Long-term hypothyroidism in patients started on levothyroxine during pregnancy

Introduction

Hypothyroidism diagnosed during pregnancy, whether subclinial or overt, is associated with adverse pregnancy and neonatal outcomes (1). Existing guidelines unanimously recommend the treatment of overt hypothyroidism (OH) in this setting, but some controversy remains about the management of subclinical hypothyroidism (SCH). The European Thyroid Association (ETA) recommends treating SCH with levothyroxine (LT4) to normalize maternal serum thyroid-stimulating hormone (TSH) values within the trimester-specific pregnancy reference range or, when not available, to the following reference range upper limits: first trimester, 2.5 mU/L; second trimester, 3.0 mU/L; third trimester, 3.5 mU/L (2). The more recent American Thyroid Association (ATA) guidelines recommend LT4 treatment in patients with TSH levels above the pregnancy trimester-specific reference range and to be considered in women with thyroid autoimmunity (TAI) with a serum TSH ≥2.5 mIU/L (3), given the association between thyroid autoantibodies, miscarriage, and preterm birth (4). Where pregnancy trimester-specific reference ranges are not available, the ATA suggests an upper limit of about 4 IU/mL for the first trimester, based on newer data on factors influencing the upper reference range of normal TSH. The same cutoff was recently advocated by the ETA guideline for the management of thyroid disorders in women undergoing assisted reproduction (5). In contrast to the ETA, ATA does not recommend the treatment of first-trimester isolated hypothyroxinemia (2, 3).

There are also differences in the postpartum management of patients started on LT4 during pregnancy. According to the ETA, LT4 treatment could be discontinued at delivery in women with TSH less than 5 mU/L and negative thyroperoxidase antibodies (TPOAb) at diagnosis. In contrast, the ATA considers treatment discontinuation in women requiring an LT4 dose of ≤ 50 µg/day during pregnancy, after a discussion between the patient and the caregiver (2, 3). Few studies have investigated the evolution of postpartum thyroid function in women for whom LT4 was initiated during pregnancy, but the incidence of persistent hypothyroidism after delivery could be as high as one-third of the patients (6, 7, 8), calling for a more accurate management of LT4 treatment started during pregnancy.

The aim of our study was to determine the postpartum management and long-term evolution of thyroid function in women started on LT4 during pregnancy, as well as the predictive factors of persistent hypothyroidism.

Materials and methods

We retrospectively analyzed the medical records of 985 pregnant women treated with LT4 during pregnancy and followed at a single tertiary center between January 2014 and December 2020 in Brussels, Belgium. From these women, 500 were excluded because of LT4 initiation before pregnancy, 94 patients for known thyroid disease or taking drugs interfering with thyroid function, and 214 patients because of missing data. Specifically, 149/214 (69%) because of no information on LT4 dose (these were patients seen in our hospital during the last trimester of pregnancy, with treatment initiated in another hospital), 10/214 (5%) because of no information on thyroid function at LT4 initiation, 42/214 (20%) due to no information about the timing of LT4 initiation (before or during the pregnancy), and 13/214 (6%) because of follow-up during pregnancy in another hospital.

Patients treated with LT4 during or outside a previous pregnancy in whom the treatment was withdrawn were included. The final analysis included 177 patients in whom LT4 was initiated during pregnancy. In total, 106 of these patients were followed for 6 months or more after delivery and analyzed for ‘long-term’ persistent hypothyroidism. Figure 1 illustrates the study selection process in detail.

View Full Size

Flowchart of patients started on LT4 during pregnancy and their postpartum long-term thyroid status. LT4: levothyroxine; long-term follow-up: at least 6 months postpartum follow-up.

Citation: European Thyroid Journal 13, 3; 10.1530/ETJ-24-0051

• Download Figure
• Download figure as PowerPoint slide

Results

Predictive factors of persistent hypothyroidism

In the univariate logistic regression, TSH level at LT4 initiation and the maximal LT4 dose during pregnancy were predictive for the risk of long-term hypothyroidism (OR = 1.37, 95% CI: 1.08–1.73, P = 0.009 and OR = 1.04, 95% CI: 1.01–1.06, P = 0.003, respectively), while gestational age, pre-conception BMI, or TPOAb level or positivity at LT4 initiation were not (Table 4). However, only the maximal LT4 dose during pregnancy (OR = 1.03, 95% CI: 1.00–1.05, P = 0.003) was associated with long-term hypothyroidism in multivariate logistic regression. ROC analysis showed that a maximal LT4 dose of 68.75 µg/day during pregnancy (87% specificity, 42% sensitivity; AUC = 0.649, 95% CI: 0.53–0.76; P = 0.013) and a TSH level ≥ 3.8 mU/L at treatment initiation (68.5% specificity, 77% sensitivity; AUC = 0.723, 95% CI: 0.53–0.91; P = 0.019) were the optimal cutoffs for predicting long-term hypothyroidism. (Fig. 2A and B)

View Full Size

Receiver operating characteristic (ROC) curve for maximal LT4 dose (µg/day) during pregnancy (a) and TSH level (mU/L) at LT4 initiation (b) predictive for long-term hypothyroidism. LT4, levothyroxine.

Citation: European Thyroid Journal 13, 3; 10.1530/ETJ-24-0051

• Download Figure
• Download figure as PowerPoint slide

Table 4

Uni- and multivariate logistic regression analysis of factors influencing the risk of long-term hypothyroidism. Values are odds ratio with 95% CI. Values in bold indicate statistical significance.

	Univariate		Multivariate
	OR (95% CI)	P	OR (95% CI)	P
Gestational age	0.944 (0.877–1.017)	0.132	–
Pre-conception BMI	0.924 (0.841–1.016)	0.102	–
TSH (mU/L)	1.371 (1.083–1.736)	0.009	1.229 (0.952–1.585)	0.113
TPOAb ULN (continuous)	1.036 (0.969–1.107)	0.301	–
TPOAb positivity (categorical)	1.875 (0.810–4.340)	0.142	–
Maximal LT4 dose during pregnancy	1.040 (1.014–1.068)	0.003	1.031 (1.003–1.059)	0.030

LT4, levothyroxine; TPOAb ULN, a ratio of the absolute TPOAb value to the ULN; TPOAb, thyroperoxidase antibodies; ULN, upper normal limit.

Discussion

In our study of a population of pregnant women representative for our area (9), we found that one-third of patients initiated on LT4 during pregnancy had long-term persistent hypothyroidism. However, this proportion could vary according to the severity of hypothyroidism at LT4 initiation and different TSH cutoffs used to define it (6, 7, 8, 10). Shields et al. found that 16.5% of patients with SCH and isolated hypothyroxinemia had persistent high TSH levels > 4.5 mU/L over 5 years postpartum follow-up, but the proportion increased to 24.5% when only women with pregnancy SCH were considered (6). The prevalence of long-term hypothyroidism increases with the increasing cut-off of diagnostic TSH, from 18% in a large Indian study of 467 women with SCH (7) when trimester-specific TSH reference range was used, up to 39% in a recent prospective Chinese study that defined SCH as TSH > 4 mU/L (8). Although we used a lower diagnostic TSH cut-off, the rather high proportion of persistent hypothyroidism found in our study could be explained by the inclusion of a significant proportion of patients with OH. At a much higher TSH level at diagnosis during pregnancy (mean TSH 13.2 mU/L), 64% of patients developed OH over >10 years follow-up in the seminal study by Haddow et al. (11).

Another confounding factor for the prevalence of persistent hypothyroidism is the continuation of LT4 treatment after delivery. In our study, there were significantly more patients who continued the treatment after delivery in the long-term hypothyroid group than in the euthyroid group. The reason for treatment continuation was not always known, and we cannot rule out therapeutic inertia, as almost half of those who continued the treatment had a TSH < 4.0 mU/L at LT4 initiation and received during pregnancy a maximal LT4 dose of 50 µg/day or less, which should have been stopped according to current recommendations (3, 12). These patients account for almost one-third of long-term hypothyroid patients and could lead to an overestimation of the persistence of hypothyroidism. In the study by Linardi et al., the prevalence of persistent hypothyroidism goes down from 75.5% to 28% when only the patients who discontinued the treatment are analyzed (10).

If discontinuation of LT4 treatment after delivery is not considered, this could lead to long-term overtreatment, especially when the indication for LT4 initiation during pregnancy is not justified. In our study, LT4 was started in 13.5% of patients although they were euthyroid according to 2014 ETA criteria. If the 2017 ATA criteria had been applied, 50/177 (28%) patients would have been treated. Recent evidence does not support LT4 treatment in euthyroid women, whether TPOAb was positive or not, including those with a miscarriage history (13, 14). Moreover, LT4 treatment was initiated in 9.5% of patients in our study for isolated hypothyroxinemia, according to 2014 ETA guidelines. Although isolated hypothyroxinemia in pregnancy has been associated with adverse maternal metabolic profile and obstetrical outcomes (15, 16) and intellectual disability in offspring (17), current evidence does not support LT4 treatment for this indication (16).

In our study, we found that the TSH level at treatment initiation and the maximal LT4 dose during pregnancy are predictive of persistent hypothyroidism. Previous studies have found associations between long-term hypothyroidism and TSH > 5 mU/L at diagnosis (6, 7, 10), LT4 dose at the end of pregnancy (7), TPOAb (6, 7, 8), or TgAb (8) positivity at diagnosis and at 6 weeks postpartum follow-up (8), gestational age at treatment initiation (8), multiparity (10), or maternal age (7). TSH was significantly higher at LT4 initiation in patients who were long-term hypothyroid in our study, and a TSH level > 3.8 mU/L was predictive of this evolution. This value is also close to the 4.0 mU/L cutoff beyond which epidemiological studies have shown an association between thyroid function and adverse maternal and neonatal outcomes (18, 19, 20), and beyond which the 2017 ATA guidelines suggest initiating LT4 treatment during pregnancy where pregnancy trimester-specific reference range is not available. In addition, a maximal LT4 dose of 68.57 µg/day was found predictive of long-term hypothyroidism in our study. This dose, larger more than the 50 µg cutoff suggested by ATA guidelines (3), might be explained by the inclusion of OH patients with higher peri-conception TSH levels and greater substitution need during pregnancy. In our long-term hypothyroid patients, as in the study by Li et al. (10), LT4 treatment was initiated at an earlier gestational age, demonstrating the lack of adaptation of these patients to the metabolic stress of the pregnancy.

We were not able to reproduce previous findings about the predictive value of TPOAb regarding long-term hypothyroidism (6, 7, 8). TAI impairs the thyroidal response to hCG (21) as a function of TPOAb and/or TgAb levels (22, 23), with a higher risk of hypothyroidism during and after pregnancy at higher antibody levels (23). Although TPOAb positivity was present in almost half of patients started on LT4 in our study, and this proportion was numerically higher in long-term hypothyroid patients, TPOAb was not associated with persistent hypothyroidism. This could be explained by the fact that most of our patients had TPOAb in the lowest tertiles. In two previous studies that showed an association between TPOAb and persistent hypothyroidism (6, 8), antibodies were measured at a more advanced gestational age than in our study (28 and 22 weeks vs 13 weeks), and the persistence of a significant TPOAb level at a late stage of pregnancy could indicate a more aggressive autoimmune phenotype with a higher impact on subsequent thyroid function.

This study also draws attention to the real-life management after LT4 initiation during pregnancy. Although the control of thyroid function is recommended 6 weeks after delivery (2, 3), almost one-third of our patients had no follow-up, compared to 16.5% in the study by Neelaveni et al. (7) and 45% in the study by Li et al. (8), and slightly more than half had long-term follow-up. In our study, being followed up by an endocrinologist seems to be of added value for the long-term management of these patients. As there were no differences in characteristics at LT4 initiation and during treatment between patients with or without follow-up, one might expect the same long-term evolution in the latter and misdiagnosis of persistent hypothyroidism in women susceptible to being pregnant again. However, a postpartum follow-up might not be necessary in some patients, for instance, in those with isolated hypothyroxinemia, given the low prevalence of persistent hypothyroidism in this group (8, 12). A normal thyroid function at 6 weeks post partum does not exclude the risk of long-term hypothyroidism, which can develop in 28% of patients followed for a median of 11 (7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19) months (8), especially when thyroid antibodies are positive. Consequently, as suggested by the 2014 ETA guidelines, the follow-up should be continued at least for 1 year after delivery.

Our study has several limitations related to its retrospective nature, which could increase the risk of selection and information bias. One-fifth of our pregnant patients on LT4 was excluded because there was not enough information about the reasons, timing, and importance of LT4 treatment. These patients might have had different clinical characteristics from those included. The group of patients included was small, and information on LT4 treatment management after delivery was not always available. We did not study the impact of TgAb on the occurrence of long-term hypothyroidism, as they were measured in only 55% of patients at LT4 instauration, nor the impact of iodine supplementation or ferritin levels as possible confounding factors for thyroid dysfunction during pregnancy. Also, there was not enough information about the occurrence of postpartum thyroiditis (PPT) in our population. Hypothyroidism can persist in up to 40% of women with PPT, especially those TPOAb positive (24), and we cannot rule out that PPT was the cause of LT4 restoration in some of our patients.

Based on our findings, LT4 treatment initiated during pregnancy could be reasonably discontinued in patients with TSH < 4.0 mU/L at diagnosis who received less than 68.75 µg/day LT4. We suggest continuing LT4 treatment in patients diagnosed with OH during pregnancy and in patients with SCH defined as a TSH > 4.0 mU/L during the first trimester of pregnancy, given the high prevalence of persistent hypothyroidism (8). The dose to be continued could be reduced by one-third to one-half of the maximal dose required during pregnancy, depending on the initial TSH elevation, and based on the increased requirement for thyroid hormones during pregnancy (25, 26). If the treatment was prescribed for isolated hypothyroxinemia, it should be discontinued after delivery. Thyroid function should be checked at 6 weeks post partum and until at least 1 year after delivery.

Conclusion

One-third of patients started on LT4 during pregnancy have long-term hypothyroidism, but this proportion could be overestimated by the unjustified continuation of treatment in some patients immediately after delivery. We observe deviations from current recommendations concerning the indications of LT4 initiation during pregnancy, the management of LT4 treatment after delivery, and the follow-up strategy. The TSH level at treatment initiation and the LT4 dose during pregnancy could guide the decision for continuing long-term LT4 treatment.

Declaration of interest

KGP received lecture fees from the IBSA Institut Biochimique SA, Berlin-Chemie AG and the Merck company between 2016 and 2022. From September 2023 on, he is the secretary of the European Thyroid Association (ETA). KGP is on the editorial board of European Thyroid Journal. KGP was not involved in the review or editorial process for this paper, on which he is listed as an author.

Funding

This work did not receive any specific grant from any funding agency in the public, commercial, or not-for-profit sector.

Statement of ethics

This study protocol was reviewed and approved by the local Ethics Committee (Comité d’Ethique Hospitalo-Facultaire des Cliniques Universitaires St-Luc, Université Catholique de Louvain, Brussels, Belgium). Written informed consent was not required due to the retrospective nature of the study.

Data availability statement

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from MCB upon reasonable request.

Author contribution statement

MCB wrote the first draft. SD collected the clinical data and DM and SMC performed all statistical analyses. CD, RF, OA, DM, and FD cared for patients and KGP provided revisions for the manuscript. MCB and FD led the clinical study. All authors reviewed and approved the final manuscript.