Thyroid-stimulating hormone, anti-thyroid antibodies, and pregnancy outcomes.

Thyroid disease complicates approximately 4% of all pregnancies. 1 Overt hypothyroidism has been linked to various pregnancy complications such as preeclampsia, gestational diabetes (GDM) and preterm delivery (PTD). 2 , 3 These adverse pregnancy outcomes contribute significant burden on families and the health care system and have important implications for the future health of the child. However, it is unclear whether less severe forms of thyroid disease, specifically, subclinical hypothyroidism (SCH), are also linked to obstetric complications. Subclinical hypothyroidism is defined as an elevated thyroid stimulating hormone (TSH) with normal thyroxine (fT4), 4 and is the most common form of thyroid dysfunction in pregnancy. 5 Though several studies have evaluated the relationship between SCH in pregnancy and various adverse pregnancy outcomes, results have been conflicting. 5 – 9 Previous studies have been limited by assessment of thyroid function during early pregnancy, as opposed to preconception, when alterations to TSH can occur secondary to the presence of human chorionic gonadotropin and other hormone changes. 10 Thyroid autoimmunity, characterized by the presence of thyroid auto-antibodies, is also common and has been variably associated with adverse pregnancy outcomes. 5 , 11 In 2011, the American Thyroid Association recommended a TSH level of <2.5mIU/L as ideal in early pregnancy. 1 None of the studies referenced above, however, examined obstetrical outcomes in women with TSH of <2.5 versus ≥ 2.5 mIU/L. Furthermore, studies evaluating the relationship between preconception thyroid levels and pregnancy outcomes are lacking. Thus, our objective was to determine the association between pre-pregnancy anti-thyroid antibodies, SCH, and adverse obstetrical outcomes including PTD, preeclampsia, and GDM.

SCH and thyroid antibodies were not associated with adverse pregnancy complications among healthy fecund women with an ongoing pregnancy >20 weeks estimated gestational age. Furthermore, among women with thyroid autoimmunity, there were no differences in PTD, preeclampsia, or GDM even when stratifying by TSH level. These results support current recommendations that low-risk asymptomatic women should not be routinely screened for thyroid dysfunction or autoimmunity. Previous research on SCH has explored a variety of adverse pregnancy outcomes with inconsistent results. In part this may be due to varied definitions of SCH among studies. Our results for PTD are in agreement with findings from a recent systematic review and meta-analysis 18 that reported no relationship between PTD and SCH, though an increased risk of PTD was noted among women with overt hypothyroidism in that study. Though available data regarding SCH and preeclampsia is mixed, our results are consistent with a prior retrospective study and meta-analysis that also showed no difference in rates of preeclampsia 6 or gestational hypertension in relation to SCH. 5 Though an association between SCH and severe preeclampsia (but not gestational hypertension or mild/moderate preeclampsia) was reported in one large study, it is of note that SCH was defined as TSH > 4.13; 8 as such the results are not directly comparable to our study. Our findings also differ from those of a large Chinese cohort that observed a link between SCH and gestational hypertension. 7 However, their results were not adjusted for BMI which may have confounded their findings and they did not examine preeclampsia as an outcome 7 . In contrast, overt hypothyroidism has been associated with severe preeclampsia 19 – 21 , and adverse cardiovascular effects, 20 , 22 , 23 which in turn may be associated with vascular-related pregnancy complications. Our findings are also in agreement with several investigations of SCH and GDM. Many groups found no association between SCH and GDM, including two recent cohort studies. 7 , 9 A meta-analysis examining SCH and pregnancy outcomes also did not detect a link between GDM and SCH (OR 1.38; 95% CI 0.97, 1.96). 5 Here, we extend these findings confirming no relationship between SCH and GDM, even with thyroid function measured prior to pregnancy. Studies regarding thyroid autoimmunity and pregnancy complications also have yielded mixed results. In this study, we found no difference in PTD, GDM, or preeclampsia among women with thyroid autoimmunity, even among those with an additional finding of subclinical hypothyroidism. Our results are consistent with several groups who also found no relationship between thyroid antibodies and GDM and preeclampsia; however, these studies found a positive relationship between these antibodies and PTD 11 , 24 – 26 which is in contrast to our study. Of note, in one of these studies an association was only observed with the outcome of very preterm birth (<34 weeks) and no adjustments for potential confounders were made. 25 However, anti-TPO positivity was associated with PTD in other studies, including a recent meta-analysis, that adjusted for relevant confounding factors 11 , 24 , 26 though some of the prior studies assessed for the presence of only TPO antibodies or used differing definitions of positive antibody status. These differences in study design may account for some of the differences in findings among studies. This study has several strengths. First, the original trial was prospective, allowing for preconception assessment of fT4, TSH and thyroid antibody levels, and thus the assessment of thyroid function was not affected by changes in thyroid function during pregnancy. An important difference between previous work and the current study is our assessment of thyroid function prior to pregnancy. As thyroid function is known to change during early pregnancy, this could have important implications for interpreting these findings. 10 Furthermore, we were able to evaluate the clinical utility of using a cut point of a TSH ≥ 2.5 mIU/L. Our results extend previous work by utilizing preconception measurements and exploring the association with a more strict and more common definition of SCH. We were also able to adjust for potential confounders in our analyses. Moreover, all outcomes were assessed prospectively and our population was comprised of young, healthy women with a history of pregnancy loss. As such, these results are generalizable to a large population of women. Additionally, the original trial had a high compliance rate, with the majority of women completing study follow-up (88%). There are some limitations to our study as well. We had a small number of adverse outcomes making our ability to detect small differences somewhat limited. Additionally, the majority of this population consisted of Caucasian, middle-class, well-educated women, which may limit generalizability. In conclusion, we found no relationship between PTD, GDM, or preeclampsia and preconception SCH and thyroid autoimmunity, defined as presence of either anti-TPO or anti-TG antibodies, among women with proven fecundity. To date, the American College of Obstetricians and Gynecologist does not recommend screening of TSH or anti-thyroid antibodies in low risk, asymptomatic patients, and our data support this guideline. These findings can offer reassurance to women with subclinical hypothyroidism or thyroid autoimmunity and their healthcare providers.

There were no differences in age, race, and education level in participants with TSH <2.5 compared to those with a TSH ≥ 2.5. Women with a TSH level ≥ 2.5 had a higher BMI and were more often nulliparous compared to women with lower TSH [ Table 1 ]. Women with positive tests for anti-thyroid antibodies were similar in age and BMI compared to those testing negative. Of the 1018 women without thyroid antibodies 962 were non-Hispanic white (94.5%), compared to 98.2% non-Hispanic white among women with thyroid antibodies (p = 0.04). There was no difference in prior number of live births or pregnancies between groups. Among women who became pregnant during the study with an ongoing pregnancy > 20 weeks, there was no increased risk of PTD (RR 078; 95% CI 0.41, 1.48), GDM (OR 1.27; 95% CI 0.50, 3.20) or preeclampsia (OR 1.16; 95% CI 0.63, 2.15) in women with TSH <2.5 versus those with TSH ≥ 2.5 either before or after adjustment for potential confounders [ Table 2 ]. Similarly, anti-thyroid antibodies also were not associated with any of these adverse obstetric outcomes. We further explored if any differences were noted when stratifying by indication for preterm delivery (spontaneous versus medically-indicated), and found no differences. We also evaluated obstetric outcomes in the cohort of women with both SCH (TSH ≥ 2.5 mIU/L) and thyroid autoimmunity (n= 45), since this cohort may be at elevated risk. However, the combination of these two assay results was not associated with PTD (OR 0.77; 95% CI 0.19, 3.13), GDM (RR 3.24; 95% CI 0.30, 34.55) or preeclampsia (RR 1.53, 95% CI 0.38, 6.09) [ Table 3 ].

This was a secondary prospective cohort analysis from the Effects of Aspirin in Gestation and Reproduction (EAGeR) trial. The EAGeR trial was a multi-center, double-blind, randomized, placebo-controlled trial that examined the effect of low dose aspirin on live birth. 12 , 13 Women (n=1228) with a history of one to two previous pregnancy losses and attempting pregnancy were recruited from four U.S. medical centers 2007–2011. A detailed description of the study design and methods has been described previously. 12 Institutional Review Board (IRB) authorization was obtained at the data coordinating center and at all clinical centers and each participant provided written informed consent. Patient safety was monitored by a Data Safety and Monitoring Board and the trial was registered with ClinicalTrials.gov, number NCT00467363 . Participants were women aged 18–40 years, with regular menstrual cycles (21–42 days in length) who were actively attempting to conceive. Although they had a history of one or two confirmed prior pregnancy losses, they did not have diagnosed infertility, pelvic inflammatory disease, tubal occlusion, endometriosis, anovulation, uterine abnormality, or polycystic ovarian syndrome, or any major medical disorder. 12 Women in the study were followed through six cycles attempting pregnancy. If they did not conceive in six cycles or experienced a second periconception loss during the study, their study participation was discontinued. If they became pregnant, they were followed prospectively throughout the pregnancy and delivery. Fertility monitors (Clearblue Easy Fertility Monitor; Inverness Medical, Waltham, MA) were used to assist with scheduling study visits and timing intercourse. Preconception TSH, fT4, thyroglobulin antibody (anti-TG), and thyroid peroxidase antibody (anti-TPO) levels were measured in serum collected at a baseline visit prior to conception. Samples were stored at −80°C after collection until the time of assay. TSH was measured via the TSH reagent sandwich immunoassay method (Roche Diagnostics, Indianapolis, IN). The reference range was 0.4 – 5 mIU/L, and the interassay laboratory coefficients of variation (CVs) were 2.1% at 1.596 mIU/L and 2.9% at 9.037 mIU/L. fT4 was measured using a fT4 reagent/competitive immunoassay (Roche Diagnostics, Indianapolis, IN) with a reference range 0.7 – 1.85 ng/dL. Anti-TG levels were measured with a TSH reagent sandwich immunoassay (Roche Diagnostics, Indianapolis, IN) and the interassay CVs were 7.2% at 91.4 IU/mL and 6.7% at 171 IU/mL. Anti-TPO levels were measured using an anti-TPO reagent competitive immunoassay (Roche Diagnostics, Indianapolis, IN) and the interassay CVs were 17% at 31.5 IU/mL and 11.9% at 76.13 IU/mL. Results for anti-TG were considered positive if anti-TG ≥ 115 IU/mL and results were considered positive if anti-TPO ≥ 35 IU/mL, according to the reference ranges at our laboratory. Reproductive, medical, and obstetric history was obtained at baseline via questionnaire and from medical record abstraction. A clinically confirmed pregnancy was defined as evidence of a continuing intrauterine pregnancy on ultrasound at 6–7 weeks’ gestation. Gestational age was determined by an ultrasound conducted in early pregnancy (mean 6.9 weeks’ gestation) for 97% (697/720) of clinically confirmed pregnancies among women who completed the trial; for the remaining 3% (23/720) of pregnancies, gestational age was determined using menstrual cycle dating from home-based fertility monitors provided by the study. The dates and details of pregnancy outcomes were assessed by post-partum phone interview and through medical record abstraction by trained research staff. Primary outcomes for this analysis included PTD, GDM, and preeclampsia. PTD was defined as delivery between 20 weeks and zero days and 36 weeks and six days’ gestation. Cases of PTD were prospectively identified during the study and had further review of abstracted medical records by a maternal-fetal medicine physician to vet and categorize the outcome as spontaneous, medically indicated, or a preterm birth of uncertain indication. Spontaneous preterm birth was defined as any preterm birth preceded by spontaneous labor (cervical change or 4 cm or greater cervical dilation in the presence of contractions), preterm premature rupture of membranes, or both. Medically indicated preterm birth was defined as any preterm birth not classified as spontaneous preterm birth for which at least one medical indication for delivery was noted in the medical record. In this cohort of patients, the major indication for medical delivery was preeclampsia. The overall number of patients with medically indicated preterm birth was small (N=28). Of these, 11 (39%) were related to preeclampsia. Other indications included placenta previa/abruption, preterm labor, chorioamnionitis, and non-reassuring fetal heart rate tracing. The remaining preterm births were categorized as a preterm birth of uncertain indication. Physicians providing prenatal care to patients made a diagnosis of preeclampsia or GDM based on standard clinical and laboratory criteria. Trained EAGeR research staff abstracted these diagnoses from participants’ delivery records. Women with an abnormal fT4 (n=12) or whose fT4 was not recorded (n=23) were excluded. The remaining participants were categorized into two groups based on TSH level: TSH <2.5 or ≥ 2.5 mIU/L. Though subclinical hypothyroidism is most often defined as TSH above the normal range (4.5–5.0 mIU/L) with a normal fT4, 14 the National Academy of Clinical Biochemistry (NACB) found that 95% of patients without any symptoms of thyroid dysfunction actually had a TSH below 2.5 mIU/L. We also evaluated participants for the presence or absence of anti-TG and/or anti-TPO. If women were positive for at least one antibody, they were included in the thyroid autoimmunity group. Descriptive statistics were used to compare characteristics among groups using Fisher’s exact test and t-tests where appropriate. Risk ratios (RR) and 95% confidence intervals (CI) for PTD, GDM, and preeclampsia by TSH level (<2.5 or ≥ 2.5 mIU/L) and by thyroid autoimmunity status were estimated using log-binomial regression adjusted for age and body mass index (BMI). We also evaluated pregnancy outcomes in women with thyroid autoimmunity stratified by TSH level. As pregnancy outcomes are conditional upon becoming pregnant, we restricted our analysis to women who had an ongoing pregnancy > 20 weeks estimated gestational age. As such, we utilized inverse probability weights to control for potential selection bias introduced by restricting the analytical cohort since low-dose aspirin treatment was shown to be associated with the probability of pregnancy among women with a single recent loss 13 , 15 , 16 . Of note, low dose aspirin treatment was not associated with GDM, PTD, or preeclampsia, and we also did not observe associations between thyroid exposures and pregnancy loss or fecundity 17 . Weights were based on factors associated with becoming pregnant, including maternal age, parity, marital status, number of prior losses, and treatment assignment. Weighted log-binomial regression was used to estimate RRs and 95% CIs. When evaluating PTD, we also stratified by indication (spontaneous versus medically-indicated). The number of cases was too low to estimate the RR in some models; therefore, odds ratio (OR) is reported in those instances. Analyses were conducted in SAS version 9.4 (SAS Institute, Cary, NC).