Levothyroxine to increase live births in euthyroid women with thyroid antibodies trying to conceive: the TABLET RCT

Article history

The research reported in this issue of the journal was funded by the EME programme as project number 09/100/10. The contractual start date was in June 2011. The final report began editorial review in June 2018 and was accepted for publication in November 2018. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The EME editors and production house have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the final report document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

none

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Dhillon-Smith et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Clinical background

Miscarriage, the loss of a pregnancy before 24 weeks of gestation, affects one in five women, making it the commonest complication of pregnancy. It substantially affects the physical and psychological well-being of women: research1 shows that the level of distress associated with miscarriage can be equivalent to that of a stillbirth of a term baby.

In addition, preterm birth, the delivery of a baby between 24 and 37 completed weeks of gestation, occurs in 6–10% of pregnancies. Preterm birth is responsible for up to 85% of newborn deaths. 2 Of those who survive, approximately 10% suffer long-term disability. The human cost of preterm birth is, therefore, enormous; the financial cost of preterm birth is estimated at £939M per year in the UK. 2 This includes health-care costs (including neonatal care), education and costs to the parents.

The prevalence of measurable circulating antithyroid autoantibodies to thyroglobulin or thyroid peroxidase (TPO) in women of childbearing age in the developed world is 5–15%; that of overt hypothyroidism is estimated to be 0.3–0.5% and that of subclinical hypothyroidism is estimated to be 2–3%. 3,4 Prevalence rates are similar during pregnancy. 4,5

Pregnancy may trigger progression to a relative hypothyroid state in women with thyroid peroxidase antibodies (TPOAbs). This is because of an increased demand for thyroid hormone during pregnancy and because women with thyroid autoimmune disease are less able to sustain this increased demand.

To understand the relationship between thyroid autoantibodies and adverse outcomes, systematic reviews of the literature were conducted.

Association between thyroid antibodies and miscarriages

A systematic review, published in the British Medical Journal (BMJ), identified 31 studies, including a total of 12,126 women and three reviews. 6 Thirteen studies were in recurrent miscarriage populations, nine were in infertile populations and nine were in unselected or other populations. The quality of the studies was judged to be generally good on the Newcastle–Ottawa Scale,7 with most studies (22/29; 76%) establishing good comparability of the antibody-positive and -negative cohorts. Of the 31 studies, 28 showed a positive association between thyroid antibodies and miscarriage. A meta-analysis of the results from 19 cohort studies demonstrated more than a tripling in the odds of miscarriage in the presence of thyroid antibodies [odds ratio (OR) 3.9, 95% confidence interval (CI) 2.48 to 6.12] (Figure 1). This strong and statistically significant association between thyroid antibodies and miscarriage was observed in all three population subgroups. A ‘dose–response’ relationship between thyroid antibody positivity and the number of miscarriages was observed. There was also a similar magnitude of increased risk of miscarriage in each of the three subpopulations identified.

FIGURE 1.

Association between thyroid autoantibodies and miscarriage. Reproduced with permission from Thangaratinam et al. 6 This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.

Reproduced with permission from Thangaratinam et al. 6 This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.

Association between thyroid antibodies and preterm birth

The same BMJ systematic review6 identified five studies examining the association between thyroid antibodies and preterm birth, including a total of 12,566 women and one review. All five were cohort studies, and all were judged to be of good quality on the Newcastle–Ottawa Scale. 7 All studies showed a positive association between the presence of thyroid antibodies and preterm births. A meta-analysis showed a more than twofold increase in the odds of preterm birth in the presence of thyroid antibodies (OR 2.07, 95% CI 1.17 to 3.68) (Figure 2).

FIGURE 2.

Association between thyroid autoantibodies and preterm births. Reproduced with permission from Thangaratinam et al. 6 This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.

Reproduced with permission from Thangaratinam et al. 6 This is an open-access article distributed under the terms of the Creative Commons Attribution Non-commercial License, which permits use, distribution, and reproduction in any medium, provided the original work is properly cited, the use is non commercial and is otherwise in compliance with the license. See: http://creativecommons.org/licenses/by-nc/2.0/ and http://creativecommons.org/licenses/by-nc/2.0/legalcode.

Effectiveness of levothyroxine treatment

Prior to commencing the Thyroid AntiBodies and LEvoThyroxine (TABLET) trial, only two randomised trials,8,9 comprising a total of 187 women, had examined levothyroxine treatment for women with TPOAbs. Both studies were in euthyroid women with thyroid autoantibodies; one was in unselected women8 and the other in women scheduled to have in vitro fertilisation (IVF) treatment. 9 One study9 used a dose of 1 µg/kg/day of levothyroxine and the other study8 used a titrated dose of levothyroxine. The quality of the studies was deemed to be satisfactory (Jadad Quality Scores of 5/5 and 3/5). Both studies showed a reduction in miscarriage rates (36% and 75% relative reductions) and, when the results were pooled, there was a statistically significant 52% reduction in miscarriages with levothyroxine treatment [relative risk (RR) 0.48, 95% CI 0.25 to 0.92]. One of the two studies reported on preterm birth;8 this study (n = 115) found a 69% reduction in preterm births with levothyroxine treatment (RR 0.31, 95% CI 0.11 to 0.90).

The full search terms used in the BMJ review can be found in Appendix 1. The search was updated to 2018 to include any further trials that had been published on this topic.

Since the initiation of the TABLET trial, a further two randomised trials10,11 have evaluated levothyroxine treatment for thyroid antibodies. Vissenberg et al. 10 have designed a double-blind, randomised controlled trial (T4-Life) evaluating the use of levothyroxine in TPOAb-positive euthyroid women who have had recurrent miscarriages (defined as two or more consecutive losses). This trial is still in the recruitment phase and so the results are not yet available. A further trial by Wang et al. 11 has evaluated euthyroid TPOAb-positive women undergoing IVF treatment. The details of this study, and its findings and interpretation, are reported in Chapter 5.

Levothyroxine is a commonly used drug in obstetric–endocrine clinics (and, indeed, general internal medicine clinics), and has a well-established safety profile. The three randomised studies8,9,11 did not find any safety concerns for the mother or the baby. Specifically, there were no instances of hyperthyroidism (from overtreatment with levothyroxine). However, as the total number of women across all three trials was 787 and follow-up was only to the end of pregnancy, these trials would not have been suitable for assessing rare or long-term adverse events (AEs). We therefore carried out a literature search to identify studies of potential harm of levothyroxine treatment in pregnancy by using medical subject heading (MeSH) terms and keywords to capture AEs and combined this with search terms to capture levothyroxine and pregnancy studies. This safety review identified 1026 studies, of which 191 were reviews. Most studies evaluated the use of levothyroxine in hypothyroid pregnant women, and found no clear or consistent evidence of serious adverse effects on the mother or the baby, provided there was appropriate monitoring and dose titration. 12,13 A comprehensive literature review, which was interpreted and graded by an international panel of endocrinologists, found that the potential risk of treating subclinical hypothyroidism with levothyroxine was limited to the development of subclinical hyperthyroidism. 14 Although this review may not directly apply to the euthyroid population, the absence of any serious side effects in this review provides reassurance on the safety of levothyroxine, particularly at the proposed dose of 50 µg per day.

The pathophysiological consequences of thyroid antibodies

The exact mechanisms to explain the observed associations between thyroid antibodies and miscarriages or preterm birth are largely unknown. Two mechanisms have been postulated:

1. It has been suggested that the presence of thyroid antibodies may reflect a generalised activation of the immune system and specifically, a dysregulated activity of the immune system at the fetal–maternal interface. The presence of TPOAbs in several non-thyroidal autoimmune diseases supports this hypothesis of global immune dysfunction. 15 Furthermore, there is evidence that there is an alteration in cytokine expression by peripheral T-lymphocytes in TPO-positive individuals outside pregnancy. 16

2. Alternatively, the presence of thyroid antibodies in euthyroid women could be associated with a subtle deficiency in thyroid hormone availability (a fall in circulating free thyroid hormones within the reference ranges) or a lower capacity of the thyroid gland to adequately rise to the increased demand for augmented synthesis of thyroid hormones required in pregnancy. Indeed, the mean serum thyroid-stimulating hormone (TSH) values, although being within normal range, are significantly higher in thyroid antibody-positive women than in women without thyroid antibodies (TSH in TPO-positive women, 2.14 ± 0.84 mlU/l; TSH in TPO-negative women, 1.33 ± 0.32 mlU/l). 17

How may levothyroxine alter the pathophysiology?

Higher concentrations of thyroid hormones within the normal reference range can directly enhance innate and adaptive immunity in normal, healthy individuals. 18 Pregnancy is an inflammatory process involving a shift in the regulation of cytokine networks within the local placental–decidual environment. Dysregulation of local inflammatory processes may be associated with miscarriage and premature delivery. 19 The main regulators of inflammation within the decidua are a whole host of cells of ‘bone marrow lineage’. 20 In particular, uterine natural killer cells, which are a major source of angiogenic growth factors and cytokines, have been shown to regulate vascular remodelling. 21 Thyroid hormones can potentially influence (1) angiogenic growth factor and cytokine production,22,23 as well as (2) trophoblast proliferation, survival and invasion. 24,25 Thus, thyroid hormones may influence the maternal immune regulation both in general and at the fetal–maternal interface, as well as specifically affect trophoblast and decidual cell behaviour.

Aims and objectives

The primary aim of the TABLET trial was to test the hypothesis that, in euthyroid women with TPOAbs, levothyroxine (a dose 50 µg taken orally once daily), started preconceptually and continued to the end of pregnancy, increases the proportion of women who attain a live birth at or beyond 34 completed weeks of gestation by at least 10% compared with placebo.

The secondary aims were to:

• Test the hypothesis that levothyroxine improves secondary outcomes such as ongoing pregnancy at 12 weeks, gestation at delivery and survival at 28 days of neonatal life (the full list of outcomes is given in Chapter 2).

• Explore subgroup effects of levothyroxine in prognostic subgroups (including maternal age, number of previous miscarriages, initial serum TSH concentration and women who were having infertility treatment).

• Test the hypothesis that levothyroxine, compared with placebo, does not incur substantial adverse effects to the mother or the neonate.

The parallel mechanistic study had specific aims, which are detailed in Chapter 4.

Trial design

The TABLET trial was a randomised, double-blind, placebo-controlled multicentre trial of levothyroxine in euthyroid women with TPOAbs, conducted to determine if levothyroxine can reduce miscarriage and premature births in women. The trial had a favourable ethics opinion from the South West 3 Multicentre Research Ethics Committee (reference number 11/SW/0036).

Eligibility (inclusion and exclusion)

Participants were recruited from early pregnancy units, recurrent miscarriage clinics and infertility clinics in the participating NHS hospitals across the UK. Participants had to meet the following eligibility criteria (see Recruitment for more details on the recruitment process):

• women trying to conceive

• a history of one or more miscarriage(s) or primary or secondary infertility

• aged 16–40 years at randomisation

• biochemically euthyroid [TSH 0.44–3.63 mlU/l; free thyroxine 4 (T4) 10.0–21.0 pmol/l using the appropriate analyser]

• TPOAb positive according to local laboratory reference ranges

• willing and able to give informed consent.

Participants could not be included if any of the following criteria were applicable:

• current treatment for any thyroid disorder [past treatment was considered on an individual basis (see below)]

• taking amiodarone or lithium therapy

• contraindications to levothyroxine therapy – thyrotoxicosis, hypersensitivity to levothyroxine, or any of its excipients

• participation in any other blinded, placebo-controlled trial of investigational medicinal products (IMPs) in pregnancy

• previous or current diagnosis of cardiac disease.

Women who had previously been treated for thyroid disorders were considered on a case-by-case basis. It was left to the discretion of the principal investigator whether or not a woman with a history of thyroid disorder could be safely offered participation in the trial. The rationale for exercising this discretion was because it was agreed that women who may have received very short-term treatment a long time ago and whom have since had normal thyroid function and not required treatment long term should not be automatically discounted from participation in the trial. These women would have been classified as ‘euthyroid’ for some time and it was not deemed clinically unsafe for these women to participate if they were TPO positive.

Recruitment

The TABLET trial recruited from three main populations: those with a history of one or two miscarriages, those with recurrent miscarriage (defined as three or more consecutive losses) and those under investigation or treatment for infertility. Recruitment was via a two-step process. Women were initially invited to be screened for TPOAbs and thyroid function tests (TFTs) and then those who were found to be positive for TPOAbs, with normal thyroid function, were introduced to the TABLET randomised controlled trial. Further details are given in the following sections.

Screening of potential participants

Potential participants were identified and approached by clinic doctors, nurses and research staff. The routes of initial approach for screening are shown in a flow diagram (Figure 3). All participants were clearly advised that participation in the trial was entirely voluntary with the option of withdrawing from the trial at any stage, and that participation or non-participation would not affect their usual care. All women were approached by staff who were appropriately trained in Good Clinical Practice and specifically trained in taking consent for this trial. To be invited for screening, the woman must have been willing and able to give informed consent and to provide a blood sample (of 10 ml) for thyroid antibody and thyroid function testing (TPOAbs and measurement of serum TSH and free T4).

FIGURE 3.

Routes of initial approach for screening.

The aim was to approach women at the optimum point, before their subsequent conception. For women who had a recent miscarriage, the initial approach was carried out after the miscarriage had been confirmed on the early pregnancy unit or when they were admitted to the ward for management. For women who were under infertility services, the initial approach was made at a routine clinic appointment and women with recurrent miscarriage were approached in the recurrent miscarriage clinics.

Some women were screened on an early pregnancy unit following an acute pregnancy loss, when it was their third (or more) miscarriage. These women were categorised into the recurrent miscarriage population. Potential participants were provided with a short screening patient information sheet and given time to consider their involvement.

Figure 4 shows the potential pathways that were followed by all of the screened participants. The co-ordinating midwife/nurse at each centre was responsible for contacting TPO-negative women to inform them that they were ineligible for the TABLET trial and provide reassurance about normal TFTs. A small number of asymptomatic women were identified as having abnormal TFTs via the screening process. It was advised that the local principal investigator would make decisions on further investigations and/or treatment for these women based on the degree of thyroid abnormality and local guidelines. If TPOAbs were positive and TSH and free T4 concentrations were within the normal range for the trial [see Eligibility (inclusion and exclusion) for limits], the woman was sent a TABLET trial participant information sheet and an appointment was arranged to discuss participation at a subsequent clinic visit at which final eligibility checks could be performed. For those who had recently suffered a miscarriage, the woman’s desire to conceive again was explored and only those who indicated that they intended to try again, within the next 12 months, were invited to participate. It was made clear to participants that they could withdraw from the trial at any time. Consent was confirmed in writing. This included consent for future evaluation of themselves and the child and the health records of both through the Office for National Statistics or equivalent.

FIGURE 4.

Screening of potential participants. GP, general practitioner.

Randomisation

Eligibility criteria were confirmed prior to obtaining consent, and demographic and prognostic factors on the Randomisation Notepad were gathered. Following this, the woman could be randomised into the trial. Randomisation was conducted through a secure online randomisation service provided by the University of Birmingham Clinical Trials Unit (BCTU). Following this, trial and bottle numbers were allocated.

Treatment allocations

Blinding

Participants, investigators, research midwives/nurses and other attending clinicians all remained blind to the trial drug allocation for the duration of the trial.

In the case of any serious adverse events (SAEs), management and care of the women was initiated as though the woman was taking levothyroxine. Cases that were considered serious, unexpected and possibly, probably or definitely related [i.e. possible suspected unexpected serious adverse reactions (SUSARs)] were unblinded only at the trial office by the trial co-ordinator. The attending clinician and local principal investigator were not made aware of the actual trial drug. If a participant was withdrawn from the trial as a result of abnormal TFTs (see Thyroid hormone monitoring and criteria for stopping trial treatment) and if the drug allocation was required for the continued medical management of the withdrawn participant, clinicians were advised to contact the TABLET trial office or use the online access to gain unblinding information. Any instances of this were recorded on the trial database.

Scheduled trial appointments

Trial participants returned to the randomising hospital at two further intervals while trying to conceive and for routine antenatal appointments. At each visit, blood samples were taken for TSH and free T4 level (see Thyroid hormone monitoring and criteria for stopping trial treatment). If conception did not take place by the end of the 12th month, the woman was asked to perform a pregnancy test and ensure that it was negative prior to stopping trial medication (Figure 5).

FIGURE 5.

Thyroid drug supply and monitoring timelines.

Compliance and treatment withdrawal

Withdrawal from trial

Participants could voluntarily withdraw their consent to trial participation at any time. If a participant did not return for a scheduled visit, attempts were made to contact her and, when possible, review compliance and AEs. We made an attempt to document all reasons for self-withdrawal. If a participant explicitly withdrew consent to have any further data recorded, their decision was respected and recorded on the electronic data capture system. All communication surrounding the withdrawal was noted in the patient’s records and no further data collected for that patient.

Outcomes and assessment

Statistical considerations

Trial oversight

Trial oversight was provided by a TSC (chaired by Professor Jane Norman, University of Edinburgh) and a DMC (chaired by Professor John Lazarus, University of Cardiff).

The TSC provided independent supervision for the trial, providing advice to the chief and co-investigators and the sponsor on all aspects of the trial throughout the trial. The DMC adopted the DAMOCLES charter30 to define its terms of reference and operation in relation to oversight of the TABLET trial. The DMC met on an approximately 6-monthly basis during the period of recruitment. The patient safety and treatment efficacy aspects were reviewed at each meeting and a decision to continue or stop trial recruitment was based on the criteria defined prior to the trial starting.

Participant flow

Participant flow through the trial is illustrated in Figure 6. A total of 19,556 women were screened for eligibility for the trial, including assessment of whether or not they were TPO antibody positive and biochemically euthyroid. The overall prevalence of TPOAb positivity was found to be 9.5%. Ultimately, 1420 (7%) women were eligible for randomisation. Of those who were ineligible, 16,162 (83%) were TPO antibody negative, 1492 (8%) were not euthyroid and were referred for treatment based on local guidelines and 482 (2%) did not complete screening for entry into the trial.

FIGURE 6.

Flow of participants through the trial. From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright © 2019. Massachusetts Medical Society. Reprinted with permission.

From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright 2019. Massachusetts Medical Society. Reprinted with permission.

A total of 952 out of 1420 (67%) women proceeded to randomisation, with 476 allocated to levothyroxine and 476 to placebo. Six participants in each group were either withdrawn from the trial or lost to follow-up, meaning that 940 (98.7% of those randomised) participants were available for analysis of the primary outcome.

Recruitment

Recruitment took place over 51 months in 49 UK NHS hospitals (Figure 7) from November 2011 to January 2016 (see Appendix 3). With agreement from the TSC, recruitment finished slightly over the sample size target, at 952 recruits. This enabled us to maximise the available power to detect any differences between groups, should one exist, in the primary outcome. Given the lower than anticipated rate of loss to follow-up, 940 participants with available data meant that we had 89% power to detect a difference under the original assumptions set out in Chapter 2. The original sample size calculation assumed 80% power. Site contributions to recruitment are given in Table 2.

FIGURE 7.

Sites in the UK. This map data have been reproduced with the permission of Google (Google Inc., Mountain View, CA, USA) from Google Maps (2015) for non-commercial purposes. Map data: © 2015 Google. For more details, see: www.google.com/permissions/geoguidelines/.

This map data have been reproduced with the permission of Google (Google Inc., Mountain View, CA, USA) from Google Maps (2015) for non-commercial purposes. Map data: © 2015 Google. For more details, see: www.google.com/permissions/geoguidelines/.

Baseline data

The baseline demographic characteristics of participants in the two groups were comparable to the minimisation algorithm, ensuring balance for the factors indicated in Table 3.

Compliance to treatment

Given the long duration for which a participant could potentially be taking the trial medication (12 months to conceive, plus, potentially, an additional 9 months of pregnancy if conception was successful), as well as the known pharmacokinetics of levothyroxine,32 intermittently missing tablets would not affect the thyroxine levels in the body. Therefore, a pragmatic approach was taken, which defined pill-taking of > 75% as good compliance.

In those women for whom compliance data were reported, compliance was found to be good; however, compliance reporting overall was poor. There was a trend of compliance reducing in those trying for a pregnancy from 3 months to 9 months preconception, but then increasing in early pregnancy (at 6–8 weeks). A summary of the compliance to treatment allocation is shown in Table 4.

Results overview

The TABLET trial found no evidence of differences in the primary or any of the key secondary outcomes between the group randomised to receive levothyroxine and the group randomised to placebo. Differences were seen in serum TSH concentration (which was reduced with levothyroxine) and free T4 levels (which increased with levothyroxine) at every time point observed (suggesting the biological effect of levothyroxine), but this did not translate to a clinical benefit to the participants randomised to levothyroxine.

Primary outcome results

Overall, 354 out of 940 participants (38%) experienced a live birth at ≥ 34 weeks of gestation. The live birth rate in the levothyroxine group was 37% (176/470) and the rate in placebo group was 38% (178/470), translating to a RR of 0.97 (95% CI 0.83 to 1.14; p = 0.74) and an absolute risk difference of –0.4% (95% CI –6.6% to 5.8%) (Table 5).

Secondary outcome results

Secondary maternal outcomes: pregnancy outcomes

Similar numbers of women in each group became pregnant: 266 out of 470 (57%) in the levothyroxine group and 274 out of 470 (58%) in the placebo group (see Table 5). In this population, the rate of miscarriage was similar in both groups: 75 out of 266 (28%) in the levothyroxine group and 81 out of 274 (30%) in the placebo group (RR 0.95, 95% CI 0.73 to 1.23). This difference was not statistically significant (p = 0.68). The median gestational age at the time of miscarriage was 8 weeks (IQR 6–10 weeks) in the levothyroxine group and 9 weeks in the placebo group (IQR 7–10 weeks). Ten women in each group delivered before 34 weeks’ gestation (RR 1.02, 95% CI 0.43 to 2.42; p = 0.96), meaning that the number of live births (at ≥ 24 weeks), overall, was 186 in the levothyroxine group and 188 in the placebo group. Overall, there was no evidence of any difference in the time pregnancy ended, through either miscarriage or successful delivery (Figures 8 and 9) (hazard ratio 1.03, 95% CI 0.87 to 1.23; p = 0.72). The results of other pregnancy outcomes appeared similar in both groups, with no significant differences.

FIGURE 8.

Kaplan–Meier curve: time from conception to pregnancy end. From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright © 2019. Massachusetts Medical Society. Reprinted with permission.

From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright 2019. Massachusetts Medical Society. Reprinted with permission.

FIGURE 9.

Kaplan–Meier curve: time from conception to birth (live birth at ≥ 24 weeks).

Neonatal outcomes

The distribution of gestational age at delivery in those women with a live birth was very similar in both groups overall (Figure 10) (hazard ratio 1.09, 95% CI 0.89 to 1.34; p = 0.43). Live births were delivered at 38⁺⁶ weeks and 39⁺⁰ weeks of gestation, on average, in the levothyroxine and placebo groups, respectively (mean difference: –1 day, 95% CI –4 to 3 days; p = 0.65). There were 62 (17%) preterm births (< 37 weeks) observed, but the numbers were very similar in both groups (15% vs. 18%; RR 0.83, 95% CI 0.53 to 1.31; p = 0.43). Birthweights appeared similar in both groups (mean difference: 35 g, 95% CI –168 to 97 g; p = 0.60), with no evidence of any differences in the numbers, large or small, for their gestational age (plus other covariates listed in Table 5). No differences were noted in Apgar scores.

FIGURE 10.

Thyroid-stimulating hormone over time by group. From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright © 2019. Massachusetts Medical Society. Reprinted with permission.

From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright 2019. Massachusetts Medical Society. Reprinted with permission.

Thyroid function data

As expected, differences were seen in levels of serum TSH concentration (reduced with levothyroxine) and free T4 levels (increased with levothyroxine) at every time point observed (Table 6 and Figures 10 and 11), demonstrating a biological effect of levothyroxine treatment. Levels of both TSH and free T4 dropped during pregnancy, but clear differences between groups remained throughout.

TABLE 6 - Thyroid function outcome data

The analysis includes adjustment for baseline value and centre (to allow for variations in assay platform). For continuous outcomes; mean difference > 0 indicates higher levels in the levothyroxine group.

From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright © 2019. Massachusetts Medical Society. Reprinted with permission.

Thyroid function pre-randomisation, pre-pregnancy and during pregnancy	Levothyroxine	Placebo	Mean difference,a 95% CI; p-value
Serum TSH concentration (mlU/l)
Pre-randomisation
Mean (SD), n (log-transformed)	0.674 (0.422), 476	0.652 (0.418), 476
Median (IQR)	2.10 (1.51–2.74)	2.01 (1.45–2.70)
Time (months) after randomisation pre pregnancy
3
Mean (SD), n (log-transformed)	–0.111 (1.389), 298	0.698 (0.687), 288	–0.822 (–0.990 to –0.654); < 0.001
Median (IQR)	1.33 (0.74–2.00)	2.11 (1.50–2.97)
6
Mean (SD), n (log-transformed)	0.250 (1.271), 172	0.716 (0.643), 172	–0.514 (–0.721 to –0.308); < 0.001
Median (IQR)	1.64 (1.10–2.39)	2.10 (1.50–2.60)
9
Mean (SD), n (log-transformed)	0.520 (0.705), 50	0.658 (0.343), 54	–0.17 (–0.38 to –0.04); 0.102
Median (IQR)	1.73 (1.28–2.53)	1.94 (1.62–2.45)
Time (weeks) during pregnancy
6–8
Mean (SD), n (log-transformed)	0.141 (0.967), 177	0.573 (0.766), 184	–0.421 (–0.584 to –0.259); < 0.001
Median (IQR)	1.36 (0.85–1.98)	2.05 (1.38–2.80)
16–18
Mean (SD), n (log-transformed)	0.177 (0.634), 141	0.385 (0.626), 142	–0.216 (–0.346 to –0.085); 0.001
Median (IQR)	1.31 (0.94–1.70)	1.60 (1.12–2.20)
28
Mean (SD), n (log-transformed)	0.123 (0.552), 134	0.327 (0.519), 130	–0.203 (–0.319 to –0.088); 0.001
Median (IQR)	1.30 (0.81–1.61)	1.50 (1.10–1.95)
Serum free T4 level (pmol/l), mean (SD), n
Pre-randomisation	14.6 (1.9), 476	14.5 (2.0), 476
Time (months) after randomisation pre pregnancy, mean (SD), n
3	17.3 (3.9), 298	14.5 (2.6), 288	2.7 (2.2 to 3.2); < 0.0001
6	16.3 (3.5), 172	14.6 (2.2), 172	1.7 (1.2 to 2.3); < 0.0001
9	16.4 (4.1), 51	14.3 (1.9), 54	1.8 (0.7 to 2.9); 0.0012
Time (weeks) during pregnancy, mean (SD), n
6–8	16.6 (3.1), 177	14.9 (2.5), 184	1.7 (1.2 to 2.2); < 0.0001
16–18	14.2 (1.8), 140	13.3 (1.8), 142	1.0 (0.6 to 1.4); < 0.0001
28	13.0 (1.7), 134	12.3 (1.7), 130	0.8 (0.4 to 1.1); < 0.0001

FIGURE 11.

Free T4 over time by group. From the New England Journal of Medicine. Dhillon-Smith et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright © 2019. Massachusetts Medical Society. Reprinted with permission.

From the New England Journal of Medicine. Dhillon-Smith RK, et al. 31 Levothyroxine in women with thyroid peroxidase antibodies before conception. Vol. 380, pp. 1316–25. Copyright 2019. Massachusetts Medical Society. Reprinted with permission.

Ancillary analyses

Introduction

A widely held view is that TPOAb positivity identifies women who display altered immune responses towards a pregnancy and, hence, are vulnerable to adverse pregnancy outcomes. Even in the non-pregnant state, TPOAb positivity has been associated with autoimmune diseases and peripheral T-lymphocytes show a significantly increased Th1 (cell-mediated immunity) to Th2 (humoral immunity) ratio of immune responses,16 particularly in women with a history of reproductive failure. 35 During pregnancy, the maternal immune system plays a key role in normal placentation by allowing the implantation of a ‘foreign body’ while balancing adequate trophoblast invasion with protection of the maternal tissues. 21 Normal pregnancy involves a shift to a Th2-predominant phenomenon, whereas Th1 predominance is associated with miscarriage and premature delivery. 19 It is therefore conceivable that, with TPOAb positivity, an unfavourable pre-pregnancy immune profile or an inappropriate immune response during early pregnancy could lead to an adverse pregnancy outcome.

It has also been postulated that TPOAb positivity induces mild maternal thyroid dysfunction leading to relative thyroid insufficiency for a given individual, despite circulating thyroid hormone levels being within the normal population range. Pregnant euthyroid women with TPOAb have been found to have TSH levels nearer the upper end of the normal range. 36 Thyroid function variations have been associated with changes in immune function even within normal physiological ranges. Higher concentrations of T3 and T4 within physiological ranges could directly enhance innate and adaptive immunity in normal, healthy individuals. 18 At the same time, subclinical hypothyroidism has been associated with increased risk of miscarriage37 and preterm birth. 38 Whether these clinical outcomes are mediated mainly by a direct effect of thyroid hormone on uteroplacental tissue or also indirectly through changes in maternal immune responses that could affect the pregnancy is not known.

The evidence thus far suggests that pregnancy is sensitive to mild changes in both immune function and thyroid function. In this trial, we have tested whether or not levothyroxine treatment can improve pregnancy outcomes in TPOAb-positive women. In this mechanistic study, we explored if levothyroxine exerts its effects on pregnancy outcome through changes in immune responses.

We hypothesised that (1) women with TPOAb positivity display a different composition of circulatory chemocytokines in the non-pregnant state and are unable to mount an appropriate balance of Th1/Th2-type immune responses during pregnancy, which increases their risk of an adverse pregnancy outcome and (2) treatment with levothyroxine can normalise chemocytokine concentrations before conception and promote the ability to mount an appropriate immune response during pregnancy, which increases the chance of a successful pregnancy outcome.

The study aimed to address these research questions:

• Is there any evidence of an altered immune status systemically (as found in the maternal circulation) in TPOAb-positive women compared with TPOAb-negative women?

• Does levothyroxine treatment alter serum chemocytokine concentrations in TPOAb-positive women before and during pregnancy?

• Are the changes in circulatory chemocytokines brought about by levothyroxine treatment associated with an improvement in the chance of a livebirth outcome?

Methods

Results

Of the 49 trial subjects who participated in this mechanistic study, 16 provided samples at recruitment (baseline), of whom 10 had a history of infertility and six had a history of miscarriage. The remaining 33 subjects provided their first sample only after treatment had commenced. The subject numbers and characteristics of those in the mechanistic study are shown in Table 12. Overall, the subjects were fairly distributed across the two treatment groups (levothyroxine, n = 26; placebo, n = 23); 60% of subjects had a history of infertility and, overall, 40% had a live birth outcome (Table 13).

Data were not available at every time point for all subjects. Different sets of subjects were included at each time point. Of those that provided a baseline sample, only seven provided a non-pregnant sample post treatment commencement and only four participants provided a sample in the first trimester of pregnancy; thus, the study had limited power to examine the longitudinal effects of levothyroxine treatment. If conception was achieved prior to the 3-month follow-up, a pre-pregnancy sample would not have been taken. Subjects who did not achieve conception after 1 year post recruitment or who developed abnormal thyroid function were withdrawn from the trial, so no pregnancy samples were obtained. Those who miscarried after the first pregnancy visit did not have any samples taken at later pregnancy time points.

Principal components analysis

Physiologically, chemocytokines work in concert to exert their effects and each one cannot be considered in isolation. Previous studies have compared ratios of specific Th1 cytokines to specific Th2 cytokines, but even these would not be able to capture a comprehensive picture. We have, thus, used PCA, which assesses the combined contribution of groups of chemocytokines using weighted averages.

Three main clusters of chemocytokines emerged as drivers for the variation in the data. The first three principal components (PCs) cumulatively explained 52% of the variation in the data (PC1: 29%, PC2: 13% and PC3: 10%; Figure 12). The chemocytokines IL-1β, IL-2, IFN-γ, IL-17A, IL-10, IL-6 and IL-4 contributed most heavily to the first PC. TNF-α, RANTES and MIP-1α contributed most to the second PC. VEGF and ENA-78 contributed most negatively, whereas IL-8 and granulocyte-macrophage colony-stimulating factor (GM-CSF) contributed most positively, to the third PC. These data are shown in Figure 12.

FIGURE 12.

Principal component analysis. (a) The relative contributions of the various chemocytokines to the variation in data; and (b) biplot depicting PC1 and PC2 axes. Green font and arrows show the contribution of various cytokines to the axis.

Results for research question 1

Research question 1: is there any evidence of an altered immune status systemically (as found in the maternal circulation) in TPOAb-positive women compared with TPOAb-negative women?

Univariate analysis

Comparisons of TPOAb-positive subjects at recruitment (n = 16) with TPOAb-negative controls (n = 6) showed no statistically significant differences in any of the chemocytokines (data are not shown). Among TPOAb-positive women, those with a history of miscarriage had a trend of a lower RANTES concentration than those with a history of infertility, but this was not statistically significant after FDR-correction.

To address the postulation that there could be an imbalance in Th1 and Th2 immunity in TPOAb-positive women, we examined the ratio in the concentrations of the classical Th1 cytokines, TNF-α and IFN-γ, to the concentrations of the Th2 cytokines IL-10 and IL-4. There were no significant differences in the ratio of Th1 : Th2 cytokines between TPOAb-positive and TPOAb-negative groups (Table 14), although there was a general tendency for a lowering of Th1 cytokines relative to Th2 cytokines with TPOAb positivity, contrary to previous reports. When we stratified by history of miscarriage or infertility, the ratio of IFN-γ/IL-10 was significantly lower in those with a history of infertility than in those with a history of miscarriage (see Table 14).

TABLE 14 - Differences in chemocytokine concentrations in non-pregnant subjects who received levothyroxine treatment compared with non-pregnant subjects who received placebo

NS, not significant.

β is the difference in log₁₀-chemocytokine concentration with levothyroxine treatment, relative to placebo. Non-significant differences in the other chemocytokines are not shown. None of the p-values was lower than the threshold of significance using the Benjamini–Hochberg approach,42 with FDR set at 0.15.

Chemocytokine	Unadjusted			Adjusted for history of miscarriage or infertility
	βa (95% CI)	Raw p-value	FDR-corrected p-value	βa (95% CI)	Raw p-value	FDR-corrected p-value
ENA-78	0.29 (0.00 to 0.59)	0.051	NS	0.35 (0.05 to 0.64)	0.024	NS
IL-8	–0.41 (–0.81 to –0.01)	0.043	NS	–0.47 (–0.88 to –0.06)	0.027	NS

Principal components analysis

At recruitment, there was no significant difference or separation between TPOAb-negative women, TPOAb-positive women with a history of miscarriage and TPOAb-positive women with a history of infertility (Figure 13).

FIGURE 13.

Principal component analysis plot showing the weighted averages of all 17 chemocytokines [in TPOAb-negative controls (n = 6), TPOAb-positive women with a history of miscarriage (n = 6) and TPOAb-positive women with a history of infertility (n = 10)].

Results for research question 2

Research question 2: does levothyroxine treatment alter serum chemocytokine concentrations in TPOAb positive women?

Univariate analysis and logistic regression

Non-pregnant cross-sectional assessment

Following the treatment of TPOAb-positive women, non-pregnant women (results at 3 or 6 months) demonstrated no significant differences in any of the chemocytokine concentrations in those who received levothyroxine treatment compared with those who received placebo, although a non-statistically significant trend of an increase in ENA-78 and a reduction in IL-8 was observed. Because our analysis of baseline samples was suggestive of possible differences between women with a history of miscarriage and those with a history of infertility, we also performed linear regression to adjust for this factor, but this did not change the results (Table 15).

TABLE 15 - Differences in the longitudinal change in chemocytokine concentrations from recruitment to 3 or 6 months non-pregnant time points with levothyroxine treatment (n = 5) compared with placebo (n = 2)

β is the difference in the longitudinal change in log₁₀-chemocytokine concentrations between the two time points in those who received levothyroxine treatment relative to those who received placebo. Non-significant differences in the other chemocytokines are not shown.

Indicates p-values that were lower than the threshold of significance using the Benjamini–Hochberg approach,42 with FDR set at 0.15.

Note

Adjustments were made for history of infertility or miscarriage (one case from each treatment arm had a history of miscarriage).

Chemocytokine	Unadjusted			Adjusted for history of miscarriage or infertility
	βa (95% CI)	Raw p-value	FDR-corrected p-value	βa (95% CI)	Raw p-value	FDR-corrected p-value
MIP-1α	0.58 (–0.70 to 1.87)	0.297	0.46	0.92 (0.42 to 1.41)	0.007b	0.12b
IL-1β	0.25 (0.13 to 0.38)	0.004b	0.07b	0.26 (0.10 to 0.41)	0.01b	0.09b
IL-4	0.12 (0.04 to 0.19)	0.009b	0.08b	0.12 (0.04 to 0.20)	0.013b	0.07b
IL-8	0.95 (0.23 to 1.67)	0.019b	0.11b	1.07 (0.33 to 1.80)	0.016b	0.07b
G-CSF	0.53 (–0.13 to 1.18)	0.094	0.32	0.66 (0.11 to 1.21)	0.029b	0.10b
IFN-γ	0.30 (0.05 to 0.54)	0.026b	0.11b	0.32 (0.02 to 0.61)	0.04b	0.11b

The ratio of the Th1 cytokines to Th2 cytokines was also not significantly different between those treated with levothyroxine and those on placebo (data are not shown).

Non-pregnant longitudinal assessment

In those who had provided a baseline sample as well as a non-pregnant sample post treatment commencement, we were able to calculate the longitudinal change in each chemocytokine concentration in individuals between the two time points. We compared the differences in these longitudinal changes in those who had received levothyroxine treatment (n = 5) with the changes in those who had received placebo (n = 2). There was a greater increase in the concentrations of the chemokines MIP-1α; G-CSF; the Th1 cytokines IL-1β, IL-8, IFN-γ; and the Th2 cytokine IL-4 with levothyroxine treatment than with placebo treatment, after adjustment for history of miscarriage or infertility (Table 16). We were unable to assess the implications of these ‘preconception’ changes on pregnancy outcome as two cases from the levothyroxine arm and both cases from the placebo arm did not conceive.

TABLE 16 - Differences in chemocytokine concentrations with levothyroxine treatment compared with placebo in the first trimester

β is the difference in log₁₀-chemocytokine concentration in those who received levothyroxine treatment relative to those who received placebo. Non-significant differences in the other chemocytokines are not shown.

p-values that were lower than the threshold of significance using the Benjamini–Hochberg approach,42 with FDR set at 0.15.

Note

In all pregnancies (levothyroxine, n = 10; placebo, n = 7) and in only pregnancies that resulted in a live birth (levothyroxine, n = 8; placebo, n = 6), adjustments were made for history of infertility or miscarriage.

Chemocytokine	All pregnancies			Only pregnancies resulting in live birth
	βa (95% CI)	Raw p-value	FDR-corrected p-value	βa (95% CI)	Raw p-value	FDR-corrected p-value
Thrombopoietin	0.73 (0.24 to 1.22)	0.007b	0.12b	0.62 (0.07 to 1.17)	0.03	NS
IL-1β	0.56 (0.17 to 0.95)	0.008b	0.07b	0.43 (0.04 to 0.82)	0.034	NS

Early first trimester (6–8 weeks’ gestation) cross-sectional assessment

There were higher thrombopoietin and IL-1β concentrations in the levothyroxine group than in the placebo group, with adjustment for history of infertility or miscarriage (Table 17). On excluding the three cases that subsequently miscarried, thrombopoietin and IL-1β concentrations showed a similar direction of change with levothyroxine treatment, but the changes were no longer significantly different, probably because of the lack of statistical power and the removal of slightly more exaggerated changes occurring in the miscarriage cases.

TABLE 17 - Ratios of the absolute concentrations of Th1 cytokines to Th2 cytokines with levothyroxine treatment compared with placebo in the first trimester

Note

Values for both trial groups are presented as median (IQR) (25th–75th percentile). Comparisons were performed using non-parametric Mann–Whitney U-tests.

Chemocytokine ratio	All pregnancies			Only pregnancies resulting in live birth
	Placebo (n = 7), median (IQR)	Levothyroxine (n = 10), median (IQR)	p-value	Placebo (n = 6), median (IQR)	Levothyroxine (n = 8), median (IQR)	p-value
TNF-α/IL-10	9.93 (0.25–13.54)	1.30 (0.50–1.70)	0.558	6.86 (0.25–13.54)	1.50 (0.70–2.57)	0.897
TNF-α/IL-4	4.96 (0.12–6.77)	0.53 (0.16–7.61)	0.770	3.57 (0.12–6.77)	0.67 (0.27–8.56)	0.796
IFN-γ/IL-10	30.32 (2.01–37.89)	4.83 (3.61–11.44)	0.558	20.49 (2.01–37.89)	5.52 (3.37–17.41)	0.897
IFN-γ/IL-4	15.16 (1.63–18.95)	2.56 (1.42–34.75)	0.696	11.14 (1.63–18.95)	3.14 (1.97–36.88)	0.796
TNF-α + IFN-γ/IL-4 + IL-10	12.27 (0.49–17.62)	2.04 (0.60–6.73)	0.558	8.73 (0.49–17.62)	2.81 (1.30–6.91)	0.897
Thrombopoietin/G-CSF	0.57 (0.28–3.01)	4.86 (2.72–8.48)	0.019	1.30 (0.44–3.01)	5.37 (3.63–8.56)	0.01

To address the hypothesis that levothyroxine could promote the mounting of a more favourable immune response in pregnancy, characterised by a reduced Th1 response and increased Th2 response, we examined the ratios of the Th1 cytokines TNF-α and IFN-γ to the Th2 cytokines IL-10 and IL-4. Because the elevation of thrombopoietin with concomitant decrease in G-CSF has previously been associated with miscarriage,39 we also examined the ratio of thrombopoietin to G-CSF. Although, overall, there was a tendency for a lowering of Th1 cytokines relative to Th2 cytokines in early pregnancy with levothyroxine treatment, there was wide interindividual variation and no significant differences were found between the levothyroxine-treated group and the placebo group (Table 18). There were higher levels of thrombopoietin when compared with G-CSF with levothyroxine treatment, which has previously been reported to be associated with a higher risk of miscarriage. More importantly, the same trends in the ratios were also found when analyses were confined to only those who had a live birth outcome, suggesting that changes in chemocytokine concentrations are not major determinants of pregnancy outcome in this subcohort.

TABLE 18 - Differences in chemocytokine concentrations in the first trimester between pregnancies that subsequently resulted in a miscarriage (n = 3) and those that resulted in a live birth (n = 14)

β is the difference in log₁₀-chemocytokine concentration with miscarriage relative to live birth. Non-significant differences in the other chemocytokines are not shown.

p-values that were lower than the threshold of significance using the Bejamini–Hochberg approach,42 with FDR set at 0.15.

Note

Results without and with adjustment for levothyroxine treatment are shown.

Chemocytokine	Unadjusted			Adjusted for levothyroxine treatment
	βa (95% CI)	Raw p-value	FDR-corrected p-value	βa (95% CI)	Raw p-value	FDR-corrected p-value
MIP-1α	–0.86 (–1.31 to –0.41)	0.001	0.017b	–0.84 (–1.28 to –0.40)	0.001	0.017b
IFN-γ	–0.22 (–0.45 to 0.02)	0.065	NS	–0.23 (–0.46 to 0.01)	0.055	NS

Second and third trimesters cross-sectional assessment

All cases resulted in a live birth beyond 34 weeks of gestation. There were no significant differences in chemocytokine concentrations between those treated with levothyroxine (n = 8) and those who received placebo (n = 5), with and without adjustment for history of miscarriage/infertility (data are not shown). Similarly, there were no differences in the Th1 : Th2 ratios with levothyroxine treatment, apart from the thrombopoietin : G-CSF ratio in the third trimester, which was significantly higher in the levothyroxine-treated women [median 6.16 (IQR 2.15–12.57)] than in those on placebo [median 1.31 (IQR 1.19–1.59)].

Principal components analysis

Levothyroxine treatment did not significantly change the levels of the three major PCs of chemocytokines, either in the non-pregnant participants or during pregnancy (data are not shown). However, an initial history of miscarriage or infertility could potentially modify the relationship between treatment group and chemocytokine concentrations (Figure 14), especially for placebo during pregnancy. Hence, we subsequently stratified our analyses by history of miscarriage or infertility to investigate the effects of levothyroxine.

FIGURE 14.

Principal component analysis showing separation in the weighted averages of chemocytokine concentrations between various subgroups.

Changes in the levels of PCs with treatment following stratification by history of miscarriage or infertility are shown in Figure 15. There were no significant differences in PC1, PC2 and PC3 at baseline (recruitment), and in non-pregnant women (3 months or 6 months) between treatment groups in both women with a history of miscarriage or women with a history of infertility. During pregnancy, there were no significant differences with levothyroxine treatment in PC1 and PC3, but PC2 showed a consistent trend of being higher with levothyroxine treatment in the women with a history of miscarriage but an opposite trend of being lower with levothyroxine in women with a history of infertility, although differences were not statistically significant. The differences in PC2 during pregnancy with levothyroxine treatment may simply be reflective of trends which were already present prior to any intervention, as suggested by similar PC2 trends observed at recruitment. This needs to be interpreted with caution because different cases are represented at recruitment and in pregnancy, so longitudinal trends cannot be assessed.

FIGURE 15.

Differences in the three main principal components between levothyroxine treatment and placebo, when stratified by history of miscarriage or infertility at each time point. Baseline (recruitment), non-pregnant (3 months or 6 months post recruitment) and during pregnancy (trimester 1: 6–8 weeks, trimester 2: 16–18 weeks and trimester 3: 28 weeks). Different women are included at each time point according to availability of samples for analysis. Box plots indicate the IQR and whiskers indicate 1.5 × IQR. The number of subjects in each group and p-values are given at the bottom of the figure. Tri1, trimester 1; Tri2, trimester 2; Tri3, trimester 3.

Discussion and conclusions

This trial has demonstrated that serum chemocytokine concentrations of TPOAb-positive women do not differ from the serum chemocytokine concentrations of TPOAb-negative women outside pregnancy. Although levothyroxine treatment of TPOAb-positive women resulted in some changes in serum chemocytokine concentrations, both in the non-pregnant state and in early pregnancy, these changes did not influence the chance of a subsequent live birth outcome.

This trial has measured only chemocytokines in the serum, which comprise secretions from peripheral blood mononuclear cells (PBMCs), adipose tissue and other tissues. In uteroplacental tissues during pregnancy, local T cells are major producers of cytokines, and non-lymphoid decidual cells and trophoblast are also significant contributors. 19,21,43 Thus, serum chemocytokine changes in this trial may not reflect changes in cytokine expression by PBMCs or changes occurring locally at the maternal–fetal interface, and this may account for differences in the findings of this trial from previously reported studies.

The finding of no difference in individual serum chemocytokines between TPOAb-positive and TPOAb-negative status, and a general tendency for lower Th1 : Th2 ratios with TPOAb positivity, is in contrast to the findings previously reported in CD3+/CD4+ T cells obtained from a similar population of women with a history of infertility or recurrent miscarriage. 35 They reported an increase in the ratio of T cells expressing the Th1 cytokine TNF-α to T cells expressing the Th2 cytokine IL-10 in the circulation of TPOAb-positive women compared with TPOAb-negative women. 35 However, we did find a lower IFN-λ : IL-10 ratio in TPOAb-positive women with a history of infertility than in those with a history of miscarriage. This could represent a phenotypic difference in immune responses, especially in TPOAb-positive women with a history of infertility, that has not been distinguished before between women with a history of infertility and those without such a history.

In a different population of euthyroid TPOAb-positive patients with chronic idiopathic urticaria serum concentrations of the Th1 cytokines, IFN-γ and TNF-α were increased by levothyroxine treatment which the authors thought was because of TSH suppression. 44 In this trial, no changes in IFN-γ and TNF-α were observed in the non-pregnant state in cross-sectional analyses, but an increase in IFN-γ (with no change in TNF-α) and other chemocytokine changes were observed in a longitudinal analysis with levothyroxine treatment. Differences in trial findings may be accounted for by little or no TSH reduction in the levothyroxine-treated subcohort [the mean reduction in TSH concentration between recruitment and preconception follow-up was 0.0195 mlU/l (SD ± 1.88) in the levothyroxine-treated group and 0.287 mlU/l (SD ± 0.96) in the placebo group] and being limited to only a young adult female population in this trial.

We had found that, in early pregnancy, there was a suggestion of increased serum thrombopoietin and IL-1β with levothyroxine treatment, but no differences were observed in the second or third trimesters between treatment groups. Another study45 similarly reported a lack of change in a range of serum cytokines with levothyroxine treatment of TPOAb-positive women during the first and second trimesters of pregnancy; however, this study45 did not measure thrombopoietin and IL-1β. Furthermore, we observed an elevated thrombopoietin : G-CSF ratio in the first trimester with levothyroxine treatment, a change that has previously been associated with miscarriage,39 yet in this trial there was no correlation with pregnancy outcome. We have also previously reported that tri-iodothyronine treatment in vitro of isolated decidual cells from the first trimester changed the secretion of a range of chemocytokines,40 but this has not been reflected by the systemic serum chemocytokine measures in vivo with levothyroxine treatment in this trial.

Overall, the findings of a lower MIP-1α concentration in women who subsequently miscarried than in those who had a live birth, and no change in all other chemocytokines or in Th1 : Th2 cytokine ratios evaluated, are in contrast to the findings reported in other studies. 40 The difference between this trial and the others is that this trial population is confined to only TPOAb-positive women, who may display a different immune response to pregnancy compared with TPOAb-negative women. The PCA findings show that alterations occur across a range of chemocytokines with levothyroxine treatment in a way that is influenced by a history of miscarriage or infertility. However, these alterations showed no association with subsequent pregnancy outcome. All of this suggests that, in women with TPOAb positivity, differences in the specific chemocytokines that we have measured are not major determinants of a live birth outcome.

The main strength of this mechanistic trial is that it is embedded in the context of a large randomised controlled trial that has the power to provide definitive clinical outcomes, which cannot be derived from smaller sample sizes that are typical of studies analysing a range of chemocytokines in this field. Complementarity of the trial results provides greater confidence in drawing conclusions from the chemocytokine data based on this limited subset. A major limitation is the small sample size and missing time points, which limits statistical power and the ability to conduct longitudinal analyses. Additional assessments of cytokine expression in PBMCs and in cells at the maternal–fetal interface would have provided a more comprehensive picture of the changes in immune responses occurring with levothyroxine treatment.

In conclusion, treatment of TPOAb-positive women with levothyroxine resulted in some changes in chemocytokine concentrations in the non-pregnant state and in very early pregnancy, but these changes had no bearing on whether or not the pregnancy resulted in a live birth outcome.

Trial strengths

To our knowledge, this trial is the largest ever randomised placebo-controlled clinical trial to report on the treatment effects of pre- and post-conception levothyroxine for women with a history of miscarriage or infertility and thyroid autoantibodies. The trial design ensured internal validity, enabling the results to be interpreted with confidence.

Randomisation and minimisation were effective in achieving balanced treatment allocations at baseline. A computer-generated allocation sequence, allocation concealment and blinding prevented investigators from knowing the allocation of the next participant based on prior treatment assignments. Possible confounding factors such as number of previous miscarriages, maternal age and TSH levels were similarly distributed between treatment groups. The TABLET trial also avoided performance bias by blinding the participants and care providers to treatments.

To determine the minimally important clinical difference to prompt a change in practice, health-care professionals, patients and representatives of the Miscarriage Association were surveyed. A consensus of a 10% increase in live birth rates beyond 34 weeks’ gestation evolved from this consultation, and this was used to derive a target sample size for the trial of 760 participants with primary outcome data. The evidence for levothyroxine in the literature at the start of the trial suggested that a much greater effect could be anticipated, with a halving of the risk of miscarriage rate estimated in a meta-analysis of two comparable trials8,9 (RR 0.48, 95% CI 0.25 to 0.92). 6 If all other assumptions were correct, then the trial would have > 99% power to detect differences of ≥ 15%.

The sample size assumed a rate of live births at ≥ 34 weeks of gestation of 55% in the placebo group, acknowledging that not all women would become pregnant within 12 months of randomisation. We planned to randomise 900 women in total (450 participants to each group) and we actually exceeded this number, accruing 952 women. Furthermore, we anticipated that we would fail to capture the primary outcome in 15% of women, because of loss to follow-up or withdrawal. Ultimately, we had an outcome for 98.7% women, equally reported in each treatment group; hence, we consider the findings of this trial to be methodologically and statistically robust.

We originally restricted the population to solely women who had experienced a miscarriage. We expanded the eligible population to include women undergoing fertility treatment, as there was comparable evidence from cohort studies that there was an association between miscarriage and thyroid autoantibodies. 6 This allows the results to be generalisable to both populations. It was anticipated that the broadening of the criteria might result in a lower rate of the primary outcome, as a greater proportion of the fertility population might fail to conceive within 1 year than we had originally factored into the rationale in the sample size calculation. Once we had considered the low apparent loss to follow-up rate and the potential power from a range of live births at ≥ 34 weeks in the control group, we considered the original target of 900 women would be sufficient and did not adjust the target sample size. Ultimately, we observed an overall rate of 38% (354/940) for the primary outcome, with only a small absolute difference of 0.4% between groups. Estimated uncertainty around this estimate was, at most, 6.6%, allowing us to rule out missing a clinically meaningful difference, which had been defined as 10% at the outset of the trial.

The trial design offered a number of other strengths with respect to data collection and analysis. The treatment of participants in a large number of trial centres and by a large number of practitioners allowed intervention impact to be evaluated without confounding by individual variance in clinical practice and local reference ranges for thyroid function. The outcome measures selected were routine variables widely used by clinicians who are familiar with early pregnancy care. This ensured that the outcomes were well understood and easy to record. Almost all of the outcome data recorded during the TABLET trial were objective outcomes (rather than subjective descriptions), and the trial was blinded, so there was no risk of incurring assessor bias.

The trial intervention was deliverable in the context of customary care without major impacts on health service structure. The mode of administration of the IMP was designed to reflect the preferences expressed by patients, and most of the data collection could be performed during routine antenatal and postnatal appointments of the trial participants.

Limitations and critique

We consider the trial to have been designed and conducted in order to be methodologically robust. Nevertheless, there are some limitations that could affect the results observed.

The potential criticisms of this trial include (1) variation in the threshold of abnormality of TPOAbs at time of randomisation, (2) a possibly suboptimal dose, (3) a possibly inadequate duration of treatment prior to conception, (4) dilution of the treatment effect by factors such as breaches of protocol and (5) inappropriate thresholds for cessation of treatment in each trimester. We examine these issues individually in the following sections.

Findings in the context of existing literature

Since the initiation of TABLET trial, a study by Wang et al. 11 has explored the effect of levothyroxine on miscarriage among women with normal thyroid function and thyroid autoimmunity undergoing IVF. They randomised 600 women to receive either levothyroxine or placebo. The levothyroxine dose was either 25 µg or 50 µg and then titrated according to TSH levels during pregnancy. They found no difference in miscarriage rates or live birth rates between the groups. A further randomised study by Nazarpour et al. 47 explored the effects of levothyroxine in TPOAb-positive women who were euthyroid or had subclinical thyroid disease. They found that treatment with levothyroxine did result in a lower preterm birth rate. This study was much smaller than the TABLET trial, with only 131 TPOAb-positive women. Given that the euthyroid and subclinical hypothyroid women (with TSH concentrations of up to 10 mlU/l) were combined in the same group, we felt that the data were not comparable with those in this trial.

When adding the results of this trial and the trial by Wang et al. 11 to those of the Negro et al. 8,9 trials, the pooled results show no significant reduction in miscarriage with levothyroxine treatment compared with control (Figure 16).

FIGURE 16.

Effect of levothyroxine treatment in reducing miscarriage in euthyroid women with TPOAbs. M–H, Mantel–Haenszel.

The slightly higher miscarriage rate observed in this trial (28% in the levothyroxine group and 30% in the placebo group) further confirms that women with TPOAbs are at a higher risk of miscarriage than the general population.

Patient and public involvement

In the TABLET trial, patient and public involvement was utilised at several stages of the trial design, development and monitoring. This included questionnaires for patients to assess the acceptability of the intervention, and engagement in the development of patient-facing literature for participants. The TSC included a representative of the Miscarriage Association. We believe that these roles were important to ensure appropriate communication with trial participants and project oversight throughout the duration of the research. Dissemination of the results will be supported by the Miscarriage Association.

Interpretation

These findings show that women with a normal thyroid function who are positive for TPOAbs do not benefit from levothyroxine treatment commenced preconceptually for any of the key clinical outcomes that we observed. There were no significant differences between levothyroxine and placebo for neonatal or maternal secondary outcomes. These findings are not consistent with the findings of two smaller and poorer-quality controlled studies,8,9 that reported benefit from levothyroxine in the same population. However, they are in keeping with more recent findings seen in the larger randomised controlled trial by Wang et al. 11 The TABLET trial is, to our knowledge, by far the largest of all randomised controlled trials conducted in this field; therefore, it has conclusively answered the question of whether or not levothyroxine is beneficial for women with TPOAb.

The TABLET trial has added to the available safety data regarding the use of levothyroxine during pregnancy, with some suggestion that there may be harm associated with its use in euthyroid women. This is discussed in more detail in Chapter 6, Implications for health care.

Generalisability

Centres participating in the trial were geographically spread across the UK, improving the generalisability of the results for euthyroid TPOAb-positive women actively trying for a pregnancy.

All women in the trial belonged to a ‘selected population’, whether it was history of miscarriage or infertility. Therefore, the results of this trial may not represent that of true unselected ‘low-risk’ women with no gynaecological or obstetric risk factors. Given that it would be impractical to test all women preconceptually for thyroid function and TPOAbs, we feel that the pragmatic approach to this trial (testing women who access health-care preconception) makes the results generalisable to all euthyroid women with TPOAb of reproductive age.

The exclusion criteria were kept to a minimum and the heterogeneity of the population was well reflected by trial participants.

Implications for health care

The key findings of the TABLET trial are clear and sufficiently generalisable to inform clinical practice. On the basis of the results of this trial, levothyroxine treatment commenced preconceptually does not have clinically significant benefits in euthyroid women with TPOAbs.

Levothyroxine is a widely used drug in the treatment of overt hypothyroidism and has not been found to have harmful effects on mother or fetus in this group. 13 There is evidence to suggest that women with TPOAbs have higher levels of TSH and are at higher risk of progression to subclinical hypothyroidisim in pregnancy and in later life. 48 Subclinical hypothyroidism itself is linked to adverse obstetric outcomes such as miscarriage and preterm birth. For this reason, clinicians are moving towards empirically treating women with subclinical hypothyroidism with levothyroxine. This is most evident in the fertility setting where women with TSH levels of > 2.5 mIU/l, with or without TPOAbs, are being commenced on levothyroxine preconceptually. These approaches are based on the rationale that, despite the lack of convincing evidence of efficacy of levothyroxine in reducing obstetric risks, the potential benefits are considered to outweigh any potential risks from the medication itself. This belief, however, has recently been contested by a large retrospective cohort study of > 5000 women with subclinical hypothyroidism published in BMJ by Maraka et al. 49 This study found that, although treatment with levothyroxine contributed to a significantly lower odds of pregnancy loss than in untreated women, there were higher odds of preterm delivery, gestational diabetes and pre-eclampsia. 49 Similarly, for our trial we noted higher rates of gestational diabetes (11% vs. 9%; p = 0.62) and pre-eclampsia (5% vs. 3%; p = 0.28) in the levothyroxine group, although these were not statistically significant.

There was also a higher number of SAEs reported in the levothyroxine group than in the placebo group (28 vs. 16 cases, respectively), as well as a higher rate of maternal admissions to HDUs (6% vs. 3%, respectively; p = 0.27), but none of these were attributed to the trial medication.

The subset of women recruited into the mechanistic study demonstrated that treatment with levothyroxine resulted in some changes in chemocytokine concentrations in the non-pregnant state and in very early pregnancy, but these changes had no bearing on whether or not the pregnancy resulted in a live birth outcome.

Recommendations for research

In our opinion, no further research is required to evaluate the role of levothyroxine therapy in reducing miscarriage and preterm birth rates for euthyroid women with TPOAbs.

Questions that remain unaddressed relate to the effects of levothyroxine in women who have subclinical hypothyroidism. The authors of the existing published trials that evaluated euthyroid TPOAb-positive women8,9,11 and the authors of T4-LIFE10 (yet to be published) have verbally agreed to share their data to allow for an individual patient data meta-analysis to be conducted. This means that all existing trial data on the topic of subclinical hypothyroidism and TPOAbs will be collated. External funding will be sought for this work. The treatment of mild subclinical hypothyroidism (TSH levels of 2.5–4.5 mIU/l) is a highly controversial area, particularly in the fertility setting, and so collating high-quality data for women with/without TPOAbs in this group will be invaluable to shaping clinical practice.

The mechanistic study was unable to demonstrate any clinically relevant changes in serum chemocytokine concentrations in the non-pregnant state or in pregnancy.

Principal investigators

Ash Alam (Arrowe Park Hospital, Wirral), Amaju Ikomi (Basildon Hospital), Derek Tuffnell (Bradford Royal Infirmary), Ayman Ewies (City Hospital, Birmingham), Fadi Alfhaily (Colchester General Hospital), Simon Wood (Countess of Chester Hospital), Laura Hipple (Cumberland Infirmary), Rukma Bhattacharya (Derriford Hospital, Plymouth), Seema Sen (University Hospital North Durham), Sridevi Sankaran (Frimley Park Hospital, Surrey), Vincent Bamigboye (Furness General Hospital), Padma Manda (James Cook Hospital, Middlesbrough), Richard Russell (Liverpool Womens Hospital), Jag Samra (New Cross Hospital, Wolverhampton), Alpa Shah (Newham General Hospital), Sachchidananda Maiti (North Manchester General Hospital), Judith Moore (Queens Medical Centre, Nottingham), Karen Watkins (Royal Cornwall Hospital), Kannamannadiar Jayaprakasan (Royal Derby Hospital), Lisa Joels (Royal Devon and Exeter Hospital), Jo Fiquet (Royal United Hospital, Bath), Zoey Robinson (Royal United Hospital, Bath), Jane Shillito (St James’ Hospital Leeds), Nigel Simpson (St James’ Hospital Leeds), Catherine Bass (St Peter’s Hospital, Chertsey), Saikat Banerjee (St Peter’s Hospital, Chertsey), Judith Hamilton (St Thomas’ Hospital, London), Tom Holland (St Thomas’ Hospital, London), Tristan Richardson (The Royal Bournemouth Hospital), Meenakshi Choudhary (The Royal Victoria Infirmary, Newcastle), Rita Arya (Warrington Hospital), Colin Johnston (Watford General Hospital), Ismail Wong (Whipps Cross University Hospital, London) and Jagadeeswari Karuppaswamy (Wrightington Hospital).

Research fellows, research nurses and clinical assistants

Justin Chu (Academic Clinical Lecturer, Birmingham Womens Hospital), Hoda Harb (Academic Clinical Lecturer, Birmingham Womens Hospital), Oonagh Pickering (Research Nurse, Birmingham Womens Hospital), Bev Hammond (Research Nurse, Burnley General Hospital), Matthew Milner (Research Nurse, Burnley General Hospital), Fiona Kinney (Research Nurse, City Hospital, Birmingham), Aine Turner (Research Nurse, Colchester General Hospital), Jean Bvumbe (Research Nurse, Guy’s Hospital, London), Charlotte Yearwood-Martin (Research Nurse, Guy’s Hospital, London), Sinead Morgan (Research Fellow, Kings College Hospital, London), Pamela Corlett (Research Nurse, Liverpool Womens Hospital), Elizabeth Kane (Research Nurse, Liverpool Womens Hospital), Lesley Harris (Research Nurse, Liverpool Womens Hospital), Rebecca Denyer (Research Clinical Assistant, New Cross Hospital, Wolverhampton), Karen Evans (Research Nurse, New Cross Hospital, Wolverhampton), Sharon Kempson (Research Nurse, New Cross Hospital, Wolverhampton), Joanne Hutchinson (Research Nurse, Newham Hospital), Helen Millward (Research Nurse, Princess Royal Hospital, Telford), Anna Molnar (Research Nurse, Queens Medical Centre, Nottingham), Liz Barnes (Research Nurse, Queens Medical Centre, Nottingham), Lucinda Wilson (Research Nurse, Queens Medical Centre, Nottingham), Kat Rhead (Research Nurse, Royal Bolton Hospital), Alice Rossi (Research Nurse, Royal London Hospital, London), Lilith Haynes (Research Nurse, St Bartholomew’s Hospital, London), Sharon Nettleton (Research Nurse, St James’ Hospital, Leeds), Jayne Wagstaff (Research Nurse, St James’ Hospital, Leeds), Jane Budd (Research Nurse, St Mary’s Hospital, Manchester), Lucy Dwyer (Research Manager, St Mary’s Hospital, Manchester), Christine Hughes (Research practitioner, St Mary’s Hospital, Manchester) Hannah Phillips (Research Nurse, St Mary’s Hospital, Manchester), Colleen Hunt (Research Assistant, St Michael’s Hospital, Bristol), Carole Shahin (Research Nurse, St Michael’s Hospital, Bristol), Alison Kirby (Research Midwife, St Michael’s Hospital, Bristol), Jennifer Hatton (Research Nurse, St Thomas’ Hospital, London), Debbie Finucane (Research Midwife, St Thomas’ Hospital, London), Victoria Murtha (Research Nurse, The Royal Victoria Infirmary, Newcastle), Michelle Perkins (Research Nurse, The Royal Victoria Infirmary, Newcastle), Debbie Bullen (University Hospital, Coventry), Lindsay Prue (University Hospital, Coventry), Rachel Crone (Research Nurse, Warrington Hospital), Lindsay Roughley (Research Midwife, Warrington Hospital) and Zandile Maseko (Research Nurse, Whipps Cross University Hospital, London).

We would like to acknowledge Adaikalavn Ramasamy (Singapore Institute for Clinical Sciences) and Hsin Fang Chang (National University Health System, Singapore) for their contribution to the mechanistic element of the trial. Their role involved assisting with statistical analyses.

Institutional support

We acknowledge, with thanks, the trial funder: the National Institute for Health Research Efficacy and Mechanism Evaluation programme in the UK.

We thank all those not otherwise mentioned who have contributed to the TABLET trial.

Contributions of authors

Dr Rima K Dhillon-Smith (Academic Clinical Lecturer) co-ordinated the practical conduct of the trial, including leading recruitment at Birmingham Women’s Hospital. She also co-ordinated the management of follow-up and contributed to the TMG. She was responsible for completion of data gathering, providing data quality assurance, co-ordination of the analysis and the writing groups, and produced the final report.

Mr Lee J Middleton (Statistics Lead) designed and conducted the statistical analysis of the primary and secondary trial outcomes, produced reports for the DMC and contributed to the TMG. He was also responsible for writing the methods and results section of the final report.

Mrs Kirandeep K Sunner (Senior Trial Manager) co-ordinated the practical conduct of the trial, including the management of follow-up, and was responsible for completion of data gathering. She also attended meetings of the TSC and contributed to the TMG and preparation of the final report.

Mrs Versha Cheed (Medical Statistician) helped with the design and conducted the statistical analysis of the primary and secondary trial outcomes, produced reports for the DMC and contributed to the TMG. She was also responsible for writing the methods and results section of the draft report.

Mrs Krys Baker (Unit Research Co-ordinator) co-ordinated the practical conduct of the trial and also attended meetings of the TSC. She also contributed to the TMG and preparation of the draft report.

Ms Samantha Farrell-Carver (Trials Assistant) co-ordinated the practical conduct of the trial, including the management of follow-up, and was responsible for completion of data gathering. She also attended meetings of the TSC and contributed to the TMG.

Ms Ruth Bender-Atik (National Director of the Miscarriage Association) provided advice and oversight to conduct patient and public involvement activities and contributed to the TSC.

Dr Rina Agrawal (Consultant Obstetrician and Gynaecologist, Specialist in Reproductive Medicine) was jointly responsible for the TABLET trial recruitment and data collection at University Hospital Coventry Assisted Conception Unit.

Dr Kalsang Bhatia (Consultant Obstetrician and Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at Burnley General Hospital.

Mr Edmond Edi-Osagie (Consultant Gynaecologist and Specialist in Reproductive Medicine) was responsible for leading the TABLET trial recruitment and data collection at St Mary’s Hospital, Manchester.

Mr Tarek Ghobara (Consultant Obstetrician and Gynaecologist, Specialist in Reproductive Medicine) was jointly responsible for the TABLET trial recruitment and data collection at University Hospital Coventry Assisted Conception Unit.

Dr Pratima Gupta (Consultant in Obstetrics and Gynaecology) was responsible for co-ordinating the TABLET trial recruitment and data collection at Birmingham Heartlands Hospital.

Mr Davor Jurkovic (Consultant Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at University College Hospital, London.

Mr Yacoub Khalaf (Consultant Gynaecologist and Specialist in Reproductive Medicine) was responsible for leading the TABLET trial recruitment and data collection at Guy’s Hospital, London.

Dr Marjory MacLean (Consultant Obstetrician) was responsible for leading the TABLET trial recruitment and data collection at Ayrshire Maternity Unit.

Professor Chris McCabe (Professor of Molecular Endocrinology) was involved with the design of the mechanistic study and reviewed the draft mechanistic report.

Dr Khashia Mulbagal (Consultant Obstetrician and Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at Royal Bolton Hospital.

Miss Natalie Nunes (Consultant Obstetrician and Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at West Middlesex University Hospital.

Dr Caroline Overton (Consultant Obstetrician and Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at St Michael’s Hospital, Bristol.

Professor Siobhan Quenby (Professor of Obstetrics) was responsible for leading the TABLET trial recruitment and data collection at University Hospital Coventry.

Mr Rajendra Rai (Consultant Obstetrician and Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at St Mary’s Hospital, London.

Mr Nick Raine-Fenning (Associate Professor of Reproductive Medicine) was responsible for leading the TABLET trial recruitment and data collection at Queen’s Medical Centre, Nottingham.

Dr Lynne Robinson (Consultant in Obstetrics and Gynaecology, Specialist in Reproductive Medicine) was responsible for co-ordinating the TABLET trial recruitment and data collection at Birmingham Women’s Hospital.

Dr Jackie Ross (Consultant Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at King’s College Hospital, London.

Mr Andrew Sizer (Consultant in Reproductive Medicine) was responsible for the TABLET trial recruitment and data collection at Princess Royal Hospital, Assisted Conception Unit, Telford.

Ms Rachel Small (Lead Midwife and Nurse for Early Pregnancy Care) conducted the TABLET trial recruitment and data collection at Birmingham Heartlands Hospital.

Dr Alex Tan (Clinical Research Fellow) was involved with the systematic review and design of the trial.

Mr Martyn Underwood (Consultant Obstetrician and Gynaecologist) was responsible for leading the TABLET trial recruitment and data collection at Princess Royal Hospital, Telford.

Professor Mark D Kilby (Professor of Fetal Medicine) was involved with trial design, contributed to the TMG and reviewed the draft report.

Dr Kristien Boelaert (Consultant Endocrinologist) contributed to the TMG and reviewed the draft report. She also provided oversight of the management of any abnormal test results.

Professor Jane Daniels (Professor of Clinical Trials) was involved with the design and conduct of the trial, contributed to the TMG and was part of the core writing group for the draft report.

Professor Shakila Thangaratinam (Professor of Maternal and Perinatal Health) was involved with the systematic review and the design and conduct of the main trial, contributed to the TMG and reviewed the draft report.

Dr Shiao-Yng Chan (Consultant in Obstetrics and Clinician Scientist) was involved with the design and conduct of the main trial and contributed to the TMG. She also designed and performed the data analysis and write-up for the mechanistic study, as well as reviewing the draft report.

Professor Arri Coomarasamy (Professor of Gynaecology and Chief Investigator of the TABLET trial) designed the trial, chaired the TMG, contributed to the TSC and took overall responsibility for the project. He also edited the final report.

All authors contributed substantially to the development of the research question and trial design, implementation, analysis and/or interpretation of data, and submission of the final report.

Publications

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

Dhillon-Smith RK, Middleton LJ, Sunner KK, Cheed V, Baker K, Farrell-Carver S, et al. Levothyroxine in women with thyroid peroxidase antibodies before conception. N Engl J Med 2019;380:1316–25.

Data-sharing statement

All data requests should be submitted to the corresponding authors for consideration. Access to available anonymised data may be granted following review.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research. The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, the MRC, NETSCC, the EME programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the EME programme or the Department of Health and Social Care.