Statistics from Altmetric.com
Population-based studies showed that non-alcoholic hepatic steatosis is less common in women than in men. Anti-oestrogens double the risk of non-alcoholic steatohepatitis,1 which may be interpreted as indirect evidence for a protective role of endogenous oestrogens against steatohepatitis. Experimental evidence further suggests that oestradiol causes a reduction of hepatic steatosis in mice unable to produce oestrogens.2 We thus hypothesised that the menopause as a natural state of oestrogen deficiency might increase the risk of hepatic steatosis.
The Study of Health in Pomerania is a cross-sectional population-based survey in northeast Germany.3 We restricted our analyses to 808 women aged 40–59 years without seropositivity to hepatitis B antigen and anti-hepatitis C virus. The study was approved by the ethics committee of the University of Greifswald. All participants gave written informed consent.
Menopause was defined as 12 consecutive months of amenorrhoea or cessation of menstruation after iatrogenic intervention. Serum glutamate oxaloacetate transaminase, glutamic pyruvic transaminase and γ glutamyl transpeptidase levels were measured photometrically. Levels exceeding the 75th centile for each marker were considered to be raised. In liver ultrasound, a hypoechogenic pattern was defined as the presence of a bright pattern with contrast between hepatic and renal parenchyma.4 Hepatic steatosis was defined as the presence of a hypoechogenic ultrasound pattern and at least two raised levels among three serum transaminase levels. Multivariable statistical analyses were performed by logistic regressions. Sets of confounders were defined using directed acyclic graphs.5
Descriptive statistics showed the association between menopausal status and liver-related characteristics, which remained stable after adjustment for age (table 1). A first sufficient set of covariables included age, smoking and unmeasured confounders (fig 1A). Controlling analyses for age and smoking only marginally affected the relationship investigated (table 1, model 1). In a second directed acyclic graph (fig 1B), alcohol use and nutritional factors were further assumed as determinants of hepatic steatosis and also included in the sufficient set of covariables. This extended model did not substantially affect the association of interest (table 1, model 2). Model 2 was then supplemented by covariables without using any causal diagrams. These analyses gave non-conclusive results (table 1, model 3). Exclusion of all women with alcohol consumption >20 g/day (n = 122; 15.1%) changed the major results only marginally.
Our findings are in good agreement with a Korean ultrasound study6 in which the menopause was also a determinant of hepatic steatosis in women. Our results are further consistent with analyses of the National Health and Nutrition Examination Survey III.7 Non-alcoholic hepatic steatosis as defined by increased concentrations of liver enzymes was twice as common in postmenopausal as in premenopausal women.7
Oestrogens lead to a preferential fat accumulation in the gluteofemoral region8 and, conversely, the loss of oestrogens during menopause transition is associated with an increase in central fat.9 In obese women with hepatic steatosis, liver volumes correlate highly with visceral adipose tissue.10 The relationship between menopausal status and hepatic steatosis found in our study may thus reflect the changes in fat distribution during menopausal transition.
Causal diagrams enable the accurate selection of covariables from a variety of potential variables influencing the studied association. It has been shown that an arbitrary inclusion of variables in multivariable models may not only lead to an underestimation of the association of interest but also introduce bias in the analyses.5 According to our directed acyclic graphs, metabolic factors should be regarded as mediators rather than confounders for the relationship investigated.
We conclude that menopausal status is associated with hepatic steatosis. The prognostic relevance of this association with respect to incident cardiovascular disease and diabetes has to be proved.
Funding: This work is part of the Community Medicine Research (CMR) net of the University of Greifswald, Germany, which is funded by the Federal Ministry of Education and Research (grant no. ZZ9603), the Ministry of Cultural Affairs and the Social Ministry of the Federal State of Mecklenburg-West Pomerania. The CMR encompasses several research projects, which are sharing data of the population-based Study of Health in Pomerania (http://www.community-medicine.de).
Competing interests: None.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.