Search results and study characteristics
The procedure of literature selection is based on the PRISMA statement[28] described in Figure 1. In total, 973 records were identified through different databases, of which 7 records were identified after reviewing the reference lists of the retrieved articles. After duplicate studies (n=289) were ruled out, 691 remaining studies have glanced over. 656 studies were excluded after screening the title and abstract. After reading the full-text of the remaining 35 articles carefully, 17 studies were excluded (9 articles were excluded after full-text reviewed, 4 articles were excluded without comparators included, 2 articles without suitable outcomes, and 2 articles were ruled out for unable to get full-text). Thus, 18 reports[5-22] published between 1982 to 2020 were included in this meta-analysis.
A total of 422,384 patients were included in our study. Details about the basic characteristics of the included studies in this systematic review are summarized in Table 1. Of the included studies, 6 were from the USA, 2 were conducted in Greece, and 1 each was from Germany, China, Sweden, Brazil, Czech, Turkey, Italy, Poland, Denmark, and the UK, respectively. The countries were further divided into 2 groups to investigated the region characteristic of BTD on the risk of BC. Greece, Germany, Sweden, Czech, Italy, Poland, Denmark, and the UK belong to the Europe group, while the USA, China, Brazil, and Turkey belong to the Non-Europe group. The studies varied in sample size from 9 to 139,124. According to the NOS assessment system[25], all of the included articles were considered to be high-quality (Additional file 1: Supplement figure 2).
BTD increases the risk of BC
As shown in Figure 2a, a total of 15 studies were included in the present meta-analysis. Our pooled analysis of these reports showed that a higher risk of developing BC for people with BTD (OR:1.27, 95%CI:1.09–1.48, I2=80.5%, n=15). After stratifying by different kinds of BTD, we found that autoimmune thyroiditis (AITD)[5, 8-11] (OR:2.56, 95%CI:1.95–3.37, I2=0.0%, n=5), and goiter[5, 9, 11, 12, 15] (OR:2.13, 95%CI:1.19-3.79, I2=80.6%, n=5) were positively associated with the risk of BC. However, both hypothyroidism[5, 6, 16, 19-22] (OR:0.82, 95%CI:0.64-1.04, I2=85.0%, n=7) and hyperthyroidism[5-7, 9, 11, 15, 18-22] (OR:1.07, 95%CI:0.93-1.24, I2=24.9%, n=11) had no correlations with BC risk. Considering the varied causes of hyperthyroidism, GD is the most common cause[29, 30]. Hyperthyroid patients with positive TRAb were therefore singled out as GD group to further investigated the underlying relationship between them. Our result showed a positive relationship between GD[7, 11] (OR:4.17, 95%CI:1.08-16.05, I2=0.0%, n=2) and BC risk (Figure 2b). Thus, GD increased the risk of BC was newly proposed in the present study.
To further explore the heterogeneity between the included studies, subgroup analysis was applied by different regions (Europe and Non-Europe) (Figure 2c). BTD was positively associated with the risk of BC both in Europe (OR:1.31, 95%CI:1.03-1.65, I2=73.3%) and Non-Europe (OR:1.27, 95%CI:1.01-1.61, I2=84.4%) subgroup. However, after stratification by different regions, the heterogeneity between the included studies did not show any decrease. It indicated the region characteristic did not exist for the including studies.
BTD and aggressiveness of BC
In addition to discuss the relationship between BTD and the risk of BC, we further investigated whether the existence of thyroid dysfunction exerted an influence on the aggressiveness of BC[31]. A total of 4 pieces of research[13, 14, 16, 17] were included eventually. Subgroup analysis was conducted on different aggressiveness markers of BC, grade Ⅱ and lymph gland metastases subgroups were excluded for only one article available. Our pooled result demonstrated no relationship between BTD and grade Ⅲ subgroup (HR:0.77, 95%CI:0.13-4.58, I2=85.9%, n=2), tumor>20mm subgroup (HR:0.87, 95%CI:0.18-4.13, I2=89.7%, n=2), estrogen receptor-negative subgroup (HR:1.03, 95%CI:0.80-1.32, I2=77.2%, n=4) and progesterone receptor-negative subgroup (HR:1.19, 95%CI:0.83-1.71, I2=82.8%, n=3). The result is shown in Figure 3a. 2 studies[13, 17] cohered with our pooled result that there is no significant relation between BTD and the aggressiveness of BC. To future investigated whether there was a difference among different kinds of BTD and aggressiveness of BC, a subgroup analysis was conducted by different kinds of BTD. After retrieving related literature comprehensively, the existing articles mainly focused on hyperthyroidism and hypothyroidism. The present synthesis analysis didn’t find a relationship in hyperthyroidism (HR:1.28, 95%CI:0.92-1.80, I2=83.9%, n=4) and hypothyroidism (HR:0.99, 95%CI:0.88-1.10, I2=38.6%, n=2) subgroup (Figure 3b).
Considering the high heterogeneity of the included studies, we then further detected whether a difference has existed between different regions (Europe and Non-Europe). After stratification by region, the heterogeneity between the included studies did not show any decrease. However, a positive relationship was observed in Europe subgroup (HR:2.05, 95%CI:1.32-3.17, I2=86.4%, n=2) (Figure 3c).
Sensitivity analysis and publication bias
The publication bias detection of the literature included was analyzed using the Harbord test[32] (Figure 4). Figure 4a shows no publication bias for the relationship between autoimmune thyroiditis and BC risk (p=0.857). Similarly, no publication bias was observed in hypothyroidism (p=0.287) and hyperthyroidism (p=0.754) subgroups. However, publication bias existed in the goiter subgroup with a p-value of 0.019. Sensitivity analysis[33] was used to verify the reliability of the result of meta-analysis in the goiter subgroup. After removing the study reported by Bach et al[5], publication bias is inexistent with a p-value of 0.949. The other sensitivity analysis results were consistent with the primary analysis.