Abstract Aim Variation in genomic sequences that control gene expression in target tissues is increasingly recognized as the underlying basis of disease susceptibility, especially cancer. This is particularly relevant to common genetic variation impacting on cancer risk. Exogenous factors may also influence gene expression, potentially leaving a transcriptomic signature of environmental risk factors in normal tissues. Therefore, understanding endogenous and exogenous influences over gene expression patterns in normal tissue is critical to the study of functional genomics and its relationship to disease susceptibility. Here, we investigated demographic and sampling variables that could impact on gene expression in normal colorectal mucosa. Method We prospectively collected normal mucosa from 424 patients undergoing colorectal surgery or outpatient assessment through surgical stripping of normal mucosa from resected colorectal specimens or rectal biopsy. Gene expression was assessed using Illumina HT-12 microarrays and analysed against demographic (age, gender, BMI) and sampling factors (general anaesthesia, cleansing bowel agents, sample site, time to RNA preservation) using adjusted linear regression modelling. Results Age, gender, smoking status, sampling under anaesthetic, sample site and sampling method were associated with differential gene expression in the adjusted model. BMI or use of cleansing bowel preparation did not impact gene expression. Age was associated with differential expression of 16 genes and significant enrichment in pathways relevant to tumourigenesis, including immune process, cell proliferation, adhesion and death. Sample site was associated with differential expression of 2515 genes, with 1102 genes more highly expressed in the proximal colon (proximal to splenic flexure). Gender and sampling under anaesthetic were associated with differential expression of 99 and 851 genes respectively. Increased time to RNA preservation (45-90 minutes) was associated with enrichment of pathways consistent with tissue ischaemia including ‘response to wounding’, ‘apoptotic process’ , and ‘response to oxygen levels’ . Conclusions Demographic and sampling factors influence gene expression patterns in normal colorectal mucosa, often with a large magnitude of effect. Meanwhile, greater time to RNA preservation is associated with patterns of gene expression consistent with tissue ischaemia which questions the generalisability of assessment of gene expression patterns generated from post-mortem studies. These results highlight the importance of fully adjusted expression analyses and may indicate mechanisms underlying age, gender and site-specific differences in CRC incidence, progression and outcome.

Factors influencing patterns of gene expression in large bowel mucosa