September 2007: Volume 33, Number 9
Thyroid Disorders in Pregnancy
New Guidelines Recommend Against Universal Screening, Suggest a New TSH Reference Limit
By Deborah Levenson
Identifying and managing thyroid disorders during pregnancy can be a tricky business because thyroid disorders can adversely affect a fetus, while at the same time being pregnant can spur thyroid disorders in the mother. The most common thyroid problem, hypothyroidism, or a deficiency of thyroxine (T4), is believed to affect 2–5% of pregnant women and has been associated with diminished intellectual capacity and developmental delay in children. In addition, a mounting body of evidence has implicated the disease in miscarriage, preterm delivery, preeclampsia, and placental abruption in mothers. Amid these potentially deleterious outcomes, an ongoing debate exists among laboratorians, obstetricians, and endocrinologists over whether to screen all pregnant women for thyroid stimulating hormone (TSH), which is affected by T4 levels, and thyroid peroxidase antibody (TPOAb), or to test only those with existing disease or at high risk of disease.
Currently, the answer as to which expectant mothers should be screened depends on whom you ask. For example, both the National Academy of Clinical Biochemistry (NACB) and the American Association for Clinical Endocrinology (AACE) recommend universal thyroid testing for pregnant women, while the American College of Obstetricians and Gynecologists (ACOG) recommends testing only those pregnant women with existing thyroid disorders and those at high risk of developing them during pregnancy.
Even endocrinologists disagree, as draft guidelines from The Endocrine Society (ES) show. This expert group’s recommendations, published online in draft in the Journal of Endocrinology and Metabolism (JCEM) in June (doi:10.1210/jc.2007–0141) and published in final form in the August issue (JCEM 2007; 92: s1–s47), do not recommend screening asymptomatic pregnant women. “We’d love to have universal screening, but the evidence did not allow us to make that recommendation,” explained Leslie J. De Groot, MD, Professor of Medicine at Brown University in Providence, R.I. and chair of the ES committee that wrote the guidelines. “We don’t actually know what would be done with screening results. We know what to do if the problem is clinical hypo- or hyperthyroidism, but not if the patient has positive antibodies and normal function. And we’re not certain about what to do for subclinical hypothyroidism.”
Universal Screening: New Studies
Findings reported in two recent papers have added more fuel to the debate over universal screening of pregnant women. One by British researchers concludes based on a single-site study that targeted testing of pregnant women in high-risk groups missed 30% of those who had subclinical disease (JCEM 2007; 92: 203–207). Researchers measured serum TSH and free T4 in 1,560 pregnant women at their first prenatal visit and divided 1,327 subjects tested for TPOAb into high- or low-risk groups based on personal or family history, previous therapy for thyroid disorder, or history of nonthyroidal automimmune disorder to determine whether testing only a high-risk group would pick up most pregnant women with thyroid dysfunction. Using a TSH cutoff of 4.2 mIU/L, they found the prevalence of elevated TSH was 6.8% in the high-risk group, versus 1% in the low-risk group (95% CI, 3.3–12.6).
Others interpret these findings differently, raising basic questions about the role and scope of screening, noted Rhoda Cobin, MD, Clinical Professor of Medicine at The Mount Sinai School of Medicine in New York, N.Y., immediate past president of the American College of Endocrinology and Past President of the AACE. “The thing that worried everyone is that 12 or 30% of the patients with TSH greater than 4.2 mIU/L came from the low-risk group, which would not have been found using targeted case finding. But overall, these women represented only 0.7% [of the study cohort]. So if an obstetrician has 100 pregnant patients, that’s akin to missing fewer than one of them. The question then becomes, should you miss any of them?”
The other study by a team of Italian researchers led by Roberto Negro, MD, concludes that pregnant women with normal TSH levels but who were TPOAb positive, developed impaired thyroid function and that levothyroxine treatment lowered their risk of miscarriage and premature delivery (JCEM 2006; 91: 2587–2591). In this study, researchers followed 984 women, 11.7% with positive TPOAb results, to determine the rate of obstetrical complications in the treated and untreated groups. The groups treated with levothyroxine—including the TPOAb group with normal TSH levels— and the TPOAb-negative control group had similar miscarriage rates, 3.5% and 2.4% respectively, while the group of antibody-positive women who did not receive treatment had a miscarriage rate of 13.8% (95% CI, 1.13–2.25 and 95% CI 2.59–9.48, respectively). The untreated group also had a higher premature delivery rate of 22.4% (95% CI, 1.18–2.34), versus the levothyroxine group’s rate of 7% (95% CI, 1.18–2.34) or the control group’s 8.2% rate (95% CI, 7.93–18.7). Based on these findings, the authors recommend screening women early in pregnancy for TPOAb and thyroid function and administering levothyroxine at least to women who have a TSH value greater than 2.0 mIU/L and/or a high titer of the thyroid antibodies. “The importance of this article is the apparent success rate in improving outcome with levothyroxine therapy in antibody-positive patients, even if they have normal TSH levels,” Cobin commented.
“This study had everyone buzzing because it shows even a small elevation in TSH is a risk factor. Preterm delivery is hugely costly, and this study shows that risks can be mitigated with treatment,” noted laboratorian Carole Spencer, PhD, Professor of Medicine at Keck School of Medicine at University of Southern California in Los Angeles. Spencer helped write the NACB’s 2002 guidelines, Lab Support for the Diagnosis and Monitoring of Thyroid Disease.
However, the ES guidelines call for more research on the effects of levothyroxine treatment, noting that the Italian study isn’t enough to suggest that drug therapy is safe, efficacious, and cost effective. “One study does not make enough data to advise on what to do with millions of women. The problem is that we don’t have adequate studies documenting the effects of treatment. Also, physicians aren’t always sure what that treatment should be,” De Groot explained.
Managing Thyroid Disease
More Recommendations for Prenatal Care
Draft Guidelines from The Endocrine Society on managing thyroid disease in pregnancy, published in the Journal of Endocrinology and Metabolism (2007; 92: s1–s47), make a number of recommendations, which are ranked according to strength of evidence and net benefit with a letter grade A through D. An “I” indicates insufficient evidence to recommend for or against. Overall evidence is ranked as good, fair, or poor. Recommendations also carry GRADES system scores of strong, moderate, low, or very low.
- If a subnormal serum TSH is detected during gestation, distinguish hyperthyroidism from both normal physiology during pregnancy and hypermesis gravidarum. Differentiation of Graves’ disease from gestational thrytoxicosis is supported by evidence of autoimmunity, a goiter, and presence of TSH-receptor antibodies (TRAb). (A, good, strong)
- TRAb freely cross the placenta and can stimulate the fetal thyroid. These antibodies should be measured before pregnancy or by the end of the second trimester in mothers with current Graves’ disease, a history of the condition and treatment with 131I or thyroidectomy, or with a previous neonate with Graves’ disease. Women who have a negative TRAb do not require antithyroid drug therapy (ADT) have a very low risk of fetal or neonatal thyroid dysfunction. (B, fair, strong)
- Measure thyroid function in all patients with hyperemesis gravidarum. (B, poor, moderate)
- Gestational hyperthyroidism with clearly elevated thyroid hormone levels (free T4 above the reference range or total T4>150% of top normal pregnancy value and TSH <0.1 mIU/L) and evidence of hyperthyroidism may require treatment as long as clinically necessary. (I, poor, low)
- Women known to be thyroid peroxidase antibody positive should have a TSH performed at 3 and 6 months postpartum. (A, good, strong)
- Postpartum Screening with TSH is recommended for women with type 1 diabetes mellitus at 3 and 6 months postpartum. (B, fair, strong)
- Women with postpartum depression should be screened for hypo-thyroidism and appropriately treated. (B, fair, moderate)
More Data a Long Way Off
Two large, ongoing multicenter trials are poised to settle the controversy over universal screening because they are expected to provide data about the effects of levothyroxine treatment. However, those results are years away. The Controlled Antenatal Thyroid Screening Study (CATS) in the U.K., led by John Lazarus of Cardiff University Centre for Endocrine and Diabetes Sciences, aims to assess intellectual development of 22,000 3-year-old children of mothers who had free T4 levels below the 2.5th percentile and/or TSH above the 97.5th percentile. A trial sponsored by the National Institute of Child Health and Human Development, led by obstetrician Brian Casey, MD, at University of Texas Southwestern Medical Centers (Dallas) started in October 2006 and is expected to look at the effects of levothyroxine on intellectual function of 1,000 5-year-old children of mothers diagnosed with subclinical hypothyroidism or hypothyroxinemia during the first half of pregnancy. Completion is expected in 2014.
New TSH Reference Range Guidance
While the endocrinologists and obstetricians hammer things out, it’s unlikely that laboratorians will see a change in thyroid test volumes. However, the draft ES guidelines’ suggested upper limit for the TSH reference range for pregnant women and preconception—below 2.5 mIU/L in the first trimester and preconception and 3.0 mIU/L in the second and third trimesters—can help laboratorians identify more hypothyroid women. The ES guidelines also recognize difficulties with free T4 testing in pregnancy, suggest that clinicians consider the less often used total T4 test, and make several other recommendations regarding testing (See Box).
Laboratorians need to pay attention to the new TSH upper reference limit (2.5–3.0 mIU/L), Spencer advised. “TSH is the first parameter to become abnormal. It’s an early warning system,” she emphasized during a presentation at AACC’s Annual Meeting in July, noting that the higher reference limit may cause laboratorians and physicians to miss thyroid problems. The guidelines explain why. “Serum TSH values are negatively correlated with the thyrotripic activity of elevated circulating human chorionic gonaotopic concentrations, particularly in the first trimester. Therefore, by using the classical reference range for serum TSH, 0.4 mIU/mL for the lower limit and 4.0 mIU/mL for the upper limit, one might misdiagnose ‘normal’ women who already have a slight TSH elevation and conversely, one might suspect hyperthyroidism in normal women who have a low serum TSH value,” the draft guidelines advise. They also cite research that showed 28% of singleton pregnancies with a serum TSH greater than two standard deviations above the mean would not have been identified by using the nonpregnant serum TSH range (Obstetrics and Gynecology 2005; 106: 753–757).
Other ES recommendations are repeat testing of thyroid function within 30 to 40 days and T4 replacement for subclinical hypothyroidism, based on research showing long-term neurological effects in children of mothers with the condition. Noting that T4 treatment has been shown to improve obstetrical outcome but not neurological issues, the guidelines explain the recommendation is based on the authors’ opinion that “the potential benefits outweigh the potential risks.”
The ES guidelines also note problems with measuring free T4 during pregnancy. Free T4 tests are useful for distinguishing between overt and subclinical hypothyroidism, depending on whether levels are normal or clearly below normal for gestational age. Reference ranges provided by manufacturers of most free T4 measurement kits have been established using pools of nonpregnant normal sera that aren’t valid during pregnancy because free T4 assays are influenced by serum changes, mainly in thyroxine-binding globulin (TBG) and serum albumin. Recently, various authors have proposed adapting serum-free T4 reference ranges to laboratory-specific or trimester-specific ranges, but currently there is no consensus on this issue, the guidelines note. “Furthermore, current free T4 immunoassays may not in fact measure free T4 but are sensitive to binding proteins and in reality reflect total T4,” Spencer pointed out, citing recent research (Clin Chem 2007; 53: 911–915).
The guidelines will answer some questions about testing, De Groot predicted. “Of particular interest to labs is the question of what’s the proper test to measure thyroid hormone levels in pregnancy,” De Groot noted. “We couldn’t definitively answer that question, although we proposed several interim solutions based on the context in which the test is being used. Sometimes it’s legitimate to measure total thyroxin and use an adjusted normal level. Sometimes it’s okay to use a range of three thyroxin assays at the top normal level for treating hyperthyroidism. It’s sometimes, if not always, appropriate for labs to actually establish a range for use of a test in pregnant women in each trimester.”
In her presentation, Spencer suggested increased use of an older thyroid test, total T4, in pregnant patients. Unlike free T4, total T4’s reference range for nonpregnant patients has remained stable for two decades and depends on no single method, she noted, adding that studies in populations with different iodine intakes and using different methodologists have yielded “remarkably” comparable values. “Total T4 measurement during pregnancy are solid as a rock—predictably 150% of nonpregnant values,” she emphasized.
The range of normal serum total T4 changes during pregnancy because of rapid increases in TBG levels, so a pregnant woman’s total T4 level should be 50% higher than non-pregnant levels, Spencer recommended during her presentation. The nonpregnant woman total T4 range (5–12 μg/dL or 50–150 nmol/dL) should be adapted in the second and third trimester by multiplying this range by 1.5-fold.
Because manufacturers establish reference ranges in nonpregnant populations, it is increasingly difficult for laboratorians to establish reference ranges independently of manufacturers, Spencer observed. Federal regulations including the Health Insurance Portability and Accountability Act and institutional review boards prevent manufacturers from obtaining adequate, clinically defined patient specimens for establishing reference ranges for pregnancy, and manufacturers are increasingly contracting outside the U.S. to acquire unusual specimens to overcome regulatory restrictions, she said. Studies of populations outside the U.S. have different iodine intakes and often use insensitive methods for excluding autoimmune thyroid disease, resulting in reference ranges that are inappropriate for the U.S., Spencer commented.
The Root of the Differences
But what accounts for the differences in the various guidelines? Varied recommendations about universal screening arise from different standards of evidence and consideration of obstetricians’ special concerns, according to Cobin and others. Cobin, who served on the writing committees for both the ES and AACE guidelines, noted that the ES committee, which had input from several thyroid societies and ACOG, “felt that without randomized, controlled trials and cost/benefit analysis, it is premature to recommend universal screening, even though the personal sentiments of all the members were that screening should be a recommendation.”
In contrast, the AACE guidelines committee evaluated evidence differently. “The AACE committee did not feel that randomized, controlled clinical trials proving benefit are necessary to make the recommendation to screen— and therefore to treat if an abnormality is found—and that waiting for their results might put patients at risk. AACE did not look at a cost/benefit analysis, as did ES, and believed that given the risks of undetected hypothyroidism for both mother and child and low risk of harm from therapy, benefit outweighs risk,” Cobin explained.
At this time, ACOG has not signed off on the guidelines, and has declined to comment on its position. Endocrinologists and obstetricians alike emphasized, however, that universal screening is a much more complicated proposition for obstetricians. “Obstetricians see all the pregnant women, while endocrinologists see only those referred for a problem or patients already in their practices with pre-existent thyroid conditions,” Cobin noted. “Obstetricians are much more sensitive than other specialties to how guidelines can be used in malpractice cases. So ACOG holds data to the highest standard.”
In addition to their liability concerns, obstetricians have to weigh many more practical issues involved in universal screening than do endocrinologists, explained Gregory Brent, MD, Professor of Medicine at University of California, Los Angeles and Secretary of the American Thyroid Association. “Obstetricians’ primary concern is that they are correctly interpreting the test and providing the necessary treatment, but liability is also a reality for them. They are stuck with the responsibility to do the test, interpret the results, and deal with the consequences of the outcomes. Plus they are already looking at a whole, long list of screening tests,” Brent added. “It’s important to do these studies [on T4 treatment] because something that seems obviously helpful to patients isn’t always, such as giving estrogen to postmenopausal women.”
Catherine Spong, MD, obstetrician and chief of the Pregnancy and Perinatology Branch at the National Institute of Child Health and Human Development, downplayed the role of liability in obstetricians’ concerns about universal testing. “The questions are: Should women be screened and then, should they be treated? There is no data showing that treatment is beneficial. Pregnancy is a delicate state, and the fetus is exposed to treatment. As an obstetrician, you must carefully weigh the risks and the benefits. Clearly, too much thyroxine is bad. If a woman’s thyroxine levels are normal, is it good to give her extra? I don’t know. Clearly, there’s a need for more data.”
There’s no need to rush to universal screening, Brent maintained. “If all women with known thyroid disease were properly managed during pregnancy, we’d be doing better than we’re doing now. We’re likely not yet meeting even the case-finding standard,” he suggested. “Also, when treating subclinical hypothyroidism, the potential for adverse effects must be considered. The major risk is excess treatment, although it’s a relatively low risk in pregnancy. Still, it’s something to factor into any mandate for screening.”
Some obstetricians worry that universal screening may become commonplace without multiple societies’ recommendations for screening, according to Casey. “We haven’t seen an overwhelming conversion to universal screening in my area, but I’m sure practices are doing it,” he commented. Research shows that in Maine, at least, universal screening is taking hold in a significant number of practices. A study by led by James E. Haddow, MD, of the Foundation for Blood Research in Scarborough, Maine, surveyed 61 Maine prenatal practices and found routine TSH screening in 48% of them. The researchers hypothesized that these practitioners may have been influenced by previous studies of maternal thyroid deficiency.
The current controversy over screening may present a good opportunity for some company to develop a dual home test for TSH and TPOAb, Spencer suggested. “Women are motivated when they are pregnant. If such a test is available in the supermarket and would work as soon as the first missed period, it could be useful over the counter product as long as it has the technical quality and appropriate sensitivity to detect TSH over 2.5 mIU/L. But it’s probably 10 years down the line,” she predicted.