Inflamm Bowel Dis. 2020 Jul 6;izaa152. doi: 10.1093/ibd/izaa152. Online ahead of print.

Andrea N Georgiou ^1 2, Georgios Ntritsos ^2 3, Nikos Papadimitriou ^2 4, Niki Dimou ^2 4, Evangelos Evangelou ^2 5

Author information

• ¹Medical School, University of Cyprus, Nicosia, Cyprus.
• ²Clinical and Molecular Epidemiology Unit, Department of Hygiene and Epidemiology, School of Medicine, University of Ioannina, Ioannina, Greece.
• ³Department of Informatics and Telecommunications, School of Informatics and Telecommunications, University of Ioannina, Arta, Greece.
• ⁴Section of Nutrition and Metabolism, International Agency for Research on Cancer, Lyon, France.
• ⁵Department of Epidemiology and Biostatistics, Imperial College London, London, UK.

Abstract

Background: Crohn's disease (CD) and ulcerative colitis (UC) are widely associated with smoking in epidemiological studies, whereas there are conflicting results for the association between CD and UC for both coffee and alcohol consumption. Herein, we aimed to investigate whether cigarette smoking and alcohol and coffee consumption are causally associated with either CD or UC.

Methods: We utilized 540 genome-wide significant single-nucleotide polymorphisms for 3 potentially addictive substances-nicotine, alcohol, and caffeine-to assess the association of smoking, coffee, and alcohol consumption with CD and UC (12,194 CD cases, 12,366 UC cases, and 25,042 controls of European ancestry), using Mendelian randomization analysis. Mendelian randomization estimates were used to evaluate the effect of the exposure factors on CD and UC risk. Sensitivity analysis was employed to test for any directional pleiotropy.

Results: We found evidence for a positive causal association between the age of smoking initiation and UC risk and between alcohol consumption and CD risk, which disappeared after sensitivity analysis for both associations (P > 0.05). No evidence for a causal association between cigarettes per day, smoking initiation, smoking cessation, and coffee consumption variables and UC or CD was found.

Conclusions: We found no clear evidence that either genetically predicted smoking, coffee consumption, or alcohol consumption are causally associated with the risk for CD or UC, although our findings indicate a potential positive association between the age of smoking and UC and between alcohol consumption and CD.