Genetically predicted 17β-estradiol and systemic inflammation in women: a separate-sample Mendelian randomisation analysis in the Guangzhou Biobank Cohort Study

Background Many chronic diseases are characterised by low-grade systemic inflammation. Oestrogens may promote immune response consistent with sex-specific patterns of diseases. In vitro culture and animal experiments suggest oestrogens are anti-inflammatory and might thereby protect against low-grade systemic inflammation. Evidence from epidemiological studies is limited. Using a Mendelian randomisation analysis with a separate-sample instrumental variable (SSIV) estimator, we examined the association of genetically predicted 17 beta-estradiol with well-established systemic inflammatory markers (total white cell count, granulocyte and lymphocyte count). Methods A genetic score predicting 17 beta-estradiol was developed in 237 young Chinese women (university students) from Hong Kong based on a parsimonious set of genetic polymorphisms (ESR1 (rs2175898) and CYP19A1 (rs1008805)). Multivariable linear regression was used to examine the association of genetically predicted 17 beta-estradiol with systemic inflammatory markers among 3096 older (50+ years) Chinese women from the Guangzhou Biobank Cohort Study. Results Predicted 17 beta-estradiol was negatively associated with white blood cell count (-6.3 10(3)/mL, 95% CI -11.4 to -1.3) and granulocyte count (-4.5 103/mL, 95% CI -8.5 to -0.4) but not lymphocyte count (-1.5 10(3)/mL, 95% CI -3.4 to 0.4) adjusted for age only. Results were similar further adjusted for education, smoking, use of alcohol, physical activity, Body Mass Index, waist-hip ratio, age of menarche, age at menopause, use of hormonal contraceptives and hormone replacement therapy. Conclusions Endogenous genetically predicted 17 beta-estradiol reduced low-grade systemic inflammatory markers (white blood cell count and granulocyte count), consistent with experimental and ecological evidence of 17 beta-estradiol promoting immune response. Replication in a larger sample is required.