Purpose: To examine the prospective association of serum iron, copper, and zinc with cancer mortality.
Methods: The study sample included 3000 men and 3244 women free from cancer at baseline who participated in the Second National Health and Nutrition Examination Survey. Vital status at follow-up was identified by the Social Security Administration's death file and the National Death Index. Iron, transferrin saturation (TS), copper, and zinc were categorized into 4 levels using the 10th, 50th, and 90th percentiles for cutoffs. Relative risks (RRs) were derived from the proportional hazard models after adjustment for a number of potential confounders.
Results: Three hundred seven cancer deaths (ICD-9 140-195, 199-208) were identified during 83,664.4 person-years of follow-up. Cancer mortality per 1000 person-years was 3.7 (4.7 for men and 2.8 for women). For men and women combined, the adjusted RRs (95% confidence intervals, CI) for the four levels were 0.96 (0.57-1.61), 1.00 (reference), 1.12 (0.80-1.58), 1.86 (1.07-3.22) for iron; 0.97 (0.56-1.70), 1.00 (reference), 1.36 (0.99-1.87), 1.82 (1.10-3.02) for TS; 0.76 (0.44-1.31), 1.00 (reference), 1.10 (0.77-1.58), 1.89 (1.07-3.32) for copper; and 0.75 (0.50-1.13), 1.00 (reference), 0.64 (0.47-0.88), 0.84 (0.53-1.33) for zinc. When the exposures were analyzed as continuous variables, the adjusted RRs (CI) were 1.66 (1.03-2.68) for 100 microg/dl iron increase, 1.17 (1.01-1.36) for 10% TS increase, 1.98 (1.12-3.50) for 100 microg/dl copper increase, and 0.57 (0.16-1.96) for 100 microg/dl zinc increase. Sex differences in the adjusted RRs for iron, TS, and copper were suggestive.
Conclusion: People with higher serum iron, TS, or copper concentrations had an increased risk of dying from cancer.