Investigating the causal relationship between type 2 diabetes mellitus and ovarian cancer using two-sample mendelian randomization

Abstract

Ovarian cancers have registered rising cases of morbidity and mortality over the years. There is an assumption that Type 2 Diabetes Mellitus has a causal relationship with ovarian cancer due to the alarming rising incidence statistics. This research aimed at using a two-sample Mendelian Randomization design to undertake the causal relationship investigation. The specific objectives were to find out whether there exists heterogeneity, horizontal pleiotropy, and the causal relationship between Type 2 Diabetes Mellitus and ovarian cancer. Some of the epidemiologists have applied observational studies for example, cohort and case-control to investigate this causal relationship but ended up with conflicting points of view. The reason for the opposing results is that observational studies are prone to confounding errors and reverse causation, which has led to inaccurate conclusions. This research employed Mendelian randomization technique which uses genetic variants as instrumental variables, which undergo random allocation at conception hence it is therefore not affected by confounding factors. Apart from that, in this method the genetic variants were non-modifiable and thus not altered by reverse causation. The study used the inverse variance weighted technique and Mendelian Randomization-Egger method to obtain the causal estimates and test for the model sensitivity using the R software. The study indicated that there was no evidence of causal relationship between Type 2 Diabetes Mellitus and ovarian cancer (Mendelian Randomization-Egger: beta= -0.0476, standard error = 0.0619, p-value = 0.4479, Inverse Variance Weighted: beta = -0.0165, standard error = 0.0257, p-value = 0.5217). The odds ratios indicated that the two-sample Mendelian randomization had the power to detect 0.0464 and 0.0164 decrease in variability per 1 standard deviation for Mendelian Randomization-Egger and Inverse Variance Weighted respectively (Mendelian RandomizationEgger: odds ratio = 0.9536, confidence interval: 0.8447, 1.0765, Inverse Variance Weighted: odds ratio = 0.9836, confidence interval: 0.9352, 1.0345). This approach alleviated the usual problem of reverse causation and confounding factors hence depicting clearly that there is no causal relationship between Type 2 Diabetes Mellitus and ovarian cancer