, this equates to approximately 0.62 μg of T4 per day. Assuming total T4 requirements of 40–60 μg/d for most athyreotic infants, the amount of thyroxine transferred to the infant via the human lactating breast is ∼1% of the total daily requirement. Thus, maternal transfer of thyroid hormone does not have a meaningful impact on the infant's thyroid hormone status.

Studies in cows (534), sheep (535), and donkeys (536) similarly find a small but detectable concentration of T4 and T3 in the milk supply. One study demonstrated secretion of 200–300 pg/mL of T3 in cow milk. This translates into a total daily consumption of 0.2–0.4 μg/d depending the age and feeding habits of the infant. Similarly, T4 concentrations in cows’ milk demonstrate an average concentration of 0.97 ng/mL, translating into the consumption of approximately 0.75–1.5 μg T4 per day. Thus, dairy milk obtained from cows, sheep, goats, or donkeys has little impact on the thyroid health of the infant or young child.

Question 91: Should diagnostic or therapeutic radiopharmaceuticals be administered to breastfeeding or lactating women?

Because radioactive iodine is concentrated in breast milk, and the half-life of ¹³¹I is relatively long (∼8 days), use of ¹³¹I for scanning or treatment is absolutely contraindicated in the breastfeeding mother. The half-life of ¹²³I is 13 hours. If circumstances necessitate the use of ¹²³I for of thyroid uptake measurement or scans while lactating, breast milk should be pumped and discarded for several days until the radioactive iodine has cleared from the body (537).

■ RECOMMENDATION 77

 The use of ¹³¹I is contraindicated during lactation. If required, ¹²³I can be used if breast milk is pumped and discarded for 3–4 days before breastfeeding is resumed. Similarly, Tc-99m pertechnetate administration requires breast milk to be pumped and discarded during the day of testing.

 Strong recommendation, moderate-quality evidence.

Question 92: Are ATDs (PTU, MMI) transferred into breast milk, and what are the clinical consequences to the breastfed infant?

Both PTU and MMI can be detected in the breast milk of treated hyperthyroid women. This finding raised initial concern that consumption of these medications could prove detrimental to the health of the breastfeeding infant. However, studies first performed using PTU confirmed that only a very small amount of the drug is transferred from maternal serum into breast milk. In a study of nine women given 200 mg PTU orally, milk PTU concentration was measured for 4 hours thereafter and only 0.007%–0.077% of the ingested dose was detected. The authors calculated that a lactating mother consuming PTU 200 mg three times daily would transmit only 149 μg (0.149 mg) of PTU daily to her infant (538). This is well below a therapeutic dose and deemed to pose no risk to the breastfeeding infant. Separate studies confirm normal thyroid function in breastfed infants of mothers consuming PTU (539,540). In these studies, two infants were found to have an elevated TSH within 1 week of birth, however, these values normalized thereafter despite continued breastfeeding. This suggests the cause of neonatal hypothyroidism in these infants was transplacental passage of PTU before birth, as opposed to any adverse effect of PTU transferred via breast milk. The remaining children who consumed breast milk expressed from mothers taking PTU had normal thyroid function.

Studies of MMI or CM, confirm a 4- to 7-fold higher proportion of the medication transferred into maternal milk in comparison to PTU. Approximately 0.1%–0.2% of an orally administered MMI/CM dose is excreted into breast milk (541,542). Johansen and colleagues (541) calculated that a single 40 mg dose of MMI could result in delivery of 70 μg (0.07 mg) to the breastfeeding infant. Several studies have separately investigated the effect of maternally ingested MMI or CM upon the thyroid status of the breastfeeding infant (543 –545). Virtually all have documented normal neonatal thyroid function. Furthermore, several women overtreated with MMI/CM were found to have elevated TSH concentrations (TSH 19–102 mU/L). Even in these situations, normal neonatal thyroid function was nonetheless confirmed in breastfeeding infants.

The largest study investigating the effects of maternal MMI consumption during lactation was performed by Azizi and colleagues (545). Importantly, this study assessed both neonatal thyroid function in the breastfeeding offspring, but also intellectual development and physical growth in a subset of infants. Verbal and performance IQ scores were measured in 14 children who breastfed from MMI-treated mothers, with comparison to 17 control children. Testing was performed between 48 and 74 months of age. No difference was detected in the IQ or physical development of the breastfeeding children compared to the control children (543).

Together, these data have led experts to confirm the safety of low to moderate doses of both PTU and MMI/CM in breastfeeding infants. However, given the relatively small size of the studied population, maximal daily doses of 20 mg MMI or 450 mg PTU are advised (352).

Question 93: What is the approach to the medical treatment of maternal hyperthyroidism in lactating women?

■ RECOMMENDATION 78

 Excepting treatment decisions specifically made on the grounds of improving lactation (discussed above), the decision to treat hyperthyroidism in lactating women should be guided by the same principles applied to nonlactating women.

 Strong recommendation, low-quality evidence.

Question 94: When medical treatment of maternal hyperthyroidism is indicated, what medications should be administered?

■ RECOMMENDATION 79

 When antithyroid medication is indicated for women who are lactating, both MMI (up to maximal dose of 20 mg/d) and PTU (up to maximal dose of 450 mg/d) can be administered. Given a small, but detectable amount of both PTU and MMI transferred into breast milk, the lowest effective does of MMI/CM or PTU should always be administered.

 Strong recommendation, moderate-quality evidence.

Question 95: How should breastfeeding children of mothers who are treated with antithyroid medications be monitored?

■ RECOMMENDATION 80

 Breastfed children of women who are treated with ATDs should be monitored for appropriate growth and development during routine pediatric health and wellness evaluations. Routine assessment of serum thyroid function in the child is not recommended.

 Weak recommendation, moderate-quality evidence.

Question 96: What are the iodine nutritional considerations in lactating women?

Iodine is an essential nutrient required for thyroid hormone production and is primarily derived from the diet. For most breastfeeding infants, breast milk is the sole (or primary) source of nutrition and thus of dietary iodine. Therefore, adequate iodine intake in the lactating mother positively affects infant health. Iodine requirements increase during pregnancy to accommodate the maternal–fetal unit. Post partum, maternal iodine requirements increase even further, as iodine is transferred to the breastfeeding infant. The dietary goals for individual total daily iodine intake (dietary and supplement) recommended by the Institute of Medicine are 290 μg/d for women who are breastfeeding (546). The WHO recommends 250 μg/d for both pregnant women and lactating women (547).

As discussed in Section IV, dietary sources of iodine vary regionally, and the dietary iodine intake of individuals cannot be reliably ascertained either by patient history or by any laboratory measure. Owing to concerns that a portion of women in the United States are mildly to moderately iodine deficient (which can be exacerbated during lactation due to increased demand), iodine supplementation for breastfeeding mothers is recommended. The ATA has previously recommended 150 μg/d as iodine supplementation for all North American women who are breastfeeding (117). The goal of supplementation is to augment rather than replace dietary iodine intake.

Unfortunately, recommendations for iodine supplementation when breastfeeding have not seen widespread adoption. In the NHANES 2001–2006 survey, only 15% of lactating women reported ingesting iodine-containing supplements (121). Furthermore, one analysis confirmed that only 51% of U.S. prenatal vitamins contained iodine (122). Therefore, physician counseling at the time of postbirth hospital discharge is paramount.

■ RECOMMENDATION 81

 All breastfeeding women should ingest approximately 250 μg of dietary iodine daily.

 Strong recommendation, high-quality evidence.

■ RECOMMENDATION 82

 Breastfeeding women should supplement their diet with a daily oral supplement that contains 150 μg of iodine. This is optimally delivered in the form of potassium iodide (present in a multivitamin) because kelp and other forms of seaweed do not provide a consistent delivery of daily iodine.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 83

 In severely iodine-deficient, low-resource regions, where universal salt iodization is lacking and daily supplementation is not feasible, lactating women should receive one dose of 400 mg iodine as oral iodized oil soon after delivery.

 Strong recommendation, high-quality evidence.

■ RECOMMENDATION 84

 As is the case during pregnancy, sustained iodine intake while breastfeeding that exceeds 500–1100 μg/d should be avoided because of concerns about the potential for inducing hypothyroidism in the infant.

 Strong recommendation, moderate-quality evidence.

Question 97: What is the definition of PPT and what are its clinical implications?

Postpartum thyroiditis is the occurrence of thyroid dysfunction, excluding GD, in the first postpartum year in women who were euthyroid prior to pregnancy (548). This is an inflammatory autoimmune condition. In the classic form, transient thyrotoxicosis is followed by transient hypothyroidism with a return to the euthyroid state by the end of the initial postpartum year (2). The clinical course of PPT varies, with approximately one quarter of patients presenting with the classical form, one quarter with isolated thyrotoxicosis, and one half presenting with isolated hypothyroidism (431). The thyrotoxic phase of PPT typically occurs between 2 and 6 months post partum, but episodes have been reported as late as 1 year following delivery. All episodes of thyrotoxicosis resolve spontaneously. The hypothyroid phase of PPT occurs from 3 to 12 months postpartum with 10%–20% of cases resulting in permanent hypothyroidism. It should be noted, however, that a prospective study reported that 50% of women with PPT remained hypothyroid at the end of the first postpartum year (549).

Question 98: What is the etiology of PPT?

Postpartum thyroiditis is an autoimmune disorder associated with the presence of thyroid antibodies (TPOAb and TgAb), lymphocyte abnormalities, complement activation, increased levels of IgG1, increased NK cell activity, and specific HLA haplotypes (550 –552). Women who are thyroid Ab positive in the first trimester have a high risk of developing PPT, ranging from 33% to 50% (553). Women with the highest Ab titers also have the highest risk of PPT (431). The occurrence of PPT reflects the rebound of the immune system in the postpartum period after the relative immune suppression of pregnancy.

Question 99: How should the etiology of new thyrotoxicosis be determined in the postpartum period?

The major challenge is to differentiate thyrotoxicosis caused by PPT from thyrotoxicosis caused by GD. This is an important distinction because the two disease entities require different treatments and have markedly different clinical courses. The timing of onset provides some clues about etiology. In a Japanese hospital study of 42 patients who developed thyrotoxicosis within the first year after pregnancy, 86% of patients who had debut of disease within the first 3 months after delivery had thyroiditis, whereas all who developed thyrotoxicosis after 6.5 months had GD (554).

TSH receptor antibodies are positive in GD in nearly all cases and are typically negative in PPT, although some mixed-type disease is seen. An elevated T4:T3 ratio suggests the presence of PPT. Physical stigmata of GD, such as goiter with a bruit or ophthalmopathy, are diagnostic when present. The radioiodine uptake is elevated or normal in GD and low in the thyrotoxic phase of PPT, but the use of radioactive diagnostic procedures in lactating patients is rarely indicated. Due to their short half-lives, ¹²³I or technetium scans (Tc-99m) may be used in women who are breastfeeding if breast milk is pumped and discarded for several days after the scan pending the isotope used (see Section XI). As described in Section XI, the use of ¹³¹I is contraindicated in women who are breastfeeding.

Question 100: What is the prevalence of PPT?

The prevalence of PPT is approximately 5%, although it has varied markedly in different studies, with frequencies reported from 1.1% to 16.7% (555). Women with other autoimmune disorders have an increased risk of PPT. Specifically, the prevalence of PPT is 3–4 times higher in women with diabetes mellitus type 1 compared to unselected populations (556,557). The frequency of PPT is 25% in women with chronic viral hepatitis (558), 14% in women with systemic lupus erythematosus (559), 44% in women with a prior history of GD (560), and 27% in patients with antipituitary antibodies (561). Individuals who recover fully from PPT have a 70% chance of developing PPT in each subsequent pregnancy (562). Women on LT4 therapy secondary to Hashimoto's thyroiditis may develop PPT if their thyroid gland is not completely atrophic (563). Cases of PPT have been reported following pregnancy loss, but the prevalence of PPT following pregnancy loss is unknown (564).

Question 101: What symptoms are associated with PPT?

Postpartum thyroiditis is a painless condition and most women are asymptomatic or only mildly symptomatic during the thyrotoxic phase. This is because the degree of increase in thyroid hormones is typically mild, and T4 levels are usually more elevated than T3. Nevertheless, in prospective studies, reported symptoms include irritability, heat intolerance, fatigue, and palpitations (431,565 –567). The hypothyroid phase of PPT is more frequently symptomatic. Symptoms during the hypothyroid phase of PPT may include cold intolerance, dry skin, fatigue, impaired concentration, and paresthesias (566,567). One study demonstrated that patients with TPO antibodies and PPT had more symptoms than those who were TPOAb negative (568).

Question 102: Is PPT associated with depression?

Studies evaluating the relationship of PPT to postpartum depression have been inconsistent (569 –573). Two studies have reported a significant association between the presence of thyroid antibodies and depression (568,574), irrespective of thyroid function, whereas another study showed no association between the presence of microsomal antibodies and postpartum depression (575). A prospective trial of LT4 treatment versus placebo in postpartum TPOAb-positive women resulted in no difference in rates of postpartum depression between the two groups (576).

■ RECOMMENDATION 85

 All patients with depression, including postpartum depression, should be screened for thyroid dysfunction.

 Strong recommendation, low-quality evidence.

Question 103: What is the treatment for the thyrotoxic phase of PPT?

There have been no prospective studies evaluating when and how to treat PPT. Treatment of the thyrotoxic phase is guided by its transitory nature. ATDs (PTU and MMI) are ineffective in treating the thyrotoxic phase of PPT because it is a destructive thyroiditis in which thyroid hormone synthesis is not increased. Symptoms are typically mild. In rare cases when symptoms are clinically significant, propranolol at the lowest possible dose to alleviate symptoms may be used. The thyrotoxic phase of PPT must be differentiated from recurrent or de novo GD.

■ RECOMMENDATION 86

 During the thyrotoxic phase of PPT, symptomatic women may be treated with β-blockers. A β-blocker that is safe for lactating women, such as propranolol or metoprolol, at the lowest possible dose to alleviate symptoms is the treatment of choice. Therapy is typically required for a few weeks.

 Strong recommendation, moderate-quality evidence.

■ RECOMMENDATION 87

 ATDs are not recommended for the treatment of the thyrotoxic phase of PPT.

 Strong recommendation, high-quality evidence.

Question 104: Once the thyrotoxic phase of PPT resolves, how often should TSH be measured to screen for the hypothyroid phase?

■ RECOMMENDATION 88

 Following the resolution of the thyrotoxic phase of PPT, serum TSH should be measured in approximately 4–8 weeks (or if new symptoms develop) to screen for the hypothyroid phase.

 Strong recommendation, high-quality evidence.

Question 105: What is the treatment for the hypothyroid phase of PPT?

In women with significant symptoms, those currently lactating or women who are actively attempting pregnancy, treatment should be started (431). LT4 treatment should be considered during the hypothyroid phase of PPT if the patient is mildly symptomatic, or recommended if the patient is considering another conception. If treatment is delayed, thyroid function should be rechecked every 4–8 weeks until euthyroidism is restored and women should be counseled to use contraception.

■ RECOMMENDATION 89

 LT4 should be considered for women with symptomatic hypothyroidism due to PPT. If treatment is not initiated, their TSH level should be checked every 4–8 weeks until thyroid function normalizes. LT4 should also be started in hypothyroid women who are attempting pregnancy or who are breastfeeding.

 Weak recommendation, moderate-quality evidence.

Question 106: How long should LT4 be continued once initiated?

The length of time that LT4 should be continued has not been systematically evaluated. Guiding principles are to maintain a euthyroid state in women who are attempting pregnancy or pregnant. Tapering LT4 doses in order to determine whether the hypothyroid phase of PPT was transitory or permanent can begin by 12 months post partum. Tapering should be gradual and TSH should be monitored every 6–8 weeks.

■ RECOMMENDATION 90

 If LT4 is initiated for PPT, discontinuation of therapy should be attempted after 12 months. Tapering of LT4 should be avoided when a woman is actively attempting pregnancy or is pregnant.

 Weak recommendation, low-quality evidence.

Question 107: How often should thyroid function testing be performed after the hypothyroid phase of PPT resolves?

The impact of PPT on long-term thyroid function has been evaluated in seven studies (549,562,577 –581). These data demonstrate that 10%–50% of women in whom the hypothyroid phase of PPT initially resolves will ultimately go on to develop permanent hypothyroidism. Factors associated with an increased risk of developing permanent hypothyroidism are multiparity, thyroid hypoechogenicity on ultrasound, greater severity of the initial hypothyroidism, higher TPOAb titers, greater maternal age, and a history of pregnancy loss.

■ RECOMMENDATION 91

 Women with a prior history of PPT should have TSH testing annually to evaluate for the development of permanent hypothyroidism.

 Strong recommendation, high-quality evidence.

Question 108: Does treatment of thyroid Ab–positive euthyroid women during pregnancy prevent PPT?

Two randomized placebo-controlled controlled trials have evaluated the efficacy of iodine or LT4 treatment during pregnancy to prevent the development of PPT in thyroid Ab–positive women. Neither intervention decreased the incidence of PPT (154,582). A single trial has suggested a benefit of selenium supplementation for the prevention of PPT in TPOAb-positive pregnant women, though this has not been independently validated. Furthermore, selenium use has been associated with an increased risk of type 2 diabetes (156). Currently there is not sufficient evidence to recommend selenium supplementation during pregnancy in thyroid Ab–positive women (see Recommendation 12).

■ RECOMMENDATION 92

 Treatment of euthyroid thyroid Ab–positive pregnant woman with either LT4 or iodine to prevent PPT is ineffective and is not recommended.

 Strong recommendation, high-quality evidence.

Question 109: Does identification and LT4 treatment of women with subclinical hypothyroidism or thyroid autoimmunity reduce pregnancy complications and pregnancy loss?

Retrospective studies have frequently suggested a positive correlation between treatment of mild elevations of maternal serum TSH and decreased risk of pregnancy loss. For example, a retrospective study from Belgium assessed the effect of TSH screening in a small cohort of pregnant women. LT4 treatment of TPOAb-positive women with TSH >1 mU/L led to a reduction of pregnancy loss rates from 16% to 0% (180). A separate historical cohort analysis studied >1000 pregnant women in the United Kingdom treated with long-term LT4. Pregnancy loss rates were significantly higher when TSH was >4.5 mU/L (and especially when >10 mU/L) compared with when TSH was <2.5 mU/L (318). The retrospective design and lack of TPOAb data are two important limitations of this study.

Limited RCT data on the effects of treatment of subclinical hypothyroidism in pregnancy are available to date. Negro and colleagues (28) randomized euthyroid (TSH <4.2 mU/L) TPOAb-positive women to treatment with LT4 or no treatment. This intervention demonstrated a significant reduction in preterm deliveries (from 22% to 7%) and miscarriage rates (from 14% to 3.5%). However, limitations of this study include its small sample size and the fact that average time of LT4 initiation was late in the first trimester (mean 10.3 weeks), at a time when many pregnancy losses had already occurred.

Two more recent prospective RCTs have investigated the benefits of universal screening of pregnant women with subsequent LT4 treatment of women with TSH >2.5 mU/L. Negro and colleagues (584) randomized 4562 women to universal screening versus screening of women at high risk for thyroid disease. Women identified to have TPOAb positivity and TSH concentrations >2.5 mU/L were treated with LT4. The timing of treatment initiation was 8.8 weeks gestation. Even though this is well into the first trimester, limiting assessment of the outcome, this is the earliest randomized intervention trial published to date. In the primary analysis, there was no significant difference in adverse events detected between the universally screened cohort versus those who underwent high-risk screening. The main limitation of this study was its design given that all high-risk patients were treated in both study arms, limiting the ability to detect a treatment effect. However, in a secondary analysis of the low-risk cohort, a significant decrease in adverse obstetrical and neonatal outcomes was detected when a screening and treatment strategy was applied (OR 0.43 [CI 0.26–0.70]). How to best interpret this finding, however, remains unclear because this study used a composite endpoint for assessing adverse outcomes. For these reasons, this study alone is often viewed as insufficient to warrant adoption of a universal screening mandate.

A second investigation employed cluster randomization at two centers in China, comparing a process of universal screening at one hospital (and treatment when TSH >2.5 mU/L) to a traditional nonscreening approach in the second institution. In the latter group, serum was obtained from women in early pregnancy, but banked and not analyzed until the postnatal period. The authors concluded that identifying and treating elevated TSH values early in pregnancy reduced pregnancy loss rates (OR 0.343 [CI 0.21–0.56]) and macrosomia (OR 0.46 [CI 0.28–0.74]) (585). However, these data should be viewed with caution because the timing of enrollment at the two centers varied substantially, likely influencing the reporting of pregnancy loss rates. Notably, the screening arm enrolled women at week 11 of pregnancy, whereas the control arm enrolled women at week 7. Because most pregnancy losses occur during the first trimester, the earlier time at testing in the control group suggests that the much higher rates of pregnancy loss were simply due to early enrollment bias.

Importantly, two additional prospective, randomized trials investigating obstetric outcomes are currently in progress. The Thyroid Antibodies and Levothyroxine Trial (586) is a large RCT conducted in the United Kingdom, randomizing TPOAb-positive euthyroid women with a history of pregnancy loss or infertility to treatment with 50 μg LT4 versus placebo. The endpoint is successful delivery of the infant beyond 34 weeks of gestation. Separately, the T4Life trial in the Netherlands is randomizing euthyroid anti-TPOAb–positive women with a history of recurrent pregnancy losses to treatment with LT4 (starting before conception) versus placebo. The primary outcome is live birth after 24 weeks gestation (587).

Question 110: Does identification and LT4 treatment of pregnant women with subclinical hypothyroidism improve neurocognitive outcomes in offspring?

With regards to the endpoint of fetal neurocognitive development, there are mixed findings from currently available studies. Haddow and colleagues (276) first published seminal data from a large case–control study investigating the effects of untreated maternal hypothyroidism on the IQ of offspring tested at ∼8 years. This study analyzed sera from 25,216 women obtained at an average gestational age of 17 weeks. They identified 62 women with either TSH concentrations ≥99.7th percentile, or women with TSH concentrations between the 98 and 99.6th percentiles who also had low FT4 concentrations. The authors compared the IQ of the offspring (at 7–9 years) with that of 124 matched children born to euthyroid mothers. IQ scores were 7 points lower in children born to untreated hypothyroid mothers in comparison to controls. However, the retrospective nature of this study has led to questions regarding its reproducibility.

The largest prospective randomized trial of LT4 therapy investigating the treatment of maternal hypothyroidism or hypothyroxinemia in early pregnancy is the CATS study. This study randomized 22,000 women with singleton pregnancies at <16 weeks gestation to immediate thyroid function testing and LT4 treatment if serum TSH was >97.5th percentile and/or if serum FT4 was <2.5th percentile. The control arm received no treatment, though serum was obtained in early pregnancy and banked until testing was performed postpartum. The primary outcome was offspring IQ at 3 years of age. Ultimately, the trial found no significant difference in IQ between the children of 390 treated mothers compared with those of 404 untreated mothers (306). Limitations of this study include the timing of LT4 administration (median initiation >13 weeks), and the mild degree of TSH elevation (median TSH 3.1–3.8 mU/L) in its subjects. Furthermore, the predictive value of early IQ testing at age 3 years remains uncertain. With these major limitations, the negative primary endpoint nonetheless does not support universal screening for thyroid disease in pregnant women.

Two separate randomized trials evaluating the effect of LT4 treatment upon fetal neurodevelopment are currently in progress. The first is a multicenter, randomized placebo-controlled clinical trial conducted by the National Institutes of Health, which seeks to evaluate the effects of LT4 treatment in women with either subclinical hypothyroidism or isolated hypothyroxinemia on offspring cognition. Preliminary results presented as an oral abstract at the 2016 Annual Meeting of the Society for Maternal Fetal Medicine meeting showed that initiation of LT4 treatment in these women at a mean gestational age of 17 weeks had no effect on child IQ (21). Full results are anticipated in 2017. A separate, large prospective study is also underway in China. The Subclinical Hypothyroid during Early Pregnancy (SHEP) trial will screen 21,500 women planning pregnancy and seek to assess the effects of LT4 treatment initiated before conception in women with iodine deficiency, those with subclinical hypothyroidism, and those positive for TPOAb (588). The results of the study are expected between 2018 and 2020.

Importantly, if a screening paradigm is conclusively proved effective, universal screening of women in the first trimester has been shown to be cost effective (589 –592). A study by Dosiou et al. (590) established the cost-effectiveness of universal screening of pregnant women with both TSH and TPO antibodies in the first trimester compared with a strategy of high-risk screening. At base-case analysis, this study assumed that treatment of identified women with TPOAb positivity and subclinical hypothyroidism reduced pregnancy loss and preterm delivery rates as per the randomized clinical trial of Negro and colleagues (28), and that screening resulted in more prompt diagnosis of PPT and hypothyroidism later in life, in the women who developed these disorders. The study did not assume any adverse effects of untreated thyroid disease on child IQ. In sensitivity analyses, even when the benefits of screening were limited to detection and treatment of overt hypothyroidism, screening was shown to be highly cost-effective at <$8000/quality-adjusted life year (QALY). Assumption of benefit for fetal IQ made screening cost-saving. Another recent study from Spain demonstrated similar findings, with an incremental cost-effectiveness ratio of €374.28/QALY (592). The cost effectiveness of screening for thyroid disease in pregnancy compares well with that for other widely used screening practices in pregnant women such as cystic fibrosis screening ($8290/QALY) (593) and prenatal diabetes screening ($16,689–19,339/QALY) (594).

Finally, it is important to place the screening discussion in the context of worldwide opinion, and acknowledge the recommendations of other Societies. The 2011 ATA pregnancy guidelines and the 2012 Endocrine Society pregnancy guidelines significantly expanded the definition of “high risk” women from that used in earlier editions to include women >30 years of age. This expanded definition of high risk encompassed many more women in populations with a high mean maternal pregnancy age. In a 2012 study by Potlukova et al. (595), with a mean maternal age at pregnancy of 31 years, the addition of age 30 or older as a risk factor increased the proportion of women correctly identified in a case-finding strategy from 55.3% to 85.6%. However, in a recent Chinese population with mean maternal age at pregnancy of 26.6 years, testing using only high-risk criteria missed 82.4% of women with subclinical hypothyroidism, 28.6% of women with overt hyperthyroidism, and 74.6% of women with antithyroid antibodies (596). Member surveys of professional societies have shown that 42.7% of responders in Latin America and 43% in Europe perform universal screening (597,598), whereas only 21% of the Asia-Oceania Thyroid Association members do so (599), and 74% of ATA members support such an approach (600). In the recent 2014 ETA guidelines, the majority of authors recommended universal screening because of the beneficial effects of treatment for overt hypothyroidism, and the fact that the targeted approach will miss a large percentage of women with subclinical hypothyroidism (601). The Spanish Society of Endocrinology and Nutrition (602) and the Indian Thyroid Society (603) have expressed support for universal screening in early pregnancy or preconception. The Indian National Guidelines recommend testing only high-risk women (604). Finally, the American Society for Reproductive Medicine recommends TSH testing in all infertile women attempting pregnancy and in high-risk women in early pregnancy (605).

Together, these data provide a perplexing dilemma. Studies strongly suggest an increase in pregnancy loss risk associated with elevated maternal TSH concentrations, especially when elevated TPOAb are detected. Similarly, thyroid dysfunction is a prevalent condition that can be diagnosed with readily available and inexpensive tests. However, the effectiveness of LT4 therapy has not yet been conclusively demonstrated. Importantly, many have argued that screening for thyroid dysfunction must occur very early in pregnancy (e.g., 4–7 weeks of gestation) to maximize potential benefits of LT4 treatment upon pregnancy loss rates and possibly neurocognitive development. The largest prospective screening studies thus far have provided data most translatable to typical pregnancy care currently provided worldwide, with initial evaluation between 10 and 15 weeks of gestation. This is important to consider because the feasibility of any screening earlier in gestation is unclear.

Further uncertainty surrounds the choice of biochemical test to use for any screening mandate. Most studies identify two dichotomous hypothyroid populations detected in early pregnancy—those with elevated maternal TSH concentrations, but separately from those with decreased FT4 concentrations. Such populations rarely overlap. These findings suggest that the most effective screening strategies may require multimodal testing beyond just TSH evaluation. And, as mentioned above, detection of elevated TPOAb concentrations may ultimately prove the strongest risk factor for adverse outcomes, though no RCT to date has used this as the sole initial testing approach. Furthermore, the only study to date to target treatment of pregnant women with low FT4 concentrations demonstrated no benefit (306).

Therefore, while acknowledging an impressive amount of retrospective data associating thyroid dysfunction with pregnancy harm, the above uncertainties preclude the task force from recommending for or against a universal screening mandate. The outcome of future studies will prove critical for expanding our knowledge, and the task force acknowledged the wealth of data which will become available in the years ahead. In coming to this conclusion, the task force noted that the majority of patients identified through any universal screening process have TSH concentrations between 2.5 and 5.0 mU/L—a population in whom a treatment benefit is not well established. Furthermore, such a strategy could have detrimental effects, labeling many patients with a biochemical abnormality, and in many cases leading to initiation of possibly inappropriate long-term treatment.

One task force member (CD) dissented from this recommendation, feeling that universal testing for maternal TSH and anti-TPO antibodies soon after pregnancy confirmation is warranted given the existing support for the obstetrical benefits of treatment, with minimal risk of harm with appropriate monitoring. There might be an additional benefit of treatment on fetal neurodevelopment, as shown in animal studies. Unfortunately, the timing of initiation of LT4 treatment after completion of the first trimester in the RCTs that examine child IQ as the outcome is a very serious design flaw that biases these studies to producing negative results, since they miss the relevant window of opportunity to influence neurodevelopment. Nonetheless, universal screening would allow identification and treatment of all women with overt hypothyroidism, which is prevalent in 0.3%–0.5% of pregnant women (259,260,583) and asymptomatic in about 70% of patients (606). It would also allow identification and selective treatment of women with autoimmune subclinical hypothyroidism, and appropriate close monitoring of euthyroid women with isolated thyroid autoimmunity, some of whom will develop hypothyroidism during gestation or PPT.

The task force uniformly recommends that health care providers identify all newly pregnant women at high risk for thyroid disease. This would include women with a history of thyroid dysfunction, symptoms or signs of thyroid dysfunction, presence of a goiter, and known thyroid antibody positivity. Other risk factors for thyroid disease include age >30 years (145,607), history of diabetes mellitus type 1 (608) or other autoimmune disorders (609), history of pregnancy loss, preterm delivery or infertility (206,208), history of head or neck radiation (610) or prior thyroid surgery (611,612), family history of autoimmune thyroid disease or thyroid dysfunction (613,614), morbid obesity (615,616), use of amiodarone (617), lithium (392), or recent administration of iodinated radiologic contrast (618), two or more prior pregnancies (619), and residing in area of moderate to severe iodine deficiency (620). In these women, measurement of serum TSH concentration should be performed as soon as pregnancy is confirmed, with reflex anti-TPOAb if TSH is 2.5–10 mU/L (Fig. 1).

OPEN IN VIEWER

FIG. 1.

Testing for thyroid dysfunction in pregnancy. ULRR, upper limit of the reference range.

While screening women for thyroid disease preconception may also prove beneficial, there are currently no data to support such an approach, and the process of testing such a high volume of women, the majority of whom will not become pregnant, seems impractical. Universal screening for TPOAb in early pregnancy or possibly preconception may also prove an attractive alternative, but it warrants further investigation. The high prevalence of anti-TPO positivity (up to 17% in reproductive age women), the extensive findings demonstrating increased risks in the anti-TPOAb–positive population, and the fact that this test would also identify women at risk for developing hypothyroidism during gestation (20%) (147) and PPT (30–50%) (621), make this test an attractive theoretical consideration. However, no data support this testing algorithm at present.

Question 111: Should women be universally tested for thyroid function before or during pregnancy?

■ RECOMMENDATION 93

 There is insufficient evidence to recommend for or against universal screening for abnormal TSH concentrations in early pregnancy.

 No recommendation, insufficient evidence.

■ RECOMMENDATION 94

 There is insufficient evidence to recommend for or against universal screening for abnormal TSH concentrations preconception, with the exception of women planning assisted reproduction or those known to have TPOAb positivity.

 No recommendation, insufficient evidence.

■ RECOMMENDATION 95

 Universal screening to detect low FT4 concentrations in pregnant women is not recommended.

 Weak recommendation, moderate-quality evidence.

■ RECOMMENDATION 96

 All pregnant women should be verbally screened at the initial prenatal visit for any history of thyroid dysfunction, and prior or current use of either thyroid hormone (LT4) or antithyroid medications (MMI, CM, or PTU).

 Strong recommendation, high-quality evidence.

■ RECOMMENDATION 97

 All patients seeking pregnancy, or newly pregnant, should undergo clinical evaluation. If any of the following risk factors are identified, testing for serum TSH is recommended:

1. A history of hypothyroidism/hyperthyroidism or current symptoms/signs of thyroid dysfunction

 Strong recommendation, moderate-quality evidence.