Effects of initiation time of levothyroxine therapy in women with gestational subclinical hypothyroidism and negative thyroid peroxidase antibodies on the neurological development of offspring

Subclinical hypothyroidism (SCH) is characterized by an elevated thyroid-stimulating hormone (TSH) with normal thyroxine (T4) levels. The 2017 American Thyroid Association (ATA) recommends that an upper reference limit of 4.0 mIU/L should be used in the absence of pregnancy-specific TSH ranges.¹ It is well known that overt hypothyroidism exerts a profound effect on pregnancy outcomes; conflicting data have been reported regarding the association between SCH and the incidence of adverse pregnancy outcomes. To date, various observational studies have shown that SCH is associated with adverse pregnancy outcomes, including preterm delivery,² gestational diabetes,³ gestational hypertension, eclampsia,⁴ low birth weight, low Apgar score, and lower childhood IQ.^{5, 6} Additional studies have not demonstrated the association of adverse outcomes with SCH,^{7, 8} although the status of antithyroid peroxidase antibodies (TPOAbs) was not differentiated in these studies. Previous studies have shown higher impairment of positive TPOAbs (TPOAb⁺) on neurocognitive outcomes in offspring.⁹ The adverse impact of negative TPOAbs in gestational SCH (SCH-TPOAb⁻) on offspring development has not yet been identified.

No consensus has been demonstrated on the treatment of pregnant women with SCH-TPOAb⁻. The Endocrine Society recommends therapy in all pregnant women with SCH, irrespective of their TPOAb status.¹⁰ The 2017 ATA supports treatment for a specific subgroup of women with SCH who are TPOAb⁺ (SCH-TPOAb⁺) or TPOAb⁻ and possess TSH levels greater than 10 mU/L.¹ The application of levothyroxine (L-T4) treatment for SCH-TPOAb⁻ is somewhat ambiguous (TSH cutoff 4.0–10.0 mIU/L). Moreover, a limited number of studies have investigated the influence of the time frame on the treatment effect in SCH during pregnancy. Previously, it was reported from our group that standardized treatment should be used for SCH-TPOAb⁻ pregnant women prior to 8 gestational weeks; their TSH levels ranged from 4.0 to 10.0 mIU/L, which might significantly improve the intellectual development of the 2-year-old offspring.¹¹ It was hypothesized that the later treatment of patients with negative TPOAb and TSH levels between 4.0 and 10.0 mlU/L (4.0 mIU/L < TSH ≤ 10.0) would have greater adverse effects on the neurobehavioral development of their offspring.

To assess this hypothesis, electronic health records and abstracted clinical data from patients who were reviewed at the Second Affiliated Hospital of Wenzhou Medical University were retrieved from June 2016 to June 2019. Pregnant women were routinely screened at their first antenatal visit for thyroid function, such as TSH, free T4, antithyroid peroxidase, and the presence of antithyroglobulin antibodies. Women with a serum TSH concentration between 4.0 and 10 mIU/L and normal levels of free T4 and TPOAb were considered to be SCH-TPOAb⁻. Women were diagnosed with SCH-TPOAb⁻ and voluntarily selected to receive L-T4 on the second day following their diagnosis; alternatively, they did not receive levothyroxine during the total gestation period. The levothyroxine regimen was initiated at a dose of 25 µg/day. Thyroid function was evaluated every 2 weeks, and the dose was adjusted to maintain TSH levels between 0.5 and 2.0 mU/L. The protocol was reviewed and approved by the ethics and research committee of the Second Affiliated Hospital of WenZhou Medical University (2021-K-84-02). All subjects provided written informed consent. All subjects’ offspring were followed up for intellectual development until the age of 2. The intellectual development was assessed and compared using the Gesell Development Diagnosis Scale (GDDS). Descriptive statistics are presented as mean ± SD (x ± s) for normal variables. Analysis of variance was applied to determine differences among three or more groups of patients for normally distributed data. In addition, a Fisher's LSD test was employed for post hoc correction. The Spearman correlation analysis was adopted to evaluate the correlation among non-normally distributed data.

From June 2016 to June 2019, 12,971 pregnant women underwent thyroid function tests (including thyroid antibody tests) prior to the 16th week of pregnancy. In total, 402 (3.10%) women were diagnosed with SCH-TPOAb⁻ and voluntarily selected to receive L-T4 on the second day following their diagnosis or did not take levothyroxine during the total gestational period. Among the 245 pregnant women in the SCH treatment group, 105 were excluded due to treatment abandonment, failure to follow the doctor's advice to adjust the LT4 dose, non-regular thyroid function screening, or other reasons. Moreover, among the 157 pregnant women in the SCH observation group, 105 were excluded due to midway treatment, non-regular thyroid function screening, or other reasons. Eventually, the remaining 192 pregnant women met the inclusion criteria of the present study. According to the time period of L-T4 treatment, SCH-TPOAb⁻ pregnant women were divided into four groups as follows: group A (n = 41) who were treated with L-T4 prior to 8 gestational weeks, group B (n = 54) who were treated with L-T4 in the period 8 + 1 to 12 gestational weeks, group C (n = 45) who were treated with L-T4 in the period 12+1 to 16 gestational weeks, and group D (n = 52) who received no treatment. A total of 58 pregnant women who were TPOAb-negative during their pregnancy and whose serum TSH levels were detected at the cutoff (0.1–2.5 mIU/L) served as the control group (group E). Finally, 26 (63.4%), 28 (51.9%), 25 (55.6%), 28 (53.8%), and 27 (46.6%) children completed the GDDS assessment at 24 months in groups A–E, respectively.

The developmental quotient (DQ) scores of groups A–E were 101.33 ± 8.79, 95.94 ± 9.03, 94.12 ± 6.67, 92.49 ± 7.89, and 101.16 ± 9.14, respectively. DQ scores in group A were 0.17 higher than those in group E (p > 0.05), and DQ scores in groups B–D were 5.22, 7.04, and 8.67 lower than those in group E (p =  0.022, 0.003, and 0.000). Specifically, the difference in DQ between groups B and E was mainly reflected as follows: fine motor quotient (FMQ) was lower by 7.79 (p < 0.05), and language quotient (LQ) was lower by 8.31 (p < 0.05). The difference in DQ between groups C and E was mainly reflected with regard to the following aspects: FMQ was lower by 8.38 (p < 0.05), adaptability quotient (ABQ) was lower by 7.72 (p < 0.05), LQ was lower by 10.19 (p < 0.05), and individual social behavior quotient (ISBQ) was lower by 7.85 (p < 0.05). The difference in DQ between groups D and E was mainly reflected with regard to the following aspects: Gross motor quotient (GMQ) was lower by 7.00 (p < 0.05), FMQ was lower by 8.33 (p < 0.05), ABQ was lower by 9.08 (p < 0.01), LQ was lower by 13.86 (p < 0.01), and ISBQ was lower by 8.34 (p < 0.05). No statistically significant differences were noted in DQ, including GMQ, FMQ, ABQ, LQ, and ISBQ, between groups A and E (Table 1). The results of Spearman's rank correlation analysis demonstrated that DQ, ABQ, LQ, and ISBQ correlated negatively with LT4 initiation treatment time (r  =  −0.328, −0.304, −0.378, −0.309; p < 0.05).

Thyroid hormone (TH) is essential for brain cell proliferation. The fetal thyroid gland forms at week 12 of gestation and becomes increasingly functional at later stages of development. Prior to gestational week 20, TH-dependent brain development fully or partly depends on maternal TH.¹² An insufficient supply of maternal TH during this period could cause significant and irreversible neurodevelopmental defects. Maternal hormonal levels are expected to become irrelevant approximately following 10–12 weeks of gestation if the fetal thyroid produces the normal increasing amounts of thyroxine, which is absent in autoimmune diseases or congenital fetal hypothyroidism.¹³ Two large randomized controlled trials investigating the effect of treatment in women experiencing SCH failed to depict substantial improvement in the cognitive function of the offspring.^{14, 15} However, the results were limited by the late-onset intervention (i.e., 16.7 and 13.4 weeks of gestation) that missed the optimal time for fetal brain development. In the current study, women who received replacement therapy prior to 8 weeks of gestation demonstrated superior results in offspring neurobehavioral development compared with those noted in untreated or late-treated women. No statistically significant differences were found in DQ about the values of GMQ, FMQ, ABQ, LQ, and ISBQ compared with those of the normal control group. In case the optimal time for treatment was missed, the neurobehavioral development of offspring would be significantly affected. Moreover, the delay in initiating L-T4 treatment caused an additional aggravation of this adverse effect. The DQ scores of the offspring whose mothers received L-T4 intervention within 8–12 weeks of gestation (group B) and within 12–16 weeks of gestation (group C) decreased by 5.22 and 7.04 (all p < 0.05), respectively, compared with those of the normal control group. The L-T4 treatment was not performed during gestation (group D), and the DQ of the offspring was lower by 8.67 compared with the respective value obtained in the normal control group (group E). Spearman's rank correlation analysis demonstrated that DQ, ABQ, LQ, and ISBQ correlated significantly and negatively with LT4 initiation time.

The present study indicates that early standardized L-T4 treatment in pregnant women with SCH-TPOAb⁻ and TSH levels between 4.0 and 10 mIU/L may improve the intellectual development of the approximately 2-year-old offspring; larger multicenter studies and prospective approaches are required to confirm and extend these findings.

The authors declare no conflicts of interest.

The National Natural Science Foundation of China, 81771555; the Research Fund for Lin He Academician New Medicine, 19331209; the WenZhou Science and Technology Commission project, Y2023011