Association between metabolic syndrome and endometrial cancer risk: a systematic review and meta-analysis of observational studies

Introduction

Endometrial cancer (EC) is the most common gynecologic cancer in developed countries, and its incidence and associated mortality are on the rise [1]. In 2018, more than 380,000 new cases and approximately 90,000 deaths worldwide were from EC, making it responsible for 4.4% of cancer cases and 2.1% of deaths due to cancer in women [2]. The prevalence of EC varies in different regions [3], which may be attributed to disparities in the incidence of obesity, as well as other important risk factors, such as aging, early age at menarche, late-onset menopause, nulliparity, and hormone replacement therapy use [4].

As the worldwide burden of EC continues to increase, interest is growing in the development of early preventive strategies for women at increased risk [5]. Metabolic syndrome (MetS) is an aggregation of several metabolic abnormalities, which include obesity, insulin resistance, hypertension, and dyslipidemia [6]. Irrespective of the diagnostic criteria used, the world is currently facing a significant epidemic of MetS. More than 20% of adults in most Asia–Pacific countries are affected by MetS [7], and approximately one-third of the adult population in the United States has MetS [8]. Studies suggest that metabolic abnormalities may be important risk factors for the development of EC. According to the results of detailed epidemiological studies, obesity is one of the most important risk factors for EC [9], whereas other studies have reported that diabetes is also a risk factor for EC, independent of obesity [10–12]. In addition, other metabolic abnormalities, such as hypertension [13] and dyslipidemia [14], are also associated with increased EC risk.

Although there is a general understanding of the association between EC risk and metabolic abnormalities, the association with MetS has not been established. Two meta-analyses [15, 16] have linked MetS, diagnosed according to the criteria of the National Cholesterol Education Program—Third Adult Treatment Panel (NCEP-ATP III), to an increased risk of EC. However, these findings should be interpreted with caution due to several limitations. Firstly, the risk estimates provided by the studies included in the meta-analyses were not suitable in evaluating the association between MetS and EC risk. For example, a meta-analysis conducted by Esposito et al. included a study reporting the association between metabolic abnormalities (defined as at least one of the following: diabetes, hypertension, overweight/obesity, dyslipidemia), rather than MetS and EC risk. Secondly, moderate or substantial heterogeneity was observed in these two meta-analyses, but the potential sources of heterogeneity were not fully explored. Thirdly, there was no evaluation of the association between MetS according to its different definitions and EC risk. Fourthly, two studies with large sample sizes (n = 13061 [17]; n = 117074 [18]), examining the same research area, have been recently published after these two meta-analyses, suggesting that this association is worth investigating. Therefore, we performed a systematic review and meta-analysis of observational studies to identify associations between MetS, diagnosed according to different criteria, and EC risk.

Results

Discussion

Irrespective of the diagnosis of MetS (NCEP-ATP III or IDF criteria), we found that MetS was positively associated with the risk of EC in our meta-analysis of observational studies. These findings were largely consistent across subgroups defined by various study characteristics and stable in sensitivity analyses. To our best knowledge, no meta-analysis has been performed to evaluate the association between MetS, diagnosed according to IDF criteria, and EC risk.

The mechanistic understanding by which MetS promotes EC is unclear, although it may be attributed to abnormal fat metabolism, chronic inflammation, hyperglycemia, and hyperinsulinemia [23]. A previous study revealed that adiponectin, which is secreted mainly by white adipose tissue, can activate the liver kinase B1–AMP–activated protein kinase (LKB1–AMPK) signaling pathway to inhibit the proliferation and invasion of EC cells by binding to adiponectin receptors [24]. Elevated leptin levels, which are encoded by the obesity gene, have been shown to promote the proliferation and invasion of EC cells by activating a variety of signaling pathways [23]. In addition, studies have demonstrated that adipose-derived inflammatory cytokines, such as IL-6 [25] and TNF-α [23], are involved in the development and progression of EC. Elevated serum glucose levels can enhance glycolysis through the AMPK signaling pathway, thereby resulting increased invasiveness of EC cells [23]. Glycolysis cannot only generate much-needed energy for tumor cells, but also produce numerous metabolic intermediates, which can be used by tumor cells in the synthesis of biological macromolecules [23]. In a previous study [26], metformin was found to suppress EC cell proliferation, which may be attributed to the activation of the AMPK signaling pathway. However, the use of metformin in the treatment of EC is still controversial, and further research is warranted to identify critical molecules and new drugs. Insulin can activate the phosphoinositide 3–kinase/protein kinase B or mitogen–activated protein kinase/extracellular signal–regulated kinase signaling pathway by binding to insulin receptor/insulin-like growth factor-1 receptor, thereby promoting epithelial-mesenchymal transition and increasing the proliferation and invasion of EC cells [27]. Furthermore, a recent study has reported that cholesterol can activate the transcriptional activity of EC cells and promote the proliferation of EC cells [28].

In addition to the elevated burden of MetS, metabolic abnormalities, such as obesity and diabetes, also contribute to an increased risk of developing EC. Improvements in the overall metabolic state, especially weight maintenance and blood glucose control, have enormous clinical implications for both MetS and EC. Previous studies have linked EC risk to metabolic abnormalities such as obesity, diabetes, and hypertension. These disorders are collectively known as the metabolic triad of EC [23]. There is compelling evidence indicating that being overweight or obese increases the risk of EC. According to a large number of epidemiological studies performed worldwide, obesity has been identified as the convincing cause of EC by the World Cancer Research Fund International Continuous Update Project [9], and it is responsible for nearly 40% of all EC cases in the Western world [29]. Three dose-response meta-analyses [30–32] showed that an increase in the body mass index (BMI) by 5 kg/m² can increase the risk of a women developing EC by 54%, 59%, and 60%, respectively. The evidence for abdominal fatness and weight gain was weaker than that using the BMI as the measure of body fatness, although it still supported a positive association between overall body fatness and EC risk [33]. Another metabolic abnormality, diabetes, has also been found to be associated with EC risk, independent of obesity [11, 12]. Two meta-analyses [34, 35] quantitatively summarized the results of published observational studies reporting a greater risk of EC in subjects with diabetes compared with those without diabetes. In addition, the results of subgroup analysis according to the type of diabetes supported a positive association between type 1 and type 2 diabetes and EC [34]. The contribution of hypertension to EC has also been established. A previous meta-analysis [36] involving 300,598 participants and 28,385 EC cases has reported an increased risk of EC among patients with hypertension. There was no significant change in the magnitude and direction of the association between hypertension and EC risk among studies after adjusting for the BMI. However, few studies have examined the association between serum lipid profiles and EC risk. Two prospective studies [14, 19] reported a positive association between low serum high-density lipoprotein-cholesterol (HDL-C) levels and high serum triglyceride levels and EC risk, although the association between serum triglyceride levels and EC risk was not statistically significant when the BMI was included in the model [14].

Various heterogeneous factors can contribute to the risk of EC. For example, different regions of the world have distinct differences in culture, economy, and education that may affect the occurrence of EC. Although subgroup analyses by region indicated that MetS significantly increased the risk of EC in women in North America and Europe, it was difficult to determine which factors contributed to the observed positive association due to the unavailability of information. Therefore, several area-related factors, namely the economy and education, need to be considered in further research. Consistent with two previous meta-analyses [15, 16], there was considerable heterogeneity in our summary analyses. Differences in various characteristics among the included studies may have influenced the heterogeneity, including the type of design, determination of metabolic abnormalities, risk of bias, and adjustment factors. As shown in subgroup analyses, irrespective of the diagnosis of MetS (NCEP-ATP III or IDF criteria), we found that the heterogeneity was reduced among studies using the results of laboratory assays and anthropometric measurements to determine the metabolic abnormalities or proxy indicators, suggesting that the methods used to identify the metabolic abnormalities may be responsible for the significant heterogeneity in our analysis. The metabolic abnormalities were not directly identified by the results of laboratory assays and anthropometric measurements in the three included studies [18, 20, 21], but replaced by the proxy indicators. In a study [21] using linked pharmaceutical and cancer registry data, Russo et al. reported no association between MetS defined according to combined prescription patterns and EC risk. Trabert et al. identified the metabolic abnormalities based on outpatient/inpatient data rather than direct measurements of blood pressure; the levels of fasting serum triglycerides, HDL-C, fasting plasma glucose, and waist circumference, and found a positive association between MetS and EC risk [18]. We found that the heterogeneity was significantly reduced after excluding the study by Trabert et al., irrespective of whether it was based on NCEP ATP III (I² = 0.7%) or IDF (I² = 42%) criteria. Another study [20] showed that the EC risk was significantly increased for subjects with MetS diagnosed according to self-reported histories of diabetes, drug-treated hypertension, drug-treated hyperlipidemia, and obesity. In addition, according to NCEP ATP III criteria, the results of the present meta-analysis indicate that a proportion of the observed heterogeneity may be explained by differences in study design.

The present meta-analysis has several strengths. Firstly, our findings are in agreement with two previous meta-analyses that reported a positive association between MetS and EC risk. However, with a careful screening process and a large sample size, we included six observational studies investigating the association between MetS and EC risk. Therefore, our strong statistical power could detect positive associations. In addition, our results were consolidated by further subgroup and sensitivity analyses. Secondly, the positive association of MetS with EC risk was observed not only in studies using NCEP-ATP III criteria to diagnose MetS but also in those using IDF criteria. To our best knowledge, this is the first meta-analysis to quantitatively evaluate the association between MetS, diagnosed according to different criteria, and EC risk. Thirdly, there was significant heterogeneity in our study. Thus, detailed subgroup and univariate meta-regression analyses were used to assess whether results varied according to key study characteristics.

Our findings should be interpreted with caution, as there are several limitations. Firstly, clinical and methodological heterogeneity is always a concern for all meta-analyses, particularly for meta-analyses of observational studies [37]. Irrespective of the diagnostic criteria used for MetS, there was considerable heterogeneity in our meta-analysis. Although meta-regression analyses were used, the limited number of included studies may have restricted the power of meta-regression in exploring the sources of heterogeneity. Secondly, cohort, case-control, and nested case-control studies were all represented in our meta-analysis, and the methodological differences in study designs may have biased the results because of significant variations in analyses. Considering that half of the included studies were case-control studies, our results may have been subjected to recall bias. Thirdly, although the included studies attempted to control for known risk factors and we also extracted maximally adjusted risk estimates, the possibility of residual confounding could not be excluded, given that our findings originate from observational studies where residual confounding always exists [37]. In addition, single-point measurements increase the chance of random measurement errors, which may lead to attenuation of the reported associations [16]. Finally, we did not assess publication bias using the funnel plot or Egger regression test, as they have insufficient power in cases of limited studies (n < 10). Thus, we cannot eliminate the possibility that our summary results are driven by publication bias. Nevertheless, all meta-analyses are subject to publication bias due to the possibility of under-reporting negative results or failing to identify the ‘grey literature’ (i.e., the articles that are not published formally by publishers, including conference proceedings, magazine articles, and government papers) [38].

In conclusion, the results of this meta-analysis suggest MetS associates with an increased risk of EC. Considering the limitations of this meta-analysis, further studies with perspective designs are warranted to confirm our findings, with an emphasis on determining the metabolic abnormalities and adjusting for several key confounding factors (e.g., overweight/obesity or BMI/WC).

Materials and Methods

Supplementary Materials

Supplementary Figures

Supplementary Tables