Adiposity distribution and risks of twelve obesity-related cancers: a Mendelian randomization analysis

Background

There is convincing evidence that overall adiposity (as measured by body mass index) increases the risks of several cancers. Whether there are similar relationships between these cancers and the distribution of adiposity is unclear.

Methods and findings

In the absence of well-powered individual studies we utilised two-sample Mendelian randomization (MR) to examine causal relationships of five adiposity distribution traits (abdominal subcutaneous adipose tissue; ASAT, visceral adipose tissue; VAT and gluteofemoral adipose tissue; GFAT, liver fat, and pancreas fat) with the risks of 12 obesity-related cancers (endometrial, ovarian, breast, colorectal, pancreas, multiple myeloma, liver, kidney (renal cell), thyroid, gallbladder, oesophageal adenocarcinoma, and meningioma) and cancer subtypes/subsites. We then used multivariable MR to investigate whether plausible molecular intermediates could potentially be mediating the relationships identified. We used the largest available GWAS from European populations for all traits (sample size across all GWAS ranged from 8,407 to 728,896 (median: 57,249); cancer GWAS ranged from 279 to 133,384 cases (median: 4,532) and 3,456 to 727,247 controls (median: 68,802)).

We found evidence that higher ASAT increased risks of endometrial cancer (inverse variance-weighted (IVW) odds ratio (IVW OR) per standard deviation (SD) higher ASAT = 1.79, 95% confidence interval (CI) = 1.18 to 2.71), liver cancer (IVW OR per SD higher ASAT = 3.83, 95% CI = 1.39 to 10.53) and oesophageal adenocarcinoma (IVW OR per SD higher ASAT = 2.34, 95% CI = 1.15 to 4.78). Conversely, we found evidence that higher GFAT decreased risks of breast cancer (IVW OR per SD higher GFAT= 0.77, 95% CI = 0.62 to 0.97) and meningioma (IVW OR per SD higher GFAT = 0.53, 95% CI = 0.32 to 0.90). We also found evidence for an effect of higher VAT and liver fat on increased liver cancer risk (IVW ORs per SD higher adiposity trait = 4.29 and 4.09, 95% CIs = 1.41 to 13.07 and 2.29 to 7.28, respectively). Multivariable MR analyses suggested that traits related to insulin signalling, sex hormones, and inflammation may play important roles in mediating the effects of adiposity distribution on obesity-related cancers.

Conclusions

Our analyses provide novel insights into the variability of the effect of adiposity distribution on cancer risk, with respect to both adiposity trait and cancer type, which would not be possible in a conventional observational analysis given the lack of available samples with all required traits measured. These findings demonstrate that adipose tissue at different anatomical locations may have differential effects on adiposity-related molecular traits. These insights enhance our understanding of the complex relationship between adiposity and cancer risk and highlight the importance of adipose tissue distribution alongside maintaining a healthy weight overall for cancer prevention.