Risk factors of subclinical hypothyroidism and the potential contribution to miscarriage: A review

Abstract Background There is no clear cutoff value for thyroid‐stimulating hormone (TSH) level that defines subclinical hypothyroidism (SCH). Moreover, TSH levels can be affected by numerous factors. Although mild SCH has been implicated in miscarriage, the relationship between TSH levels and miscarriage remains unelucidated. Methods We reviewed nine known risk factors affecting TSH levels and 28 studies investigating the potential association between mild SCH and miscarriage, examining whether these factors were considered. Main findings Among 28 studies that examined whether mild SCH (TSH > 2.5 mIU/L) contributed to miscarriage, thyroid antibodies were measured in only 15. TSH measurement methods were described in 18 studies. Although the iodinated contrast medium used in hysterosalpingography (HSG) is stored in the body for a long time and is a risk factor for mild SCH, only one study described its potential impact on TSH levels. Nine studies, which concluded that mild SCH contributed to miscarriage, had thyroid status evaluated only after the onset of pregnancy, but not before. Conclusion TSH levels can be significantly affected by patient demographics and health history, country of origin, and fertility treatment. It is important to consider these factors while evaluating mild SCH. It remains unclear how mild SCH contributes to miscarriage.

managing thyroid function in women planning or becoming pregnant. 8,9 The ATA guideline describes that there is insufficient evidence that LT4 treatment contributes to an increased pregnancy rate in SCH women and thyroid autoantibody-positive euthyroid women who are planning a natural pregnancy; however, SCH women are recommended to consider a small amount of LT4 to reduce the risk of hypothyroidism after pregnancy. In contrast, when infertility treatment is performed in in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI), controlling TSH levels to <2.5 mIU/L by LT4 is recommended to improve assisted reproductive technology (ART) outcomes. During pregnancy, the use of a population-based reference for the upper limit of TSH is suggested. If a population-based reference cannot be obtained, it is recommended that the reference value for pregnant women be 0.5 mIU/L lower than the upper limit of nonpregnant women for each TSH measurement method. In particular, pregnant women with thyroid autoantibodies are recommended to consider LT4 supplementation with a target TSH level < 2.5 mIU/L.

Many studies have assessed the effect of mild SCH in early
pregnancy on adverse events of pregnancy with a TSH cutoff value > 2.5 mIU/L.  The secretion of TSH is not only strongly influenced by various health factors, but is also known to have different baseline values due to differences in age and race. 38 There is still a question as to whether TSH > 2.5 mIU/L is effective as a common global index for mild SCH and whether LT4 treatment is needed in all cases where TSH exceeds this value.
First, we review the risk factors that affect TSH secretion.
Next, we review previous studies that examined whether mild SCH both prior to and during pregnancy contributes to miscarriage, and then discuss the relationship between the timing of evaluation for mild SCH and the miscarriage rate. Finally, we try to conclude when and how to evaluate patients who are or may become pregnant for mild SCH.

| RIS K FAC TOR S AFFEC TING TS H
TSH is a sensitive marker of thyroid function. Factors affecting TSH levels should be considered when assessing the presence and potential impact of mild SCH. Here, we review the following factors: thyroid antibodies, age, body mass index (BMI), iodine intake, daily and seasonal changes, TSH measurement methods, hysterosalpingography, ovarian stimulation, and pregnancy related hormones including human chorionic gonadotropin (hCG) and estrogen (Table 1).

| Thyroid antibodies
Anti-thyroid peroxidase antibody (TPOAb) and anti-thyroglobulin antibody (TgAb), which are thyroid autoantibodies, are risk factors for SCH and have been reported to be prevalent in infertile women and miscarriage patients. 38 49 and Vitamin D receptor gene. 50 The disease penetrance of mutations in these genes varies widely with ethnicity. 42

| Age and BMI
TSH is known to be positively correlated with both age 38,52,56,57 and BMI. 58 although Japan is an iodine-sufficient area, UIC < 150 μg/mL, which is defined as mildly iodine deficient, 8 is observed in approximately 30% of pregnant women in Japan. 71 Zimmermann and Delange 72 reviewed pregnant women's iodine intake in several European countries and found that individuals in Sweden and Switzerland were iodine-sufficient, while those in Belgium, Denmark, France, Germany, Hungary, Ireland, Italy, and Turkey were iodine-insufficient.
Even within the same country, iodine intake may vary depending on food culture differences. In China, Li et al 73  States indicated that the basic value of TSH may differ depending on race. 38 This study found that African-American women had significantly lower TSH levels than Caucasian women (1.15 ± 0.02 vs 1.55 ± 0.03) and lower frequency of SCH (TSH > 4.5 mIU/L) (1.5% vs 6.6%). Caucasians have been reported to have higher iodine intake than African-Americans, so the difference in TSH levels among races may be due to the differences in food culture. 75

| Daily and seasonal changes
TSH conforms to circadian rhythm, and TSH secretion increases up to 1.5 times at nighttime compared with daytime. 76 This circadian rhythm is age-related, 77 shows large interindividual variability, 78 and has been reported to be disturbed by an irregular sleep-wake cycle. 79 In addition, eating is known to act on the pituitary-thyroid axis and reduce TSH secretion. 80

| TSH measurement methods
Currently, third-generation TSH assays are the main method for screening for primary hypothyroidism. Most of these assays are two-site "sandwich" immunoassays that use an enzyme or chemiluminescent label. 89

| Hysterosalpingography
In hysterosalpingography (HSG), an oil-based or water-soluble iodine-containing contrast medium is used. It has been reported that patients receiving oil-based contrast media have a higher cumulative pregnancy rate after HSG than patients receiving water-based contrast media. [93][94][95] However, oil-based contrast media have a longer retention period, which significantly increases the frequency of SCH compared with water-soluble contrast media. 94 after 2-6 months from HSG with an oil-based contrast agent. 94 We also found that the frequency of SCH is significantly lower in HSG patients receiving water-soluble than oil-based contrast media: 9.5% after 1 month, 3.6% after 2-6 months. This means that the type of contrast medium used and the timing of HSG affect the frequency of SCH. Furthermore, the use of oil-soluble iodinated contrast medium may cause fetal hypothyroidism. 98 However, there are currently no data to suggest that the miscarriage rate increases after HSG with an oil-based contrast medium. Benaglia et al 12 reported that 35% of patients with initial TSH values < 2.5 mIU/L had TSH levels higher than 2.5 mIU/L after ovarian stimulation. Therefore, the question of whether the presence of mild SCH before or after ovarian stimulation contributes to adverse pregnancy outcomes, including miscarriage, is of special importance in patients undergoing ART.

| Changes in thyroid function owing to pregnancy
Gestational age and hormones that regulate pregnancy, including estrogen and hCG, affect thyroid function, including TSH levels, during pregnancy.
Estrogen, which increases during pregnancy, contributes to an increase in thyroxine-binding globulin (TBG) levels. 99  The hCG has a high structural homology to TSH and is known to exhibit TSH-like activity. 99,103,104 Serum TSH shows a transient decline at 9-10 weeks of gestation when serum hCG is at its maximum (~100 000 IU/L) and a gradual increase after the second trimester.
Glinoer 99 estimated that as hCG rises to 1000 IU/L, TSH shows a 0.1 mIU/L decrease. A decrease in TSH level is even more pronounced in twin pregnancies where hCG is higher than in single pregnancies. 15 The trimester-specific range is used as the TSH reference value during pregnancy. [8][9]99 The ATA guideline recommends using a population-based reference range, defined based on local population data, as the upper limit of TSH during pregnancy. 8 If the population-based reference cannot be obtained, it is recommended that the reference value for pregnant women be 0.5 mIU/L lower than the upper limit of nonpregnant women for each TSH measurement method. Abbreviations: ART, assisted reproductive technology; CLEIA, Chemiluminescent enzyme immunoassay; CLIA, Chemiluminescence immunoassay; ECLIA, Electro-chemiluminescence immunoassay; ET, embryo transfer; hCG, human chorionic gonadotropin; HSG, hysterosalpingography; IUI, intrauterine insemination; IVF, in vitro fertilization; LT4, levothyroxine; RPL, recurrent pregnancy loss; SCH, subclinical hypothyroidism; TAI, thyroid autoimmunity. a Patients with TSH > 2.5 mIU/L received LT4 before ovarian stimulation, but TSH level measurement did not follow. b Thyroid antibody test was performed in the study group. c Urinary iodine concentration was measured. i Sixty-one out of 89 pregnant women who had TSH > 2.5 were treated with LT4. j Thyroid function tests were performed before HSG.

| DOE S MILD SCH CONTRIBUTE TO MISC ARRIAG E?
We first selected 28 studies from the PubMed search of the relevant literature using TSH > 2.5 mIU/L as the cutoff value and the miscarriage rate as the outcome (Tables 2 and 3: one study 31 is   included in both Tables 2 and 3 Since the women studied were primarily of childbearing age, mostly in their thirties, we did not expect age to be a significant factor in the development of mild SCH in these particular studies.
However, the contribution of mild SCH to miscarriages may increase with advancing age and obesity as they are known to be risk factors. 16 Therefore, these risk factors need to be considered in this context.
Iodine intake and average temperature can be assessed according to the country in which the study was conducted. The iodine intake status of a country or region can be found in the WHO report. 17 Whether to use different TSH cutoff points for countries with sufficient vs insufficient iodine intake, or for regions with high vs low average temperature, is an interesting question that requires further consideration. Only two studies 18,31 out of the 28 studies included measurement of urinary iodine. In common medical practice, however, it is not practical to assess urinary iodine in all cases.
Our review has established that the measurement method for TSH affects its value. This means that when TSH > 2.5 mIU/L is used as the cutoff value, the frequency of mild SCH diagnosis might vary due to the differences in TSH measurement methods. Therefore, the notation of the TSH measurement method is important and should be included in research studies. Of the 28 studies reviewed here, 19 studies reported a description of the TSH measurement method (67.9%). One study only described the measurement method as "third-generation." Currently, most TSH measurement methods are of the third generation. Since there are several types of third-generation measurement, merely describing the method as "third-generation" is insufficient. Furthermore, it should be noted that even if the TSH measurement principle is the same but the assay kits' components are different, TSH values may be affected. HSG using iodine contrast media is one of the causes of transient mild SCH. However, only one study (our previous study 33 ) considered the timing of TSH measurement relative to HSG. In addition, there was no report other than ours that indicated the type of contrast medium (oil-soluble or water-soluble) used. As mentioned earlier, when an oil-based contrast medium is used, the storage period in the body is more than half a year. 96 Therefore, if a patient begins intrauterine insemination (IUI) or IVF treatment several months after an HSG, the retention of iodine may cause transient mild SCH.
Thus, at least the type of contrast medium used and the time relationship between TSH measurement and HSG should be indicated in research studies.
An increase in serum E2 concentration due to ovarian stimulation in IVF contributes to an increase in TSH concentration, whereas an increase in serum hCG concentration in early pregnancy leads to a decrease in TSH concentration. Therefore, information on the time when TSH was measured is important in evaluating mild SCH. Of the 28 studies, 27 described clearly the timing of TSH measurement relative to fertility treatment or weeks of pregnancy (96.7%). Among 25 studies comparing miscarriage rates in euthyroid women (TSH < 2.5 mIU/L) and women with mild SCH (TSH > 2.5 mIU/L), six have found an increase in miscarriage rates in the latter group (Table 2). Notably, all of them were diagnosed as having mild SCH based on TSH values after the onset of pregnancy. Conversely, in women who underwent IUI, Karmon et al 22 reported a lower miscarriage rate in women with mild SCH than in euthyroid women. However, this study evaluated mild SCH before ovarian stimulation, and TSH values in early pregnancy are unknown. Taken together, these results suggest that the presence of mild SCH in early pregnancy may be a stronger contributor to miscarriage than the presence of mild SCH when planning pregnancy. In addition, of the four studies that evaluated the effect of LT4 on mild SCH, three studies showed a reduction in miscarriage rate with LT4 treatment and they included patients with mild SCH in early pregnancy. The fourth study that did not show any benefit from LT4 treatment included patients with mild SCH before pregnancy. Since there is no compelling evidence that developing mild SCH before pregnancy contributes to miscarriage, there is also no evidence to support the recommendation that all patients with TSH > 2.5 mIU/L should receive LT4 treatment before pregnancy. TSH in women during early pregnancy is lower than that in non-pregnancy women who are planning pregnancy. Therefore, if the same cutoff value (TSH 2.5 mIU/L) is used in nonpregnant and pregnant women, aggressive treatment of nonpregnant women with TSH > 2.5 mIU/L may be an overtreatment among women with low-risk miscarriage. Thus, changes in TSH levels before and after pregnancy make it difficult to evaluate SCH in women who wish to become pregnant, and the timing of starting LT4 supplementation is being discussed.
Whether mild SCH contributes to adverse pregnancy events, including miscarriage, is a topic of great interest, and many studies conducted globally have addressed it. In order to allow other researchers to accurately compare and evaluate their data across studies, it is necessary to provide sufficient information on the exact method and timing of the evaluation for mild SCH.

| CON CLUS ION
TSH secretion is influenced by many factors. When evaluating mild SCH in infertility treatment, the effects of HSG and ovarian stimulation cannot be ignored. Our review has demonstrated that many factors affecting TSH should be considered when evaluating a patient for mild SCH, including the timing and exact method of TSH measurement, as well as the age, diet, geographical location, and ethnicity of the patient. The review of the 28 studies considered here indicates that all reports of mild SCH contributing to miscarriage involved a diagnosis in early pregnancy, as opposed to before pregnancy. The question of whether TSH > 2.5 mIU/L is an appropriate cutoff value for the diagnosis of mild SCH remains controversial, and the essential problem is the handling of so-called "gray area cases" of SCH (such as when TSH < 4.5 mIU/L). It is necessary to continue to study how mild SCH contributes to adverse events in infertility treatment and pregnancy.

ACK N OWLED G EM ENTS
We would like to thank Editage (www.edita ge.jp) for English language editing.