The association between weight at birth and breast cancer risk revisited using Mendelian randomisation

Siddhartha Kar ,Thilo Dörk,Manuela Gago-Domínguez,Daniel Considine,Håkan Olsson,Hermann Brenner,Jacques Simard,Rulla M Tamimi,Irene L Andrulis,Kevin Punie,Els Van Nieuwenhuysen,Stephen Burgess,Dezheng Huo,D Gareth Evans ,Kenneth Muir ,Zheng Wang ,Jenny Chang‐Claude ,Artitaya Lophatananon,Jingmei Li,Mikael Hartman,Sara Margolin,Roger L Milne,Rudolf Kaaks,Graham G Giles,Paolo Radice,Camilla Wendt,Paul D.p Pharoah

doi:10.1007/s10654-019-00485-7

Abstract

Observational studies suggest that higher birth weight (BW) is associated with increased risk of breast cancer in adult life. We conducted a two-sample Mendelian randomisation (MR) study to assess whether this association is causal. Sixty independent single nucleotide polymorphisms (SNPs) known to be associated at P < 5 × 10−8 with BW were used to construct (1) a 41-SNP instrumental variable (IV) for univariable MR after removing SNPs with pleiotropic associations with other breast cancer risk factors and (2) a 49-SNP IV for multivariable MR after filtering SNPs for data availability. BW predicted by the 41-SNP IV was not associated with overall breast cancer risk in inverse-variance weighted (IVW) univariable MR analysis of genetic association data from 122,977 breast cancer cases and 105,974 controls (odds ratio = 0.86 per 500 g higher BW; 95% confidence interval 0.73–1.01). Sensitivity analyses using four alternative methods and three alternative IVs, including an IV with 59 of the 60 BW-associated SNPs, yielded similar results. Multivariable MR adjusting for the effects of the 49-SNP IV on birth length, adult height, adult body mass index, age at menarche, and age at menopause using IVW and MR-Egger methods provided estimates consistent with univariable analyses. Results were also similar when all analyses were repeated after restricting to estrogen receptor-positive or -negative breast cancer cases. Point estimates of the odds ratios from most analyses performed indicated an inverse relationship between genetically-predicted BW and breast cancer, but we are unable to rule out an association between the non-genetically-determined component of BW and breast cancer. Thus, genetically-predicted higher BW was not associated with an increased risk of breast cancer in adult life in our MR study.

Highlights

The hypothesis that the risk of developing breast cancer in adulthood may be increased by factors that first act in utero—in particular by fetal exposure to higher levels of maternal estrogen—was first proposed in 1990 [1]
Of the 18 single nucleotide polymorphisms (SNPs) associated with overall breast cancer risk at P < 0.05, the direction of association with birth weight (BW) was inverse for 13 SNPs
Our study provides no evidence to support the association between higher weight at birth and ods used the 49-SNP instrumental variable (IV) and all other methods, unless otherwise specified in the plots, used the 41-SNP IV

Summary

Introduction

The hypothesis that the risk of developing breast cancer in adulthood may be increased by factors that first act in utero—in particular by fetal exposure to higher levels of maternal estrogen—was first proposed in 1990 [1]. Extended author information available on the last page of the article of intrauterine hormones on fetal growth and of the extent of the fetal mammary stem cell pool from which breast tumours may eventually arise [2,3,4,5,6] When considered overall, these studies have suggested that higher weight at birth may be associated with an increased susceptibility to breast cancer in later life [7,8,9,10,11,12,13,14,15,16,17,18,19,20], but a few studies have failed to demonstrate this association [21,22,23,24,25,26,27,28]. It has not been possible to determine whether this association is causal and to dissect whether it is birth weight per se or an external factor influencing

Methods

Results

Conclusion