LAUSR

Introduction Suboptimal levels of selenium (Se) or selenoprotein P (SELENOP), which is an antioxidant protein that also distributes Se from the liver to target tissues, may contribute to risk of colorectal cancer (CRC) development, as we previously showed using serum samples taken pre-diagnostically in the European Prospective Investigation into Cancer and Nutrition (EPIC) cohortHughes et al, 2015 Int J Cancer 136: 1149–61). However, the relationship between Se and survival outcomes following cancer diagnosis is more uncertain. Here, we examined the association of Se status with mortality from CRC and overall mortality in the same study group. Material and methods Se was measured by reflection X-ray fluorescence spectrometry and SELENOP by immunoluminometric sandwich assay. Multivariable adjusted Cox proportional hazard models were used to calculate the hazard ratio (HR) and 95% confidence interval (CI) of the association between Se and SELENOP and CRC-death and all-cause mortality. Results and discussions Higher levels of Se showed non-significant inverse associations with reduction in both CRC and overall mortality: Respective multivariable adjusted HRs for the fifth quintile versus the first quintile (HRQ5 vs. Q1) were 0.76 (95% CI: 0.52–1.11, Ptrend =0.10), and 0.82 (95% CI: 0.56–1.16, Ptrend =0.14). Higher levels of SELENOP were also not associated with a statistically significant reduction in CRC mortality (HRQ5 vs. Q1 = 0.83, 95% CI: 0.57–1.19, Ptrend =0.33). However, higher SELENOP concentrations were associated with a significant reduction in overall mortality (HRQ5 vs. Q1 = 0.70, 95% CI: 0.50–0.98, Ptrend =0.05). Similar results were also obtained by tumour site and sex. Possible interactions of potential effect modifiers and sensitivity analyses showed no considerable change in these estimates, although CRC stage data sensitivity analyses were significant for the association between Se and overall mortality (HRQ5 vs. Q1 = 0.89, 95% CI: 0.80–0.99, Ptrend =0.03). Conclusion We found no major association of Se status markers with survival after CRC diagnosis, but an association of SELENOP with overall mortality. Detailed investigation of Se metabolism is needed to further explore relevance for CRC prognosis especially for individuals of suboptimal SELENOP status.