Overtime trend of thyroid hormones and thyroid autoimmunity and ovarian reserve: a longitudinal population study with a 12-year follow up

Ovarian reserve is vital for reproductive function, and women with reduced ovarian follicles are at an increased risk of premature ovarian failure (POF) [1]. The status of ovarian reserve can be precisely measured by anti-Müllerian hormone (AMH), which is secreted by the granulosa cells of ovarian follicles [2]. Among the various factors assumed to be associated with POF [2, 3], thyroid function status and thyroid autoimmunity have in particular, been suggested [4]. Thyroid dysfunction influences the reproductive system directly by affecting oocytes, via thyroid hormone receptors on the surface of these cells [4], and indirectly through increasing prolactin secretion and disruption of GnRH function [5]. In addition it has been shown that thyroid autoimmune disease may be associated with a general autoimmunity, leading to premature ovarian failure; however the main pathophysiology linking ovarian reserves with TPO antibody and thyroid hormones has not been elucidated [6‐8]. In a cross sectional study of 5000 Belgian women (mean age: 32.0, SD: 5.5 years), there were no significant association, between ovarian reserve, thyroid hormones and thyroid autoimmunity [9]. Several studies also show an inverse association between ovarian reserves, hypothyroidism and thyroid autoimmunity in humans and animals [10, 11]; however to the best of our knowledge, the interaction between changes in thyroid functions and ovarian reserves status has not yet been reported and it is unclear whether or not women with poor ovarian reserves are at increased risk of occurrence of hypothyroidism.

We aimed to assess the trend of changes in thyroid hormones and TPO Ab among reproductive aged women by various ovarian reserve statuses, identified by age- specific AMH levels in a population based cohort study, the Tehran Thyroid Study, with an average 12 years of follow up.

Figure 1 illustrates the study flowchart; of 4174 participants of the Tehran Thyroid Study, there were 1502 women, aged between 20 and 50 years; among these, 775 met our eligibility criteria. At baseline, 203, 181, 201 and 190 subjects were in the first (Q1 or lowest level), (Q2) second, (Q3) third and fourth (Q4 or highest level) quartiles of age-specific AMH, respectively; their baseline characteristics are presented in Table 1. Initially, there was no statistically significant difference in the anthropometric measures, diastolic blood pressures, parity, smoking status, educational level, and thyroid hormone levels among participants of various quartiles. Women in the first quartile had higher levels of TPO Ab; 7.85(3.69, 23.3) IU/ml in comparison with those in the 2nd, 3rd and 4^thquartiles 5.59 (3.10, 18.66), 5.14 (3.01, 11.35), 6.32 (3.37, 13.89), respectively; p-value = 0.02.

Table 2 demonstrates the prevalence of various types of thyroid dysfunction among participants with different AMH quartiles at initiation of the study; there was no significant difference between study groups in terms of thyroid dysfunction.

According to GEE analysis (Table 3), after adjustment for BMI, parity, smoking status, and educational level, there was a statistically significant annual decrease in mean changes of FT4 in all AMH quartiles; these decreases were [− 0.28(95%CI: -0.43, − 0.1); p < 0.001], [− 0.18, (95%CI: -0.34, − 0.01); p = 0.02], [− 0.25 (95%CI: -0.42, − 0.10); p = 0.001], [− 0.25 (95%CI: -0.41, − 0.09) pmol/L; p = 0.002] for 1st, 2nd, 3rd and 4th quartiles of AMH respectively. There was a statistically significant annual increase in TPO Ab in women of 1st quartile of AMH (10.66 ± 3.60 IU/ml), after adjustment for mentioned variables; the odds of TPO Ab positivity (TPO Ab+) in these women increased by 8% (95% CI: 0.06, 0.16) per year; these significant changes in TPO Ab was not observed in other quartiles of AMH (Table 3). There was no statistically significant difference in mean changes of TSH in all age-specific AMH quartiles (Table 3).

Based on GEE analysis, overall odds of TPO Ab+ in the first quartile of AMH was 2.05 fold higher than those of the fourth [95% CI:(1.31–3.74); p = 0.01], while the overall mean changes of TSH (p-value = 0.79), showed no statistically difference between these two quartiles, after adjustment for BMI, parity, smoking, and education level (Table 4). The interaction (follow-up year × AMH quartile status) in women in the first and the fourth quartiles of AMH was not statistically significant for any of the thyroid parameters (p-value > 0.05).

Figure 2 a, b, and c illustrate trends of TPO Ab, TSH and FT4 in women in the first and fourth quartiles of AMH, adjusted for BMI, parity, smoking, and education level. TSH had a significant increase in women in the first quartile (mean TSH = 2.4 mu/L in the first visit vs. 5.2 mu/L in the last visit, P_trend = 0.04); these values were 2.25 and 3.25 mu/L, P_trend < 0.001 for the first and last visits in the fourth quartile of AMH. However, mean changes of TSH in women in the first quartile were not significantly different from women in the fourth quartile (P_interaction = 0.30)

Although FT4 showed a significant decrease in women in the fourth quartile (mean FT4 = 15.5 pmol/L in the first visit vs. 14.28 pmol/L the last visit,

< 0.001), FT4 of women in the first quartile showed no significantly different value overtime (P_trend = 0.14).

Interaction between follow-up years and AMH quartile demonstrate that as time progresses, changes in thyroid hormones of women in the first quartile of AMH do not differ or become significantly worse compared to those in the fourth quartile (Fig. 2).

There was no significant difference in incidence of subclinical hypothyroidism [HR:1.34 (95% CI:0.52,2.12)], overt hypothyroidism [HR:0.78(95% CI:0.34,1.80)] subclinical hyperthyroidism [HR:1.97(95% CI:0.59,6.63)], overt hyperthyroidism [HR:0.63(95% CI: 0.14,2.86)] and TPO Ab positivity [HR:1.49 (95% CI:0.73,3.03)] between those of the 1st versus those of the 4th quartile. Incidence rates of subclinical hypothyroidism in the 1st, 2nd, 3rd and 4th quartiles were [20(95% CI:10,27), 15(95% CI:10,23), 5 (95% CI:3,10); p = 0.002] and 14(95% CI: 10,21) per1000 person-years, respectively. This incidence in the 3rd quartile was significantly lower than in others, (p = 0.002).

The long term prospective population-based Tehran Thyroid Study provides a unique opportunity to investigate associations between overtime trend of thyroid functions and the ovarian reserve status. In the current study, we found that women with lower ovarian reserves had higher levels of TPO Ab at baseline; moreover a positive trend of this antibody was observed in these groups of women compared to those ones with better ovarian reserve status; there were more incidents of thyroid autoimmunity in the former group.

There are several studies with controversial results on association between hypothyroidism and thyroid autoimmunity with low ovarian reserve [9, 11, 21, 22]; however effects of low ovarian reserve on the overtime trend of thyroid function have not been investigated yet.

In agreement with our findings, Chen et al. (2017) in a cross-sectional study of 1044 infertile Chinese women demonstrated that idiopathic low ovarian reserve with lower serum concentration of AMH was associated with more frequent positive TPO Ab rather than thyroid function or Tg Ab positivity [21]. Saglam et al. in a case control study found that even after age adjustment, autoimmune thyroid disease(AITD) was independently associated with AMH, which was lower in AITD women than in controls [22]. In contrast, Polyzos et al. (2015) in a large retrospective cross-sectional analysis reported that thyroid autoimmunity was not associated with low ovarian reserve [9]. In an animal study, hypothyroidism in adult female rats was induced by an iodide deficient diet in combination with perchlorate supplementation for inhibition of iodide uptake by the thyroid. This condition in adult females negatively affects the ovarian follicular reserve and the size of the growing follicle population [11]. Moreover, some studies report that elevated serum concentration of TSH is associated with decreased serum level of AMH [4, 23, 24]. Weghofer et al. in a study of 225 infertile women reported that even after adjustment for thyroid autoimmunity and age, TSH < 3.0μIU/mL was associated with significantly better ovarian reserve and higher AMH than TSH ≥3.0μIU/mL [23]. In another study among 23 pairs of infertile and fertile women, it was found that both TSH levels and age were negatively correlated with AMH levels in infertile participants with standardized partial regression coefficient (β) of − 0.534 and − 0.361, respectively, but not in normal fertile ones [4]. This finding is in agreement with ours as we also found no association between serum concentration of TSH and ovarian reserve status among participants who were mostly fertile; since thyroid autoimmunity precedes thyroid dysfunction [25], it may need a longer follow up to cause significant elevation in TSH levels among women with thyroid autoimmunity.

The underlying pathophysiological mechanism of the association between thyroid autoimmunity and ovarian reserve status is not completely understood. In the Monteleone et al. study [26], for the first time the presence of thyroid antibodies in ovarian follicular fluid was demonstrated and a significantly lower oocyte fertilization and percentage of grade A embryos was found in infertile women with thyroid autoimmunity undergoing IVF compared to controls. Although this data are preliminary, but it may be assumed that TPO Ab that passes through the blood follicle barrier during follicular evolution may result in the destruction and damaging of growing follicles and oocytes via thyroid hormone receptors, on these cells [26]; a possible mechanism which needs to be confirmed by further comprehensive studies. Except for autoimmunity, thyroid dysfunction may influence the reproductive system via thyroid hormone receptors on the surface of oocytes [4], or through disruption of GnRH function due to increasing prolactin secretion [5].

Furthermore, in the present study, regardless of ovarian reserve status, we found that serum concentration of TSH had an incremental trend by age. In cross-sectional studies of individuals without thyroid disease, serum TSH concentrations increased with age [27‐29], although this age-related TSH increase was not accompanied by a fall in free T4, suggesting an alteration in the TSH regulatory system [28, 30]. In the present study, the decline in FT4 among women with higher ovarian reserves is clinically unimportant. Additionally, some studies indicate hereditary and genetic effects on concentrations of free T4 and TSH, as well as on the negative feedback set point [31, 32].

We showed that the incidence of subclinical hypothyroidism in women of the 3rd AMH quartile was lower than those in other quartiles, even the 4th; this may partly be explained by the including of women with subclinical ovarian dysfunction in the 4th AMH quartile, who had higher AMH levels and regular menstrual cycles, but had polycystic morphology in sonography and subclinical anovulation [33, 34] and also higher rates of thyroid autoimmunity and hypothyroidism [35], despite our effort to recruit fertile women with regular menstrual cycles for the purpose of the present study.

The main strength of the present study is its methodology; it is the first study that has investigated the temporal trend of changes in thyroid hormones, TPO Ab and thyroid disorders among a general population of women with different levels of age specific AMH; in addition the follow up time was 12 years, which seems a sufficient duration for the purpose of the current study. We used age-specific AMH levels, instead of crude AMH levels because this method provides more accurate and precise assessment of ovarian reserve status [36, 37]. The intra-assay and inter-assay in our data is likely to be minimal because all thyroid assessments and AMH assays were performed in the same laboratory by the same expert.

This study had some limitations as well: 1) other methods for estimation of ovarian reserve including: FSH, antral follicular count, inhibin B were not assessed in the study; however AMH has been introduced to be the most reliable [15, 16, 36]. 2) Age specific AMH was identified according to a single AMH measurement at initiation of study and we did not evaluate ovarian reserve over time; however it has been shown that AMH level remains almost constant from one cycle to another and has a high interclass correlation coefficient, as a result of which only one measurement provides a reliable estimate of age specific level [38‐40]. 3) Subjects were not assessed for the subclinical ovarian dysfunction, which may have affected the characteristics of women with higher ovarian reserves [41, 42]. 4) We did not measure Tg Ab in this study and thyroid autoimmunity was evaluated based only on serum TPO Ab measurement; however the presence of TPO Ab in serum is more frequent than Tg Ab [43]; furthermore a large population based study demonstrated a similar prevalence in both antibodies [44] and finally our study assessed only thyroid autoimmunity in relation to age specific AMH levels and autoimmunity in general was not assessed. Previous studies showed that ovarian reserve is closely linked to autoimmunity [22, 37]. In another study premature ovarian failure (POF) was significantly associated with adrenal cortex autoantibodies [45], which may be due to general autoimmunity that affects multiple organs and TPO Ab is a part of this process that has destructive effect on oocytes [26].

Large long-term studies are needed to evaluate both the causality between low ovarian reserve and appearance of thyroid autoimmunity, and the culprits that lead to the autoimmune process in women with limited ovarian reserve.

In summary this study demonstrates that women with low ovarian reserves had both higher baseline and increasing levels of TPO Ab over a12 year follow. They had also a higher incidence of TPO Ab positivity over time. A prospective comprehensive study with several time points measurement of AMH and thyroid assessment is recommended to define the evolution of the association between ovarian reserve status and thyroid function.