The effect of diabetes on the risk of endometrial Cancer: an updated a systematic review and meta-analysis

Background Previous studies conducted on the association between diabetes and the risk of endometrial cancer have reported controversial results that have raised a variety of questions about the association between diabetes and the incidence of this cancer. Thus, the aim of this systematic review and meta-analysis was to more precisely estimate the effect of diabetes on the risk of endometrial cancer incidence. Methods All original articles were searched in international databases, including Medline (PubMed), Web of sciences, Scopus, EMBASE, and CINHAL. Search was done from January 1990 to January 2018 without language limitations. Also, logarithm and standard error logarithm relative risk (RR) were used for meta-analysis. Results A total of 22 cohort and case-control studies were included in this meta-analysis, of which 14 showed statistically significant associations between diabetes and risk of endometrial cancer. Diabetes was associated with increased risk of endometrial cancer (RR = 1.72, 95% CI 1.48–2.01). The summary of RR for all 9 cohort studies was 1.56 (95% CI 1.21–2.01), and it was 1.85 (95% CI 1.53–2.23) for 13 case control studies. The summary of RR in hospital-based studies was higher than other studies. Thirteen of the primary studies-controlled BMI as a confounding variable, and the combined risk of their results was 1.62 (95% CI 1.34–1.97). Conclusions Diabetes seems to increases the risk of endometrial cancer in women, and this finding can be useful in developing endometrial cancer prevention plans for women having diabetes.


Background
A recent study conducted by Lortet-Tieulent, J and colleague show that endometrial cancer is the sixth most commonly occurring cancer in women and the 15th most commonly occurring cancer overall. There were over 380,000 new cases in 2018 [1]. Also, about 142,000 women are diagnosed with endometrial cancer annually worldwide, and about 42,000 women lose their life due to endometrial cancer. The usual curve of endometrial cancer indicates that most cases are diagnosed after menopause, and the highest incidence rate is around the seventh decade of life [2]. The disease is more than 10 times common in North America and Europe than in less developed countries [3]. The incidence and the mortality rate of endometrial cancer increased during 2006 and 2010 [4]. Estrogens, both internal and external, play an important role in increasing endometrial cancer [5]. Several studies have shown that the risk of endometrial cancer increases with older age, early menstruation, late menopause, obesity, family history of endometrial cancer (especially among close relatives), exposure to radiation, infertility (especially due to polycystic ovarian syndrome), and long-term use of estrogens for hormone therapy [4][5][6][7]. Estrogens, both internal and external, play an important role in increasing endometrial cancer. Multiple studies have claimed a positive association between diabetes and incidence of endometrial cancer with several biological mechanisms [8]. However, a previous systematic review and meta-analysis was performing by Friberg and colleges [8] but growing several publications afterwards and also considering new variables in adjusted models, we felt to design an updated systematic review and meta-analysis in order to show any posible relationship between diabetes and endometrial cancer.

Methods
This systematic review was performed according to the Meta-Analyses of Observational Studies in Epidemiology (MOOSE) and Strengthening the Reporting of Observationally Studies in Epidemiology (STROBE) guidelines for reviews of analytical observational studies (case-control and cohort) [2,9,10].

Search strategy
All original published articles were searched in international databases, including Medline (PubMed), Web of sciences, Scopus, EMBASE, and CINHAL. Search was done from January 1990 to January 2018 without language limitations. The keywords were Diabetes, Diabetes Mellitus (type 1 and 2), Insulin Dependent, IDDM, NIDDM, Noninsulin Dependent, Endometrial Stromal Tumors, Endometrial Neoplasms, and Endometrial. The selected studies were limited to observational studies on humans.
The primary search results were reviewed, and some of the articles were eliminated after reviewing their title and an abstract. Inclusion and exclusion criteria were set by 2 researchers separately (YM, FV) ( Fig. 1).

Eligibility criteria
A published study had to meet the following inclusion criteria: (1) original article, (2) case-control or cohort study, (3) human population, (4) diabetes and patients with diabetes as the main independent variable, and (5) endometrial cancer as the dependent variable. Case reports, reviews, animal studies, and case control or cohort studies with crude estimates about the effect of diabetes on the risk of endometrial cancer were removed from the tabulation. The authors resolved all disputes during the collection, compilation, and analysis of data.

Data extraction
Two researchers evaluated all included articles independently. They assessed the disagreement, if any, and in case an agreement was not reached, a third author (LS) evaluated the study. Two independent matched reviewers extracted the data according to a uniform Excel sheet. Then, a structured checklist was used to extract the following information: (1) author, (2) year of publication, (3) type of study, (4) country, (5) study population, (6) age of women, (7) sample size, (8) type of diabetes, (9) measurement, and (10) adjusted variables of association.

Statistical analysis
In the meta-analysis, 3 measures of association were used: (1) odds ratio (case-control and population-based case-control studies), (2) relative risk (cohort and population cohort studies), and (3) hazard ratio (cohort and population-based cohort studies). As the frequency of endometrial cancer was relatively low, the odds ratio in the case-control and population based case-control studies and the risk ratio in the cohort and populationbased cohort studies yielded similar estimates of relative risk (RR) [11].
Logarithm and standard error logarithm relative risk (RR) were used for the meta-analysis. DerSimonian and Laird method was used to compute the pooled estimate of relative risk (RR) with confidence interval (CI 95%) using random models [12]. Because the test for heterogeneity was statistically significant in some analyses, the random effects models were used to estimate RR. In this study, w Cochran's Q test and I2 statistic were used to evaluate statistical heterogeneity between studies [13]. In addition, a meta-regression and subgroup analysis was performed to assess the source of heterogeneity between studies. Moreover, publication bias was assessed by funnel plot and Egger and Begg's test [14,15]. Statistical analysis was performed using STATA 14.0 (Stata Corp, College Station, TX, USA), and statistical significance was set at p < 0.05.

Subgroup analysis
The subgroup analysis was conducted based on the study design, and variables adjustment (Table 3). Individual study results and the overall summary results for 8 cohorts and 7 population-based, 2 hospital-based, and 5 case-control studies investigating the effect of diabetes on the risk of endometrial cancer in women are shown in Table 3. The results indicated that the summary of RR for all the 8 cohort studies combined was 1.52 (95% CI 1.16-2.00), and heterogeneity among these studies was significant (Q = 3.03, I 2 = 70.7%; P = 0.001). The summary of RR for all the 7 population-based case-control studies was 1.55 (95% CI 1.37-1.75), however, heterogeneity among these studies was not significant (Q = 6.88, I 2 = 0.0%; P = 0.461). In addition, the summary of RR for all the 5 case-control studies was 2.31 (95% CI 1.81-2.96), but heterogeneity was not significant (Q = 6.69, I 2 = 22.7%; P = 0.270). Also, the summary of RR was higher in hospitalbased studies than in other studies [RR = 4.10 (CI 95% 2.09-8.01), heterogeneity was Q = 4.12, I 2 = 0.0%, P = 0.402]. According to the results in Table 3, the summary of RR in hospital-based studies was higher than in other studies. Also, 13 of the primary studies-controlled BMI as a confounding variable, and the combined risk of their results was 1.62 (95% CI 1.34-1.97, test for heterogeneity: Q = 4.14, I2 = 71.0%, P = 0.0001). However, 4 of the primary studies-controlled weight as a confounding variable, and the combined risk of their results was 2.45 (95% CI 1.14-5.26, test for heterogeneity: Q = 2.53, I2 = 21.0%, P = 0.021).  Also, the summary of RR of primary studies, whose results were adjusted based on BMI showed a less value compared to summary of RR of primary studies, whose results were adjusted based on weight control (1.62; 95% CI 1.34-1.97 Vs 2.45; 95% CI 1.14-5.26). Physical activity was adjusted in 4 primary studies, and the summary of RR based on controlling this variable was 1.89 (95% CI 1.22-2.94, test for heterogeneity Q = 6.08, I2 = 50.6%, P = 0.108) ( Table 3).

Discussion
The results of this meta-analysis showed that women with diabetes had a 72% increased risk of endometrial cancer compared to those without diabetes as supports the previous meta-analysis conducted by E. Friberg et al. (31) in 2007. Also, other studies have shown that diabetes increased the risk of endometrial cancer, which is in line with the results of the present study [5,6,16,23,26,32,35].  Based on subgroup analysis, the risk of endometrial cancer in case-control studies was higher than in cohort studies, and a higher risk was observed in hospital-based studies compared to population-based studies. [36][37][38].
Since the results of case-control and hospital-based studies are more prone to be affected by confounders therefore the calculated risk might be overestimated. [39][40][41].
In our meta-analysis, heterogeneity was 66.7% for overall risk, which was reduced by subgroup analysis based on type of study, so the heterogeneity for each group for RR in cohort studies, case-control studies, population-based studies, and hospital-based studies were 70.7, 22.7, 0, and 0%, respectively. Furthermore, in this study, it was found that the levels of heterogeneity in physical activity, weight, and BMI had decreased remarkably. It can be concluded from the analysis that the causes of heterogeneity in determining the overall risk of endometrial cancer in women with diabetes in the present meta-analysis were type of study, adjusted co-variables and geographical area (Fig. 2).
Obesity, which is one of the most important factors in diabetes, can cause hormonal imbalances in the body, and this in turn predisposes a person to endometrial cancer [26,[42][43][44]. One of the risk factors for type 2 diabetes is obesity, which is also a major risk factor for endometrial cancer. Although the precise mechanisms and pathways are uncertain, it could be hypothesized that endometrial carcinogenesis is that exposure of the endometrium to excess estrogen unopposed by progesterone increases the mitogen activity of endometrial cells [45,46]. In this meta-analysis the summary of RR of primary studies, whose results were adjusted based on BMI showed a less value compared to summary of RR of primary studies, whose results were adjusted based on weight control. In women with obesity the levels of estradiol and estrogen are higher than women with normal weight, [47,48], and this could be one the possible reason for the increase risk of endometrial cancer because of obesity [48]. However, results of several studies showed that other factors, such as higher insulin levels and growth factors, may also increase the risk of endometrial cancer in women with obesity [49,50]. Moreover, long-term insulin therapy may also be responsible for increased risk of endometrial cancer in women with diabetes (31).
In this study, the authors performed subgroup analysis based on type of primary studies, geographical area, and adjusted covariate. However, we could not perform subgroup analysis based on type of diabetes (type 1 and type 2) because the early studies did not specify or separate the types of diabetes. Diabetes is a chronic disease, whose diagnosis may not be accurate and specific, in which case it would lead to classification bias (non-differential misclassification). Therefore, the overall results obtained from primary studies should be interpreted with caution.
However most of included case-control and cohort studies in this meta-analysis controlled the variables of obesity and sedentariness but it is of utmost importance to consider the effect of confounding variables (sedentariness, hormonal disorders, and obesity) on determining the relationship between diabetes and risk of endometrial cancer in women. The major strength of this updated meta-analysis in compare to previous one is that more primary studies identified and included [8], therefore distinguished effects of diabetes on risk of developing endometrial cancer based on adjustments to BMI/weight presented with larger sample size (larger effect size).

Limitations
Included primary studies did not mention the duration of diabetes and type of treatment (oral anti hyperglycemic agents and/or insulin). Furthermore, identifying women with diabetes in the primary studies was almost based on their self-reports. Since the primary studies did not consider type of diabetes therefore it was not possible to estimate the possible risk separately in in type 1 and type 2 diabetes.

Conclusions
Diabetes seems to increases the risk of endometrial cancer in women, and this finding can be useful in developing endometrial cancer prevention plans for women having diabetes. Authors' contributions LS, HRB, and YM conceptualized the idea for this review, formulated the review question, and objectives, assisted with the development of the final search strategy, contributed to the data analysis/ interpretation, and writing the manuscript. ZN, ABSK, SKH and FV contributed to the conceptualization Test for heterogeneity between population-based and hospital-based case-control studies of the final review question, formulation of the review objectives, data