J Endocrinol Invest DOI 10.1007/s40618-017-0679-x

REVIEW

Does thyroid dysfunction increase the risk of breast cancer? A systematic review and meta‑analysis Y. Fang1 · L. Yao2 · J. Sun1 · R. Yang1 · Y. Chen3 · J. Tian3 · K. Yang3 · L. Tian1 

Received: 17 November 2016 / Accepted: 24 April 2017 © Italian Society of Endocrinology (SIE) 2017

Abstract  Purpose  To investigate the relationship between hypothyroidism, hyperthyroidism, thyroid hormone replacement, and the risk of breast cancer. Methods We searched the PubMed, Cochrane Library, EMbase, Web of Science, and China Biology Medicine (CBM) databases through June 2016 to identify researches that assessed the relationship between thyroid dysfunction and the risk of breast cancer together with the impact of thyroid hormone substitution treatment on incidence of breast cancer. Quality of evidence was assessed per outcome, using GRADE. Results A total of 13 population-based studies including 24,808 participants were identified as eligible for this metaanalysis. A meta-analysis of 12 researches illustrated that hypothyroidism was not related to the risk for breast cancer Electronic supplementary material  The online version of this article (doi:10.1007/s40618-017-0679-x) contains supplementary material, which is available to authorized users. Y. Fang and L. Yao contributed equally to this work. * K. Yang [email protected] * L. Tian [email protected] 1

Department of Endocrinology, Gansu Provincial Hospital, Dong gang West Road, Lanzhou 730000, Gansu, People’s Republic of China

2

Clinical Evidence Based Medicine Center, Gansu Provincial Hospital, Dong gang West Road, Lanzhou 730000, Gansu, People’s Republic of China

3

Evidence Based Medicine Center, School of Basic Medical Sciences, Lanzhou University, Lanzhou 730000, People’s Republic of China







[odds ratio (OR) = 0.83, 95% confidence interval (CI) 0.64–1.08, P = 0.162]. 10 researches illustrated that hyperthyroidism was also not related to the risk of breast cancer (OR  = 1.03, 95% CI 0.83–1.30, P  = 0.767). The impact of therapy was evaluated in six researches; there was no proof of a relationship between thyroid hormone substitution treatment and breast cancer with an overall OR of 0.83 (95% CI 0.57–1.21, P = 0.965). Conclusions  Our meta-analysis illustrated that thyroid dysfunction may not be related to increased risk of breast cancer as well as the thyroid hormone substitution treatment did not reduce the incidence of breast cancer; while this study has some confounders that might weaken the results of this meta-analysis, we believe that the findings provide valuable information for stakeholders concerned with outcomes in patients with thyroid dysfunction. Keywords  Breast cancer · Hypothyroidism · Hyperthyroidism · Thyroid hormone · Meta-analysis

Introduction Breast cancer is the most common malignancy and the most commonly diagnosed cancer in women, with an estimated 1.67 million cases and 520,000 deaths worldwide in 2012 [1]. Beatson used thyroid extract to treat breast cancer to explore the relationship between breast cancer and thyroid dysfunction in 1896, and found out that they were correlated [2]. Most scholars believed that breast cancer and thyroid disease might have common endogenous hormones or pathogenic factors. Thyroid hormones, including triiodothyronine (T3), are synthesized by the thyroid cells and released into the bloodstream. Thyroid hormone maintains the body’s basal metabolism and promotes differentiation

13



and maturation of the nervous system. Recent studies have suggested that thyroid hormone receptor α (THRα2) expression may be a regulator of signaling in breast cancer, as breast cancer patients with high levels of THRα2 expression have improved survival rates [3]. In addition, studies in recent years have also demonstrated that thyroid hormone can act on the alpha v beta 3 (avb3) receptor protein of the integrin family on the membrane surface and activate the mitogen-activated protein kinase (MAPK) pathway, thus, inducing the phosphorylation of the endoplasmic reticulum (ER) receptor on the nuclear surface, and promoting angiogenesis and breast cancer cell growth through activation of the ER-mediated pathway. Such processes can be blocked by PD98059 (a MAPK pathway blocker) and ICI 182, 780 (which can degrade the ER) [4, 5]. Tetraiodothyroacetic acid (tetrac) is an avb3 blocker which has been shown to inhibit the growth of breast cancer cells, thus, verifying that thyroid hormones can activate the MAPK and ER receptors through the avb3 receptor protein, thus, promoting the growth of breast cancer cells [6, 7]. Smyth [8] considered that regional variations in the incidence of breast cancer may be ascribed to differences in dietary iodine intake and the impacts of iodine on breast tissue cells. Funahashi et al. [9] fed mice with dimethylbenzanthracene-induced cancer with iodine-rich algae, which inhibited breast cancer progression, verifying that iodine had an antitumor effect; they speculated that this effect might be due to up-regulated expression of transforming growth factor (TGF) β in tumor cells, which induced tumor cell apoptosis. Thus, iodine may play significant roles in the pathogenesis of breast cancer. Although animal researches proved that endogenous thyroid function influenced breast development [10, 11], but epidemiologic researches of the thyroid dysfunction and breast cancer etiology had contradictory outcomes. Some researches found that patients with hypothyroidism [12–14] and hyperthyroidism [15–17] had higher risk of breast cancer, while others found patients with hypothyroidism [18] and hyperthyroidism [19] had lower risk of breast cancer. A few researches did not find any correlations between thyroid disorders and breast cancer [20–22]. Besides, further controversy has arisen regarding the possible role of thyroid supplements in the etiology of breast cancer. Kapdi and Wolfe [23] considered that thyroid supplementation increased the risk of breast cancer, especially amongst nulliparous women treated for more than 15 years. Nevertheless, the model and interpretation of this research have been critiqued and subsequent researches could not confirm a relationship between thyroid supplements and incidence of breast cancer [24–26]. There are several meta-analyses that have been published before, such as Sarlis et al.’s study [27], and Hardefeldt

13

J Endocrinol Invest

et al.’s study [28]. Sarlis et al. searched MEDLINE and manual literature searches up to 2001, and included 13 studies. However, they focused on the association between Hashimoto thyroiditis and breast cancer. Hardefeldt et al. searched Mediline, Embase, Pubmed, Current contents connect and Google from inception to 2010. They included seven studies, and the pooled results did not support a relationship between thyroid dysfunction and breast cancer. Although two relevant meta-analyses had been published, the statistical effect remains insufficient due to the lack of strict research types, incomplete database index, and small sample size. Therefore, to evaluate these questions, we conducted a systematic review and meta-analysis to (1) further explore the relationship between thyroid dysfunction and breast cancer; (2) investigate whether the use of thyroid supplements may be associated with an increased risk of breast cancer.

Materials and methods Literature search A systematic search was performed of the PubMed, Cochrane Library, Web of Science, EMbase, and China Biology Medicine (CBM) electronic databases from inception to June 2016 to identify studies which precisely described the proportion of hypothyroidism and hyperthyroidism in breast cancer patients and non-breast cancer population. There was no language restriction on the search. All searches were performed using the combination of medical subject heading terms (MeSH) and free words. The retrieval strategies were determined by performing multiple pre-retrieval assessments. The detailed searched strategies to different databases could be found in Supplementary Appendix 1. Based on the references included in the studies retrieved, we performed a secondary search and expanded the literature retrieval by manual searches. We set breast cancer and non-breast cancer as the case and control groups, respectively, according to the PICO method. Study selection Two authors (Yuan Fang and Jing Sun) were professionally trained to independently review the records. The process of study selection was as follows: (1) the records identified using the search strategy described above were imported into EndNote and sorted to remove duplicate literature from the same databases; (2) study titles and abstracts were screened individually to exclude articles with irrelevant study objects, study types, and intervention measures; (3) for those records which could not be evaluated via titles and abstracts, the full texts were retrieved so as to evaluate

J Endocrinol Invest

them in details based on inclusion and exclusion criteria. Two authors (Yuan Fang and Jing Sun) independently evaluated the included studies and solved disagreements by discussion. A third author (Ruifei Yang) made final decisions regarding study eligibility. Researches which met these inclusion standards were incorporated in the meta-analysis: (1) case–control researches which precisely described the population, included precise definitions of primary hypothyroidism and hyperthyroidism; (2) studies which reported risk point estimates as odds ratios (ORs) or in which the ORs could be calculated from the data; (3) there was the report of 95% confidence interval (CI) or the CI can be calculated from the presented data; and (4) among studies with repeated reports, one of the highest quality was included. The following studies were excluded: (1) letters, case reports, reviews and commentaries; (2) researches of drug-induced hypothyroidism and hyperthyroidism; (3) researches that assessed thyroid dysfunction according to measurement of thyroid hormone levels for women who have breast disease but without prior history of hypothyroidism or hyperthyroid; (4) literature including patients with severe hepatic and renal insufficiency or acute and chronic infectious diseases; (5) literature with repeated reports, poor quality (NOS < 6), or insufficient information.

Quality assessment and data extraction Two authors (Yuan Fang and Jing Sun) independently evaluated the methodological quality of the included researches using the Newcastle–Ottawa Scale (NOS) [29]. The NOS evaluates literature quality from three aspects: selection of study object, inter-group comparability, and exposure factor measurement of the case–control study. Next, data were extracted by two authors (Yuan Fang and Jing Sun) and entered into a standardized data spreadsheet. Characteristics of each article were extracted and summed up, such as publication year, country of origin, surnames of authors, research design, and demographics including family history of breast cancer, hormone replacement treatment, parity, ethnicity, menopausal status, and age. The novel research in this meta-analysis involved the correlation analysis of thyroid dysfunction and thyroid supplements with the risk of breast cancer; the clinical research variables included the univariate and multivariate comparisons of these three groups. These processes were independently conducted by two trained authors, who formally evaluated and performed two preliminary experiments so as to guarantee the consistency and accuracy of the evaluation. Inconsistent conclusions between the two authors (Yuan Fang and Jing Sun) were resolved through negotiation or requesting the opinion of a third author (Ruifei Yang).

Assessment of evidence quality The quality of the evidence was evaluated through the method of the grades of recommendation, assessment, development and evaluation (GRADE) based on meta-analysis results, which could be divided into four quality levels: (1) high quality; (2) moderate quality; (3) low quality; (4) very low quality. In the present study, three outcomes were rated for the quality of evidence, including the association of hypothyroid, hyperthyroid, and thyroid supplements with the incidence of breast cancer. The quality of evidence might be reduced due to the following five factors: risk of bias, publication bias, inconsistency, imprecision, and indirectness. GRADE pro software was utilized to edit and analyze the evidence grades, and for drafting.

Data synthesis Statistical analysis was performed with STATA software, version 12.0 (Stata Corp, College Station, TX). A pooled OR of the effect of thyroid disorders on the risk of breast cancer was calculated using a random effects model. Heterogeneity was evaluated by Cochran’s Q statistic with a P value of <0.05, an indication of high heterogeneity. Heterogeneity could be quantified by I2 statistic with 75, 50, and 25% that, respectively, corresponds to high, moderate, and low heterogeneity. The potential causes of high heterogeneity were explored by performing a sensitivity analysis. The results of the meta-analysis were visualized using forest plots. The significant level was set at α = 0.05.

Results Study selection A total of 4658 original studies were initially retrieved. After removing duplicate citations, 4324 remained, 4080 of which were excluded based on title and abstract for a lack of correlation with the research topic. After full-text review of the remaining 244 selected articles, we excluded 231 studies which did not meet the inclusion criteria. Finally, 13 case–control studies [21, 30–41] were enrolled. Twelve [21, 31–41] of these studies, containing data from 24,571 cases, evaluated the incidence of hypothyroidism with breast cancer, and 10 studies [21, 30–32, 34–36, 38–40] with 21,889 cases evaluated the incidence of hyperthyroidism between breast cancer and the healthy control group or the control group with benign breast disease. The literature screening process is shown in Fig. 1.

13



J Endocrinol Invest

Fig. 1  The flow chart of systematic studies search and selection procedure

Study characteristics Among the 13 included studies, the effective sample size ranged from 65 to 4575 cases (12,218 cases total), six studies [30–32, 34, 37, 41] included fewer than 200 cases, five [33, 36, 38–40] had 200–2000 cases, one study [35] included 2569 cases, and one study [21] included 4575 cases. Three studies [21, 32, 33] reported the ethnicity of the enrolled participants, which were mainly Caucasian. Four studies [21, 33, 35, 38] involved premenopausal (7142 cases) and postmenopausal women (9471 cases). Four studies [21, 30, 33, 39] reported a family history of breast cancer (2015 cases) and no family history of breast cancer (11,119 cases). Two studies [21, 33] reported the administration of estrogen replacement therapy, among which 4519 women received the therapy, while 7780 did not. Among these, two studies [21, 33] showed that estrogen replacement therapy was administered more frequently in the group of subjects with breast cancer group compared to the

13

control group. The basic characteristics of the patients are summarized in Table 1. Hypothyroidism Twelve studies [21, 31–41] included in this meta-analysis to retrospectively analyze the morbidity of thyroid carcinoma among breast cancer and non-breast cancer patients. The total sample size reached 24,571 cases, 12,106 and 12,465 of which comprised case and control groups, respectively. According to a relevant study [33], primary hypothyroidism proved to be protective in terms of breast cancer that could lower the morbidity of breast cancer. The results of the meta-analysis suggested that there was no statistical correlation between primary hypothyroidism and the morbidity of breast cancer (OR = 0.83, 95% CI 0.64–1.08, P  = 0.162) (Fig. 2). However, the I2  = 66.3% and P  = 0.001, indicating heterogeneity, with P < 0.05, suggesting statistical significance.

ND

7

As there was great heterogeneity among studies on hypothyroidism and breast cancer, we conducted subgroup analysis according to countries of study. Eight studies [31, 32, 34–37, 40, 41] conducted in Europe had a total sample size of 9377 cases, including 4679 cases and 4698 controls, respectively. The results of the metaanalysis illustrated that there was no statistical correlation between hypothyroidism and risk of breast cancer (OR  = 0.84, 95% CI 0.58–1.22, P  = 0.369). However, I2 = 0 and P = 0.542. Four studies [21, 33, 38, 39] were conducted in the American and Canadian group, with a total sample size of 15,194 cases (7427 cases and 7767 controls). The results of the meta-analysis indicated that there was no statistical correlation between hypothyroidism and the risk of breast cancer in the American and Canadian group (OR = 0.80, 95% CI 0.56–1.14, P  = 0.220); but I2  = 88.8% and P  = 0.000, suggesting great heterogeneity in this group (Fig. 2). We conducted a sensitivity analysis in this group to explore the heterogeneity, and found that I2 decreased to 0 after excluding the study by Cristofanilli et al. (2005) [33]. Similarly, the pooled result resulted in an OR of 0.93 (95% CI 0.84– 1.04, P = 0.217).

14/223 ND

ND

6 7 7 7 6 7 7 7 6 7 3666/5586 ND ND ND ND ND ND ND ND ND 1231/7666 ND ND 323/1523 ND ND ND ND ND ND 4177/3953 ND ND ND ND ND 1832/3325 330/777 ND ND

4213/1695 ND ND ND ND ND ND ND ND ND

7 6 ND 853/2194 ND 447/1707 ND 803/1416

ND 1805/276

Family history BC+/ no family history BC Premen/ postmen

Parity+ /no parity

Estrogen use+ /no estrogen use

Quality evaluation

J Endocrinol Invest

ND no data, premen premenopausal, postmen postmenopausal, BC breast cancer

56 ± 12 237 Case–control Brazil Freitas et al. [28]

ND ND 57.2 ± 1.4; 57 ND ND 64 55 ND 63 59 9257 350 400 2612 2352 358 5157 1101 250 380 Case–control Case–control Case–control Case–control Case–control Case–control Case–control Case–control Case–control Case–control USA Ireland Ireland USA UK Sweden Italy New York Prague Italy

58.6 ± 13.5 51.6 ± 12.6 130 2224 Case–control Case–control Germany USA

Ditsch et al. [30] Cristofanilli et al. [31] Simon et al. [21] Shering et al. [35] Smyth et al. [34] Brinton et al. [37] Kalache et al. [38] Adami et al. [39] Talamini et al. [33] Miriam et al. [36] Orhan et al. [32] Grani et al. [29]

Age [mean ± SD, or median (range)] years Total population Study type Country References

Table 1  Demographic characteristic of the included populations

Hyperthyroidism Ten studies [21, 30–32, 34–36, 38–40] reported on the relationship between hyperthyroidism and breast cancer, with a total sample size of 21,889 cases (10,766 cases and 11,123 controls). One study [38] concluded that hyperthyroidism increased the incidence of breast cancer; however, the results of the present meta-analysis suggested that women with hyperthyroidism had a 3% higher risk of breast cancer compared to that in the control group (OR = 1.03, 95% CI 0.83–1.30, P = 0.767). This difference was not statistically significant (Fig. 3). Thyroid supplements Thyroid hormone replacement therapy was administered to patients with hypothyroidism. Six studies [21, 32, 33, 38, 39, 41] evaluated the relationship between thyroid hormone supplements and the risk of breast cancer; of these, five studies [21, 32, 38, 39, 41] found no statistical correlation between thyroid hormone supplements and the incidence of breast cancer, while one [33] reported that thyroid hormone therapy increased the incidence of breast cancer. Therefore, we conducted a metaanalysis to investigate the relationship between thyroid hormone therapy and the risk of breast cancer. The metaanalysis showed that there was no statistical correlation between thyroid hormone replacement therapy and the risk of breast cancer (OR = 0.83, 95% CI 0.57–1.21); but

13



J Endocrinol Invest

Fig. 2  Overall odd ratio for breast cancer with hypothyroidism in a meta-analysis of twelve researches and separate odd ratios for European researches and USA/Canadian researches using a random effect

Fig. 3  Primary hyperthyroidism and the risk of breast cancer in a meta-analysis of ten researches using random effects

I2 = 82.4% and P < 0.001, indicate great heterogeneity in this outcome (Fig. 4). We, therefore, conducted a sensitivity analysis to explore the heterogeneity. I2 decreased to 0 after excluding the study by Cristofanilli et al. [33]. The pooled estimate effect changed to an OR of 1.01 (95% CI 0.88–1.17, P = 0.330) (Fig. 5).

13

Discussion The negative results of the research are consistent with most of the reports in the literature and further prove that neither disorders of the thyroid nor treatment for the conditions greatly changes the risk of breast cancer.

J Endocrinol Invest Fig. 4  Thyroid hormone replacement therapy and breast cancer risk in a meta-analysis of six researches

Fig. 5  Overall OR for breast cancer with thyroid hormone replacement therapy in a metaanalysis of five researches

Due to the large heterogeneities, subgroup analyses were conducted based on the countries of the study population and the outcomes to assess the association between hypothyroidism and the incidence of breast cancer. The result indicated that there was no heterogeneity in the European group, while there was serious heterogeneity in the American and Canadian group. Multivariate analysis was performed in one [35] and four studies [21, 33, 38, 39] in the European and the American and Canadian groups, respectively. No further meta-regression analysis or demographic comparison (menopausal status, parity, estrogen replacement therapy, family history of breast cancer, and ethnicity) was conducted between two groups because relevant data were not available.

Some studies suggested that a family history of breast cancer increased the risk of breast cancer [42, 43], which might influence the results of our meta-analysis. However, no study reported data regarding a family history of breast cancer, so we were unable to perform subgroup analysis on family history. Some studies showed that parity, age at first delivery, pregnancies, abortion history, and breastfeeding affected the occurrence of breast cancer; non-pregnant, nondelivery, and increased age at first delivery (>35 years) increased the risk of breast cancer. As was pointed out in Qiong Dai study [44], increased frequency of full-term production was associated with a reduced risk of breast cancer. The increased risk in null gravid infertile women may

13



reflect a longer-lasting hormonal abnormality. However, in the meta-analysis, only four studies [33, 35, 38, 39] conducted multivariate analysis on childbearing history, while the other articles did not consider this risk factor. Similarly, estrogen was another important factor leading to breast cancer, which is required for the complete development of the mammary gland and to promote the transformation of the mammary gland cells [45, 46]. Thereby, estrogen may play significant roles in the incidence of breast cancer. However, only one study [33] included estrogen into the multivariate analysis of hypothyroidism; the results indicated that primary hypothyroidism was associated with a reduced risk for breast cancer. Therefore, menopausal status may play an important role in the risk of breast cancer. A similar result was found in Chappa et al.’s study [47], the group of female patients (n  = 862) was divided into two subgroups on the basis of hormonal status: patients with breast cancer that arose before and after menopause. The result of the study found a significant association between breast cancer diagnosed after menopause and thyroid disease (P < 0.003). However, our meta-analysis failed to perform subgroup analysis on menopausal status because relevant data were not reported (Table 1). In addition, Niehoff [48] reported that numerous studies found that breast cancer was associated with body mass index (BMI) and that women with higher BMI and women who gained weight had greater ORs than those with lower BMI or those who maintained weight, respectively. These findings support our hypothesis that individuals with more fat tissue may store and metabolize polycyclic aromatic hydrocarbons (PAHs) to a different extent than those with lesser amounts of body fat, which may result in chronic, longterm exposure to these lipophilic carcinogens. However, only two studies [35, 39] in the present analysis mentioned BMI, resulting in some bias (Table 2). In addition, we found that only a few researches considered stages of breast cancer [32, 33, 36, 41], one of which [33] reported that primary hypothyroidism was associated with a greater possibility to develop local disease relative to women with euthyroidism (95 vs. 85.9% clinical T1 or T2 disease, respectively, P  = 0.02), with no lymph node involvement (62.8 vs. 54.4%, P  = 0.15). However, these findings conflict with those from Ditsch et al. [32] who reported that patients with a history of hypothyroidism were more susceptible to breast cancer compared with those at Tis stage and with benign breast disease, but the difference was not statistically significant. Two studies [36, 41] assessed the relationship between circulating levels of thyroid hormones or the volume of thyroid and the stage at diagnosis, regardless of a history of hypothyroidism. Although the association of autoimmune thyroid dysfunction and thyroid antibody with the incidence of breast cancer is reported in a large number of papers, they failed to

13

J Endocrinol Invest

analyze this factor separately in this meta-analysis because of a lack of data on autoimmune thyroid disease (including the thyroiditis, Graves disease, etc.) and thyroid antibodyrelated data. Two studies [31, 37] included in the metaanalysis demonstrated that the presence of high degrees of TPOAbs [31, 37] and TgAb [31] were linked to alifted risk of breast cancer; however, another study [32] discovered no association between TPOAbs’ high levels with the risk for breast cancer. At the same time, no association was found by two other included studies [21, 31] between autoimmune thyroiditis [21, 31] or Graves [21, 31] with the risk of breast cancer. One study [34], however, revealed that the incidence of autoimmune thyroid diseases were higher in breast cancer patients than in control individuals. Besides, the study [34] indicated that the mean value for serum antiTPO antibodies was higher in breast cancer patients than in the control group, whereas the difference between the groups in mean values for serum anti-thyroglobulin antibodies was not statistically significant. In addition, our finding of a non-significant increased risk of breast cancer is suggestive of an adverse effect of hyperthyroidism. Moreover, the risk of breast cancer for patients with hyperthyroidism was 3% higher than that of the control group (OR = 1.03, 95% CI 0.83–1.30); however, this difference was not statistically significant, consistent with results of the nine case–control studies [21, 30–32, 34–36, 39, 40]. Thus, women with hyperthyroidism are likely to benefit from stricter health control and lifestyle modifications, which can help early diagnosis and reduce the risk of breast cancer. Our meta-analysis did not prove that thyroid hormone replacement therapy played a statistically significant role in modifying the risk of breast cancer. In our study of the relationship between thyroid hormone supplements and breast cancer, the high inter-study heterogeneity was due primarily to one study [33], based on sensitivity analysis. When this study was excluded, the overall heterogeneity was minimized to 0. As the research reported, the research limitations include the retrospective nature of the investigation and the definition of cases with primary hypothyroidism (HYPT) on the basis of the reported information of medical records, because the misclassification bias they may introduce may influence the positive results. In addition, recent studies also reported that thyroid hormone replacement therapy was also related to the duration of the oral administration of thyroid hormone, dates of the first and last administration, and dose [49, 50]. However, the studies included in our meta-analysis did not investigate the effects of patient age or time and dose of oral administration of thyroid hormone on the incidence of breast cancer; consequently, corresponding subgroup analysis could not be performed in the present meta-analysis. These observations provide new directions for future researchers for

NS, P = ND

Prevalence of 20/1176 (1.7%) vs history of hyperthyroid- 25/1176 (2.1%) ism NP NS, P = ND

8/1176 (0.7%) vs 10/1176 (0.9%)

1176

Kalache et al. [38]

RR = 0.90 (0.1–6.4) P = ND Prevalence of 2/1362 (0.14%) history of hyperthyroid- vs 2/1250 (0.16%) ism

RRc [for women with a history of primary HT (treated and untreated)] = 1.0 (0.8–1.2), P = ND RR = 0.23 (0.0–0.7), P = ND RR = 1.04 (0.8–1.3), P = ND

1/1362 (0.07%) vs 4/1250 (0.32%) 219/1362 (16%) vs 191/1250 (15.2%)

1250

1362

Brinton et al. [37]

Prevalence of 7/200 (3.5%) vs 4/200 history of hyperthyroid- (2%) ism NP

NS

200

200

Smyth et al. [34]

NP

NS

1176

200

Prevalence of history of HT

Prevalence of history of non-treated HT Prevalence of history of treated HT

Prevalence of history of HT Prevalence of history of HT

5/150 (3.3%) vs 2/200 (1%) 3/200 (1.5%) vs 2/200 (1%)

150

Shering et al. [35]

Prevalence of 125/4575 (2.7%) vs history of hyperthyroid- 141/4682 (3%) ism

NS

NS

ORb = 0.9 (0.8–1.02), P = NS

NS

Prevalence of 465/4575 history of HT (10.2%) vs 530/468 (treated and (11.3%) untreated) Prevalence of 174/4502 (3.8%) vs history of 184/4604 treated HT (3.9%)

4682

4575

Simon et al. [21]

ORa = 0.44 (0.32– 0.60), P < 0.001

OR = 0.43 (0.33–0.57), P < 0.001

NP

RRc = 0.99 (0.6–1.7) P = ND

NP

ORb = 0.9 (0.7–1.1) P = NS

ND

NS

80/1136 (7.0%) vs 162/1088 (14.9%)

1136

Cristofanilli et al. [31]

Prevalence of 2/65 (3%) vs 0/13, 0/27 history of hyperthyroid- and 0/38 ism

ND

NS

7/65 (10%) vs 0/13, 5/27 (18.5%) and 3/38 (7.9%)

Multivariate analysis: statistical significance ORs (95% CI), P

Univariate analysis: statistical significance ORs (95% CI), P

78 (13 in situ, Prevalence of history of 27 with HT benign breast disease, and 38 healthy) 1088 Prevalence of history of primary treated HT

Comparison of incidence of hyperthyroidism between compared groups

Results 2

Multivariate analysis: statistical significance ORs (95% CI), P

Outcome 2 Univariate analysis: statistical significance ORs (95% CI), P

Results Comparison of incidence of HT between compared groups

Outcome 1

Control group (non cancer)

65

Case group (cancer)

No of patients

Ditsch et al. [30]

References

Table 2  Characterisitics of the comparison groups and studied outcomes

J Endocrinol Invest

13

13

2569

354

150

190

112

Talamini et al. [33]

Miriam et al. [36]

Orhan et al. [32]

Grani et al. [29]

Freitas et al. [28]

125

190

100

747

2588

179

Control group (non cancer)

Prevalence of history of HT

Prevalence of history of HT

Prevalence of history of treated HT

Prevalence of history of treated HT Prevalence of history of HT

Outcome 1

NS, P = ND

2/179 (1.1%) vs 2/179 (1.1%) 18/2569 (0.7%) vs 25/2588 (0.9%)

NS

9/190 (4.7%) vs 15/190 (7.89%)

NP

NS, P = 0.152 NP

4/150 (2.6%) vs 0/100

Prevalence of 1/112 (0.89%) vs 3/125 history of hyperthyroid- (2.4%) ism

Prevalence of 17/354 (4.8%) vs 19/747 history of hyperthyroid- (2.5%) ism Prevalence of 6/150 (4%) vs 1/100 (1%) history of hyperthyroidism Prevalence of 4/190 (2.1%) vs 7/190 history of hyperthyroid- (3.68%) ism

NS, P = ND

ORe = 0.9 (0.5–1.5)

  Multivariate analysis adjusting for study setting, age, and race

NP

NP

NS

NS, P = 0.42

NP

ORe = 2.2 (1.1–4.4), P = ND

ORd = 1.0 (0.7–1.4), P = ND

Multivariate analysis: statistical significance ORs (95% CI), P

NS, P = 0.24

NS, P = ND

NS, P = ND

Univariate analysis: statistical significance ORs (95% CI), P

  Multivariate analysis adjusted for body mass index, study setting, age, education, menopausal status, parity, and in overall analyses

  Multivariate analysis adjusted for family history of breast cancer, screening variables (see text), Jewish religion, Latin American birthplace, age, age at first full-term birth, menopausal status, nulliparity, height

e

d

  Multivariate analysis adjusted for age of first use of thyroid supplements, years of use, years since initial use; and last use, age of menarche and the first child, parity, weight parity status, family history of breast cancer, or history of surgery for benign breast disease

c

b

  Multivariate analysis adjusting for hormone replacement treatment, family history of breast cancer, history of pregnancy, menopausal status, and age

a

Comparison of incidence of hyperthyroidism between compared groups

Results 2

Prevalence of 70/2569 (2.7%) vs history of hyperthyroid- 65/2588 (2.5%) ism

Outcome 2

ORd = 0.6 (0.3– 1.2), P = ND

Multivariate analysis: statistical significance ORs (95% CI), P

25/354 (0.7%) vs 50/747 (6.69%)

NS, P = ND

Univariate analysis: statistical significance ORs (95% CI), P

Comparison of incidence of HT between compared groups

Results

BC breast cancer, HT primary hypothyroidism, n number, ORs odd ratios, ND no data, NS non-significant, RR relative risk, NP not performed

179

Case group (cancer)

No of patients

Adami et al. [39]

References

Table 2  continued

J Endocrinol Invest

J Endocrinol Invest

an in-depth investigation of the relationship between oral administration of thyroid hormone and breast cancer. In the studies analyzed, iodine intake was not reported explicitly, and only one study [31] reported that the population was from the same iodine intake areas. Iodine is considered to be an important causative factor in similar studies. Reports on the connection of breast cancer with diminished dietary iodine intake have implied that such deficiencies may lead to hypothyroidism predisposing to breast disease. Excessive iodine will not only lead to the risk of hyperthyroidism, but also increase the incidence of hypothyroidism through the following ways. Excessive iodine will result in the increasing apoptosis of thyroid cells, which may be done through the following ways: increase the expression of apoptosis promoting gene Bcl-2 and reduce the expression of apoptosis inhibiting gene Bax, thereby increasing the occurrence of apoptosis. Besides, the changes of oxidative damage and Bcl-2 Bax will also activate the apoptosis process of Fas mediate, leading to the increase of apoptosis and finally affect the function of the thyroid [51–53]. Therefore, iodine deficiency can lead to the increase in incidence of hypothyroidism and iodine deficiency should be a risk factor for breast cancer, which further confirmed the conclusions of this study. However, excessive iodine is a confounding factor to cause the occurrence of hypothyroidism and hyperthyroidism. Because of

the lack of reports on iodine intake in the included studies, a deeper analysis cannot be conducted. The assessments of the quality of the evidence are shown in Table 3, which indicated that the overall quality of evidence was poor for hypothyroidism and thyroid hormone replacement therapy. This result of research also suggested that more researches in the future may influence the confidence to evaluate the effect and be likely to change the estimate. The major reasons for this poor assessment were that the studies were observational studies; consequently, the risk of confounding bias was unclear. Another factor which decreased the quality of the evidence were the wide confidence intervals for the outcomes of hypothyroidism, hyperthyroidism, and thyroid supplements in incidence of breast cancer. The strengths of the research included the large-scaled comprehensive literature and database search without language restrictions. Besides, the type of thyroid disorders and the requirements for an internal control group are defined through specific diagnostic criteria, which improved the quality of the quantitative analysis. This research observed strict literature inclusion and exclusion criteria, and all enrolled literature were assigned semiquantified scores using the NOS system after strict quality control and data extraction. The corresponding computational models were selected in data analysis through

Table 3  Summary of findings for the main comparison

High quality Further research is very unlikely to change our confidence in the estimate of effect Moderate quality Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate Low quality Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate Very low quality We are very uncertain about the estimate CI confidence interval, OR odds ratio, GRADE Working Group grades of evidence a

  The basis for the assumed risk (e.g., the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI) b

  The heterogeneity was high

c

 Imprecision

13



rigid heterogeneity testing. Finally, a sensitivity analysis was performed on the results of the meta-analysis. Our study also had several limitations: (1) some factors (e.g., alcohol consumption, BMI, smoking, health activity, and lifestyle) may also influence the development of breast cancer, but most of the studies in our meta-analysis did not report details on these factors, which might weaken our confidence in our findings; (2) the enrolled studies derived from case–control studies, which could introduce a bias; (3) the sample size was not large enough; therefore, the findings should be interpreted with caution.

Conclusions The results of our meta-analysis did not show an association between thyroid dysfunction and risk of breast cancer. However, we can assume that women with hyperthyroidism may benefit from stricter health controls and lifestyle modifications which can help early diagnosis and reduce the risk of breast cancer. The evidence was of poor quality, because the significant inter-study heterogeneity was introduced through retrospective study design and most of the researches failed to conduct multivariate analysis. Although there are some confounders in this study, which may weaken the results of the meta-analysis, we believe that the findings provide useful information for stakeholders with an interest in the outcomes of patients with thyroid dysfunction. Larger prospective studies are necessary, including women with thyroid dysfunction, to clarify the potential association between thyroid dysfunction and breast cancer. Acknowledgements  The authors thank the Limin Tian (Gansu Provincial Hospital), Kehu Yang, Yaolong Chen and Jinhui Tian (Evidence Based Medicine Center of Lanzhou University) for their help and support to the methodology and meta-analysis process. Author contribution  Conceived and designed the experiments: LT. Performed the experiments: YF. Analyzed the data: JS, LY, RY. Wrote the paper: YF. Offered suggestions: LT, KY, LY. Compliance with ethical standards  Conflict of interest  The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. Ethical approval This article does not contain any studies with human participants or animals performed by any of the authors. Informed consent  No informed consent.

13

J Endocrinol Invest

References 1. Torre LA, Bray F, Siegel RL, Ferlay J, Lortet-Tieulent J, Jemal A (2015) Global cancer statistics, 2012. CA Cancer J Clin 65(2):87–108 2. Beatson G (1896) On the treatment of inoperable cases of carcinoma of the mamma: suggestions for a new method of treatment, with illustrative cases. Lancet 148(3803):162–165 3. Ditsch N, Toth B, Himsl I, Lenhard M, Ochsenkuhn R, Friese K et al (2013) Thyroid hormone receptor (TR)alpha and TRbeta expression in breast cancer. Histol Histopathol 28(2):227–237 4. Hall LC, Salazar EP, Kane SR, Liu N (2008) Effects of thyroid hormones on human breast cancer cell proliferation. J Steroid Biochem Mol Biol 109(1–2):57–66 5. Tang HY, Lin HY, Zhang S, Davis FB, Davis PJ (2004) Thyroid hormone causes mitogen-activated protein kinase-dependent phosphorylation of the nuclear estrogen receptor. Endocrinology 145(7):3265–3272 6. Glinskii AB, Glinsky GV, Lin HY, Tang HY, Sun M, Davis FB et al (2009) Modification of survival pathway gene expression in human breast cancer cells by tetraiodothyroacetic acid (tetrac). Cell cycle (Georgetown, Tex) 8(21):3562–3570 7. Hercbergs AH, Ashur-Fabian O, Garfield D (2010) Thyroid hormones and cancer: clinical studies of hypothyroidism in oncology. Curr Opin Endocrinol Diabetes Obes 17(5):432–436 8. Smyth PP (2003) The thyroid, iodine and breast cancer. Breast Cancer Res BCR 5(5):235–238 9. Funahashi H, Imai T, Tanaka Y, Tsukamura K, Hayakawa Y, Kikumori T et al (1999) Wakame seaweed suppresses the proliferation of 7,12-dimethylbenz(a)-anthracene-induced mammary tumors in rats. Jpn J Cancer Res Gann 90(9):922–927 10. Vonderhaar BK, Greco AE (1979) Lobulo-alveolar development of mouse mammary glands is regulated by thyroid hormones. Endocrinology 104(2):409–418 11. Nagasawa H, Yanai R, Nakajima Y, Namiki H, Kikuyama S, Shiota K (1980) Inhibitory effects of potassium thiocyanate on normal and neoplastic mammary development in female mice. Eur J Cancer 16(4):473–480 12. Smyth PP (1997) The thyroid and breast cancer: a significant association? Ann Med 29(3):189–191 13. Kuijpens JL, Nyklictek I, Louwman MW, Weetman TA, Pop VJ, Coebergh JW (2005) Hypothyroidism might be related to breast cancer in post-menopausal women. Thyroid Off J Am Thyroid Assoc 15(11):1253–1259 14. Finley JW, Bogardus GM (1960) Breast cancer and thyroid disease. Q Rev Surg Obstet Gynecol 17:139–147 15. Szychta P, Szychta W, Gesing A, Lewinski A, Karbownik-Lewinska M (2013) TSH receptor antibodies have predictive value for breast cancer—retrospective analysis. Thyroid research 6(1):8 16. Hellevik AI, Asvold BO, Bjoro T, Romundstad PR, Nilsen TI, Vatten LJ (2009) Thyroid function and cancer risk: a prospective population study. Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cospons Am Soc Prev Oncol 18(2):570–574 17. Tosovic A, Bondeson AG, Bondeson L, Ericsson UB, Malm J, Manjer J (2010) Prospectively measured triiodothyronine levels are positively associated with breast cancer risk in postmenopausal women. Breast Cancer Res BCR 12(3):R33 18. Sogaard M, Farkas DK, Ehrenstein V, Jorgensen JO, Dekkers OM, Sorensen HT (2016) Hypothyroidism and hyperthyroidism and breast cancer risk: a nationwide cohort study. Eur J Endocrinol 174(4):409–414 19. Loeser AA (1954) A new therapy for prevention of post-operative recurrences in genital and breast cancer; a six-years study of prophylactic thyroid treatment. BMJ 2(4901):1380–1383

J Endocrinol Invest 20. Weiss HA, Brinton LA, Potischman NA, Brogan D, Coates RJ, Gammon MD et al (1999) Breast cancer risk in young women and history of selected medical conditions. Int J Epidemiol 28(5):816–823 21. Simon MS, Tang MT, Bernstein L, Norman SA, Weiss L, Burkman RT et al (2002) Do thyroid disorders increase the risk of breast cancer? Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cospons Am Soc Prev Oncol 11(12):1574–1578 22. Angelousi AG, Anagnostou VK, Stamatakos MK, Georgiopoulos GA, Kontzoglou KC (2012) Mechanisms in endocrinology: primary HT and risk for breast cancer: a systematic review and meta-analysis. Eur J Endocrinol 166(3):373–381 23. Kapdi CC, Wolfe JN (1976) Breast cancer. Relation ship to thyroid supplements for hypothyroidism. JAMA 236(10):1124–1127 24. Wallace RB, Sherman BM, Bean JA, Leeper J (1978) Thyroid hormone use in patients with breast cancer. Absence of an association. Jama 239(10):958 25. Shapiro S, Slone D, Kaufman DW, Rosenberg L, Miettinen OS, Stolley PD et al (1980) Use of thyroid supplements in relation to the risk of breast cancer. JAMA 244(15):1685–1687 26. Danielson DA, Jick H, Hunter JR, Stergachis A, Madsen S (1982) Nonestrogenic drugs and breast cancer. Am J Epidemiol 116(2):329–332 27. Sarlis NJ, Gourgiotis L, Pucino F, Tolis GJ (2002) Lack of association between Hashimoto thyroiditis and breast cancer: a quantitative research synthesis. Hormones (Athens, Greece) 1(1):35–41 28. Hardefeldt PJ, Eslick GD, Edirimanne S (2012) Benign thyroid disease is associated with breast cancer: a meta-analysis. Breast Cancer Res Treat 133(3):1169–1177 29. Stang A (2010) Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 25(9):603–605 30. Freitas PA, Vissoci GM, Pinto RM, Lajolo PP, Jorge PT (2016) Study of the prevalence of autoimmune thyroid disease in women with breast cancer. Endocr Pract Off J Am Coll Endocrinol Am Assoc Clin Endocrinol 22(1):16–21 31. Grani G, Dicorato P, Dainelli M, Coletta I, Calvanese A, Del Sordo M et al (2012) Thyroid diseases in women with breast cancer. La Clin Ter 163(6):e401–e404 32. Ditsch N, Liebhardt S, Von Koch F, Lenhard M, Vogeser M, Spitzweg C et al (2010) Thyroid function in breast cancer patients. Anticancer Res 30(5):1713–1717 33. Cristofanilli M, Yamamura Y, Kau SW, Bevers T, Strom S, Patangan M et al (2005) Thyroid hormone and breast carcinoma. Primary hypothyroidism is associated with a reduced incidence of primary breast carcinoma. Cancer 103(6):1122–1128 34. Turken O, NarIn Y, DemIrbas S, Onde ME, Sayan O, KandemIr EG et al (2003) Breast cancer in association with thyroid disorders. Breast Cancer Res BCR 5(5):R110–R113 35. Talamini R, Franceschi S, Favero A, Negri E, Parazzini F, La Vecchia C (1997) Selected medical conditions and risk of breast cancer. Br J Cancer 75(11):1699–1703 36. Smyth PP, Smith DF, McDermott EW, Murray MJ, Geraghty JG, O’Higgins NJ (1996) A direct relationship between thyroid enlargement and breast cancer. J Clin Endocrinol Metab 81(3):937–941 37. Shering SG, Zbar AP, Moriarty M, McDermott EW, O’Higgins NJ, Smyth PP (1996) Thyroid disorders and breast cancer. Eur J Cancer Prev Off J Eur Cancer Prev Organ (ECP) 5(6):504–506

38. Moseson M, Koenig KL, Shore RE, Pasternack BS (1993) The influence of medical conditions associated with hormones on the risk of breast cancer. Int J Epidemiol 22(6):1000–1009 39. Brinton LA, Hoffman DA, Hoover R, Fraumeni JF Jr (1984) Relationship of thyroid disease and use of thyroid supplements to breast cancer risk. J Chronic Dis 37(12):877–893 40. Kalache A, Vessey MP, McPherson K (1982) Thyroid disease and breast cancer: findings in a large case–control study. Br J Surg 69(7):434–435 41. Adami HO, Rimsten A, Thoren L, Vegelius J, Wide L (1978) Thyroid disease and function in breast cancer patients and nonhospitalized controls evaluated by determination of TSH, T3, rT3 and T4 levels in serum. Acta Chir Scand 144(2):89–97 42. Streff H, Profato J, Ye Y, Nebgen D, Peterson SK, Singletary C et al (2016) Cancer incidence in first- and second-degree relatives of BRCA1 and BRCA2 mutation carriers. Oncologist 21(7):869–874 43. Collaborative Group on Hormonal Factors in Breast Cancer (2001) Familial breast cancer: collaborative reanalysis of individual data from 52 epidemiological studies including 58,209 women with breast cancer and 101,986 women without the disease. Lancet (London, England) 358(9291):1389–1399 44. Dai Q, Liu B, Du Y (2009) Meta-analysis of the risk factors of breast cancer concerning reproductive factors and oral contraceptive use. Front Med China 3(4):452–458 45. Lai LC (2002) Role of steroid hormones and growth factors in breast cancer. Clin Chem Lab Med 40(10):969–974 46. Saraiva PP, Figueiredo NB, Padovani CR, Brentani MM, Nogueira CR (2005) Profile of thyroid hormones in breast cancer patients. Braz J Med Biol Res Revista brasileira de pesquisas medicas e biologicas 38(5):761–765 47. Chiappa C, Rovera F, Rausei S, Del Ferraro S, Fachinetti A, Lavazza M et al (2016) Breast cancer and thyroid diseases: analysis of 867 consecutive cases. J Endocrinol Invest 40(2):179–184 48. Niehoff N, White AJ, McCullough LE, Steck SE, Beyea J, Mordukhovich I et al (2017) Polycyclic aromatic hydrocarbons and postmenopausal breast cancer: an evaluation of effect measure modification by body mass index and weight change. Environ Res 152:17–25 49. Mustacchi P, Greenspan F (1977) Thyroid supplementation for hypothyroidism. An latrogenic cause of breast cancer? JAMA 237(14):1446–1447 50. Morabia A, Szklo M, Stewart W, Schuman L, Thomas DB, Zacur HA (1992) Thyroid hormones and duration of ovulatory activity in the etiology of breast cancer. Cancer Epidemiol Biomark Prev Publ Am Assoc Cancer Res Cospons Am Soc Prev Oncol 1(5):389–393 51. Hsu YT, Youle RJ (1998) Bax in murine thymus is a soluble monomeric protein that displays differential detergent-induced conformations. J Biol Chem 273(17):10777–10783 52. Oltvai ZN, Milliman CL, Korsmeyer SJ (1993) Bcl-2 heterodimerizes in vivo with a conserved homolog, Bax, that accelerates programmed cell death. Cell 74(4):609–619 53. Bonny C, Oberson A, Steinmann M, Schorderet DF, Nicod P, Waeber G (2000) IB1 reduces cytokine-induced apoptosis of insulin-secreting cells. J Biol Chem 275(22):16466–16472

13