Validation of inflammatory genetic variants associated with long-term cancer related fatigue in a large breast cancer cohort

Background: Studies to date have reported several associations between single nucleotide polymorphisms (SNPs) and cancer related fatigue (CRF), but have been limited by small sample sizes, missing adjustment for relevant covariates or multiple testing, as well as varying CRF definitions, i.e. time and method of assessment. This study aimed to validate previously reported associations using the largest independent breast cancer sample to date and to evaluate further functional cytokine variants in relation to total CRF and all relevant CRF subdomains (physical, cognitive, and affective CRF). Method: 45 candidate SNPs in inflammatory pathway genes were selected based on previous reports (16 SNPs) or regulatory function (29 SNPs). Breast cancer patients recruited between 2002 and 2005 provided information on CRF at first follow-up (FU1) (N = 1389) and second follow-up (FU2) (N = 950), a median of 6.2 years and 11.7 years respectively after diagnosis. SNP associations were assessed using linear regression models on CRF scores separately for FUl and FU2. Additionally, patients with persistent fatigue (fatigued at both time-points) were compared to those never fatigued using logistic regression models (N = 684). All analyses were adjusted for relevant covariates. Secondary analyses were conducted for CRF subdomains. Results: For total CRF none of the previously reported associations were confirmed after correction for multiple testing. The p-value distribution of all SNPs was not different than the one expected by chance. Analyses of CRF subdomains yielded a significant association between TNF-alpha rs3093662 and persistent physical CRF (Odds Ratio (OR) = 3.23, 95% Confidence Interval (CI) = 1.71-6.10, p = 0.0003). Conclusion: We were unable to confirm previously reported findings, suggesting that individual SNPs are unlikely to be of clinical utility. Further investigations in well powered studies are warranted, which consider genetic heterogeneity according to subdomains of CRF.