Introduction

Clinic-based studies suggest that hypothyroidism (as documented by high blood levels of TSH, also known as thyroid-stimulating hormone/thyrotropin) may be a comorbid disorder as well as a pain exacerbating factor for primary headache disorders such as migraine. Interestingly, headache attributed to hypothyroidism was recently classified by the International Headache Society in the headache categories [1, 2]. The diagnostic criteria for headache attributed to hypothyroidism require pain resolution to have occurred within 2 months after effective treatment of hypothyroidism [3]. Although there are original and review articles describing the hormonal levels relating to thyroid function in terms of TSH, 3,5,3′-triiodo-l-thyronine (T3), and 3,5,3′,5′-tetraiodo-l-thyronine (T4) blood concentrations and migraine pathogenesis, no meta-analysis study has yet been carried out on this subject. Thus, the purpose of the current study was to evaluate the hormonal levels relating to the functional state of the thyroid gland in migraineurs in terms of serum TSH, T3, and T4 levels for migraineurs in comparison with non-migraineurs, using a systematic review and meta-analysis of original case-control studies.

Materials and methods

Protocol of the systematic review and meta-analysis

In the current study, in order to conduct a systematic review and meta-analysis of hormonal levels relating to the functional state of the thyroid gland in terms of serum TSH, T3, and T4 levels in migraineurs in comparison with non-migraineurs, the PRISMA 2009 Checklist was applied [4].

Information sources and search strategies

The online Google Scholar, Scopus, Embase, MEDLINE, PubMed, and PubMed Central biomedical databases were systematically searched by two authors (D. P-F and G. B.) independently, without restriction to the start date, and until April 2020. Published studies using the search terms “TSH,” “thyrotropin,” and “thyroid stimulating hormone,” in combination with “migraine” and “migraineur/s,” with additional keywords including “headache/s,” “hormone,” “plasma,” “serum,” “blood samples,” “circulatory,” “level/s,” and “case-control study” were identified in a systematic search manner in the aforementioned databases. Similarly as regards serum T3 concentrations, the systematic search in the aforementioned databases was performed using the terms “T3” and “triiodothyroxine” in combination with “migraine” and “migraineur/s,” with additional keywords including “headache/s,” “hormone,” “plasma,” “serum,” “blood samples,” “circulatory,” “level/s,” and “case-control study.” Similarly, as regards serum levels of T4, the systematic search in the aforementioned databases was performed using “T4” and “tetraiodothyroxine” terms in combination with “migraine” and “migraineur/s,” with additional keywords including “headache/s,” “hormone,” “plasma,” “serum,” “blood samples,” “circulatory,” “level/s,” and “case-control study.” Moreover, the search terms “hypothyroidism,” “hyperthyroidism,” and “thyroid dysfunction,” with additional keywords including “hormone”, “plasma,” “serum,” “blood samples,” “circulatory,” “level/s,” and “case-control study” being identified. The results were restricted to the English language. Moreover, the systematic reviews were independently limited by two of the authors to case-control studies fulfilling the exclusion and inclusion criteria.

Eligibility criteria

Publications were preselected if standardized biochemical methodologies, such as enzyme-linked immunosorbent assay (ELISA), chemiluminescent, and radioimmunoassay (RIA) techniques for concentration determination had been used. Only studies describing circulatory levels of TSH, T3, and T4 levels in the migraineurs and non-migraineurs were considered to calculate the pooled effect size (SMD). The SMD effect size applies when studies report the associated results in terms of a continuous measurement, such as serum TSH hormonal levels. The SMD is also known as Cohen’s d and is calculated by the following formula:

$$ \mathrm{SMD}=\frac{\mathrm{Case}\ \mathrm{group}\ \mathrm{measurand}-\mathrm{comparator}\ \left(\mathrm{controlgroup}\right)\mathrm{measurand}}{\mathrm{pooled}\ \mathrm{standard}\ \mathrm{deviation}} $$

An SMD of zero means that case and control groups have the same mean levels of the measurand. In contrast, if there is a difference between two groups, SMD is lower or greater than zero. Cohen offered the following guidelines for interpretation of the magnitude of the SMD results: small, SMD = 0.2; medium, SMD = 0.5; and large, SMD = 0.8 [5, 6]. The studies that reported the data only by presenting curves and graphs without mentioning the exact mean ± standard deviation (SD)/standard error of the mean (SEM) were also excluded. Moreover, studies which did not report the number of cases and controls were discarded.

Study selection

Only studies in which the levels of TSH, T3, and T4 hormones evaluated for migraineurs in comparison with non-migraineur subjects, i.e., case-control type studies, were considered for inclusion. Moreover, studies providing suitable and sufficient information were chosen in order for the pooled effect size to be statistically computable (including mean and standard deviation or standardized error of the mean, number of migraineurs and controls, etc.). Any disagreement was solved by the third author (A. M.). No limitation was applied to the migraine subtype and pain severity, race, and participants’ gender as reported by the included publications. Furthermore, studies were excluded if they had enrolled individuals other than migraineurs.

Data collection process

Data such as the first author of the included publications, date of publication, serum hormone levels of all individuals in both groups, reported criteria for diagnosing migraine in patients or rejection in non-migraineur subjects, total number of migraine patients and non-migraineurs, and other related information were extracted from the finally included studies.

Summary measures and synthesis of results

Stata version 14.0 statistical software (Stata Corporation, College Station, TX, USA) was used for analyses, reporting, and presenting of the plots in the current meta-analysis. Between-study heterogeneity was evaluated according to the χ2-based Q test and I2 index statistics [7]. In the current study, the estimated SMD of TSH, T3, and T4 hormones for each finally included record and also overall effect size were determined and presented with the 95% confidence interval (CI). The statistical significance of the total SMD was determined by the z-test and was considered statistically significant if p < 0.05. The data extracted from each individual study included in the current meta-analysis are presented in Table 1, and these data were used for the calculation of the pooled SMD effect size as well.

Table 1 The demographic and thyroid function test results extracted from the included studies
Full size table

Risk of bias across studies

In order to determine the risk of bias across included studies, the publications were also scrutinized and evaluated for method validation and data processing. According to the pooled SMD and each study result, the funnel plot was provided in such a way as to evaluate the extent of publication bias in the reported result by each study. Empirically, funnel plot symmetry was considered for interpretation of any publication bias among the finally included studies using visual inspections of the generated bias plot. In the prepared plot, the x- and y-axis represent the standard deviation and logarithm of the effect sizes (log of SMD), respectively.

Results

Study selection for serum TSH, T3, and T4 concentrations

In this investigation, the publication selection processes for TSH, T3, and T4 serum levels have been summarized as flowcharts and presented in Fig. 1. The initial search in the medical databases for TSH, T3, and T4 serum levels retrieved a total of 332, 329, and 330 potentially eligible research publications, respectively. These initial search records were scrutinized for detection of duplicate studies and 263, 260, and 261 papers were removed as being duplicate records for TSH, T3, and T4 serum level studies, respectively. Of the included studies, 62 articles were excluded after being considered irrelevant to the purpose of this study on TSH, T3, and T4 serum level evaluations in migraineurs. Publications were included if they considered the case-control study design. Moreover, due to insufficient data reported by some publications regarding SMD and 95% CI calculations or because of authors’ use of cross-sectional or other types of study design, two, two, and four articles were excluded for TSH, T3, and T4 serum level studies, respectively. Notwithstanding, among all headache types, only migraine type of headaches (without considering migraine subtypes) was included in the current investigation. Finally, five [8,9,10,11], three [8, 10, 11], and five [8,9,10,11] case-control studies, which are detailed in Table 1, were included in the meta-analytical processes for assessment of TSH, T3, and T4 circulatory levels, respectively.

Fig. 1
figure 1

Systematic review process followed for the selection of the studies presenting TSH, T3, and T4 hormone levels in migraineurs in comparison with controls. The numbers reported in parenthesis represent the respective number of studies chosen presenting TSH, T3, and T4 levels, respectively

Full size image

Study characteristics

The reported disease characteristics and epidemiological data were extracted from each finally included paper. The female to male ratio in migraineurs and also in non-migraineur controls and age in cases and controls are presented in Table 1. Finally, a total of 593 migraineurs and 446 non-migraineur controls were retrieved for serum TSH and T4 concentrations in this meta-analysis, up to April 2020. For serum T3 concentrations, the systematic review retrieved a total number of 515 migraineurs and 213 non-migraineur controls, up to April 2020.

Risk of bias within studies

To determine whether significant heterogeneity exists, the p value for the χ2 test of heterogeneity can be considered, a low p value suggesting that the heterogeneity is significant. The p value for the χ2 test of heterogeneity was 0.01, 0.03, and 0.13 for TSH, T3, and T4 serum levels across studies, respectively. Moreover, the I2 test for TSH, T3, and T4 blood levels was calculated as 91.134, 75.310, and 0.000, respectively. Therefore, according to the I2 test for estimation of the overall SMD calculation and presentation of the forest plot of migraineurs in comparison with controls, the random-effects model was applied for TSH and T3 levels and the fixed-effect model was used for T4 levels in migraineurs in comparison with controls.

Synthesis of results

The forest plots for TSH, T3, and T4 serum levels on the basis of the included studies are presented in Fig. 2. In these representations, the mean effect size, standard deviation, and SMD (with 95% CI) for each study and the overall effect size for TSH, T3, and T4 serum levels are presented. The final meta-analytical results showed that the weighted overall SMD for the impact of TSH, T3, and T4 blood levels in migraineurs in comparison with non-migraineur controls were as follows: SMD = 0.804 (95% CI, 0.045–1.564), − 0.267 (95% CI, − 0.660–0.125), 0.093 (95% CI, − 0.077–0.263), respectively. It is noteworthy that only the p value for the significance of the overall SMD for TSH serum level was statistically significant (p = 0.038), as examined by the z-test.

Fig. 2
figure 2

Forest plots for serum TSH, T3, and T4 hormone levels in the case-control studies. In this presentation, pooled data evaluating the effects of serum TSH (a), T3 (b), and T4 (c) concentrations in migraineurs compared with non-migraineur controls have been demonstrated under the random-effects model

Full size image

Risk of bias across studies

Notably, the funnel plots developed by the Stata version 14.0 software were seen to be moderately asymmetrical in shape, demonstrating existence of publication bias in the results reported by the finally included publications for the serum levels of TSH and T4 hormones in migraine subjects as compared with non-migraineurs. For TSH and T4 circulatory levels, the bias is mainly visible in the right-hand part of the publication bias plot indicating higher SMD quantities for migraine subjects as compared with non-migraineurs. However, the publication bias plot was nearly symmetrical for reporting serum T3 levels, demonstrating no publication bias for its levels in migraineurs as compared with controls (Fig. 3).

Fig. 3
figure 3

Funnel plot for serum TSH, T3, and T4 hormone levels in case-control studies. a–c Funnel plots for serum TSH, T3, and T4 hormone levels in the included case-control studies, respectively. The funnel plots appeared asymmetrical for TSH (a) and T4 (c) blood levels in migraineurs in comparison with non-migraineur controls under the random-effects model. However, the funnel plot for T3 (b) appeared symmetrical. In this plot, the x and y-axis represent the standardized mean differences (SMDs) and standard errors, respectively

Full size image

Discussion

In the current study, we investigated the association between the serum levels of TSH, T3, and T4 hormones in migraine subjects in comparison with non-migraineurs in order to explore the functional state of the thyroid gland through conducting a meta-analysis.

Although there are original and review articles describing the functional state of the thyroid gland in migraineurs in comparison with non-migraineurs and the functional state of the thyroid gland in migraineurs has been a matter of debate in recent original investigations, narratives, and systematic reviews, no meta-analysis study has to date been carried out on this aspect. To the best of our knowledge, this report is the first meta-analysis describing TSH, T3, and T4 serum levels in migraineurs vs. non-migraineur controls in order to provide evidence of mild thyroid dysfunction in migraineurs.

While search results for the aforementioned hormones in medical databases retrieved related publications, most of the studies report TSH and T4 blood levels and only three publications report T3 levels. This phenomenon may be due to the previously accepted definition of overt hypothyroidism (also known as clinical hypothyroidism), which is defined as TSH concentrations above the reference interval and free thyroxine concentrations below the reference interval. Subclinical hypothyroidism (also known as mild hypothyroidism), which is commonly regarded as a sign of early thyroid failure, is defined by TSH concentrations above the reference interval and free thyroxine concentrations within the reference interval [12]. It is evident that, employing the above definitions, T3 level would not have been implicated in inducing this particular type of hypothyroidism, and this lower implication could obviate its being reported in the corresponding publications. Five included studies investigated migraine and thyroid dysfunction, especially hypothyroidism, and the association was clearly documented in the majority of them.

Overall, there was great heterogeneity concerning study design, population, and outcomes in the studies reporting TSH blood concentrations across the finally included case-control studies. Moreover, regarding the small number of finally included studies, as occurred for the serum T3 levels in the current study, it may be difficult to interpret the funnel plots and draw inferences as to publication bias and evaluate whether there are missing studies.

In the present meta-analysis, the results showed a statistically significant difference between the migraineurs and the non-migraineur controls, recording higher serum TSH concentrations among migraineurs in contrast with both serum T3 and T4 concentrations, implying a possible link between a dysfunctional state of the thyroid gland and migraine pathogenesis. Hormones synthesized in the thyroid gland are vital for normal development and growth, and most crucially for that of the central nervous system, where their severe deficiency results in cretinism in the fetal and neonatal periods, a disease characterized by mental retardation, ataxia, and deafness. Importantly, these devastating disease manifestations are irreversible if they are not treated soon after birth [13]. A recent study showed that congenital hypothyroidism alters formalin-induced pain response in neonatal rats and induces hypersensitivity to noxious stimuli [14]. Moreover, another study provides basic evidence that the thyroid gland-derived hormones are involved in pain processing via the anterior cingulate cortex, presenting indications that thyroid-derived hormones affect pain/migraine development. Adult onset hypothyroidism induces hypersensitivity to noxious thermal, but not mechanical, stimulus in hypothyroid mice, while, interestingly, pain severity is relieved by T3 or T4 replacement. The underlying mechanisms for the beneficial effect of hormone therapy may involve an improvement of glutamatergic and GABAergic transmission balance in the anterior cingulate cortex of hypothyroid mice [15]. However, in a recent systematic review (which did not include a meta-analysis) on cross-sectional studies concluded that the existing data are at present inadequate to confirm with certainty a relationship between headaches and thyroid dysfunction. According to their analysis, a possible bidirectional association between headaches, and particularly migraine, and hypothyroidism could exist [16].

In contrast with the above-mentioned systematic review, which was confined to qualitative results, in the current study, the functional state of the thyroid gland in migraineurs was explored using meta-analysis to quantify the overall effect size based on TSH, T3, and T4 serum concentrations in the case-control studies using SMD. On the other hand, to the best of our knowledge, reports on thyrotropin releasing hormone (TRH) in migraineurs have been scant. It is noteworthy that in a recent study, we showed hyperprolactinemia in migraineurs [17]; to further clarify the role of the hypothalamic-pituitary-thyroid (HPT) axis and its association with migraine pathogenesis in more detail, we herein underline the importance of investigation into the levels of TRH in migraineurs, as it regulates both TSH and prolactin blood concentrations. Headache attributed to hypothyroidism seems to be rare, whereas it is more prevalent in migraine subjects also affected by hypothyroidism [3].

The Bradford Hill criteria remain one of the most trusted concepts for determining causality between two factors in medical investigations [18]. While one of the Bradford Hill criteria for causality is the strength of association (the effect size) between migraine and thyroid dysfunction, which was quite high in the current meta-analysis, all aspects of the criteria must be considered to accurately interpret causality. The Bradford Hill criteria include nine topics, as follows: (1) strength of association, (2) consistency, (3) specificity, (4) temporality, (5) biological gradient, (6) plausibility, (7) coherence, (8) experiment, and (9) analogy [18]. Since which of the two disorders, i.e., migraine or thyroid dysfunction, comes first must be revealed in accordance with the Bradford Hill criteria, further investigations should be conducted to determine the causality relationship between these two factors. On the other hand, euthyroid sick syndrome (also known as non-thyroidal illness syndrome) is described as abnormal findings on thyroid function tests that occur in the setting of a non-thyroidal illness, without preexisting hypothalamic-pituitary and thyroid gland dysfunction [19]. Recent investigations showed that altered plasma molecules (such as interleukin-1 beta, interleukin-6, and tumor necrosis factor-alpha-alpha) affect thyroid functions in migraineurs [20,21,22].

Conclusions

In conclusion, the results of the current study point to an association between migraine pathogenesis and the changing TSH hormonal levels of migraineurs in comparison with non-migraineurs. However, another possibility which could explain the results of this study is that the authors of the investigations which have been included in the present meta-analysis, at least at the time of publication, did not know the possibility of “headache attributed to hypothyroidism” and, hence, did not exclude such patients from their studies.