Serum C-reactive protein levels and colorectal cancer mortality.

Abstract

360 Background: Chronic inflammation has been causally linked to colorectal cancer (CRC), and use of NSAIDs has been associated with reduced risk. Prediagnostic C-reactive protien (CRP) levels, a highly sensitive marker of inflammation, have been weakly associated with increased CRC incidence. However, their relationship with CRC mortality has not been studied well. We hypothesized that elevated baseline CRP levels in general population will predict increased CRC mortality. Methods: This cohort study used CRP data from the Third National Health and Nutrition Examination survey, 1988-94 (NHANES III), with follow-up through 2006. Of the 15,832 eligible adults, NHANES III classified 65% as having CRP levels below detection (≤0.21mg/dL). Using this as the reference, we categorized the remaining participants in three approximately equal groups, and calculated hazard ratios for CRC, all-cancer mortality excluding CRC, and overall mortality due to non-cancer causes. Results: Median follow-up period was 14.2 years. In age adjusted (not shown) and multivariable adjusted models (Table), we observed strong, dose-response associations between CRP levels and CRC mortality. Associations between CRP levels and mortality due to other causes were much weaker. Conclusions: In this large, representative study of U.S. adults, we obtained significantly higher HRs for CRC mortality, as compared to mortality from other cancer and non-cancer causes, making these results unlikely to be explained by residual confounding or other biases. Further, since mortality is a function of both incidence and survival, it provides a more valid estimate of the prognostic value of CRP compared to incidence alone. Further evaluation of CRP may help stratify high risk groups for screening and prognosis, and potentially identify those who might benefit from anti-inflammatory therapy. [Table: see text]