Relationship between transferrin saturation and iron stores in U.S. African-american and caucasian populations

In previous analyses of transferrin saturation (TS), we found TS subpopulations consistent with population genetics for common loci that influence iron metabolism in African Americans and Caucasians (Gordeuk et al., Blood 1998:91:2175; McLaren et al., Blood 1995:96:2021). However, TS can be elevated sporadically or reduced with inflammation as well as iron deficiency. To determine if TS subpopulations have different iron stores as assessed by serum ferritin concentration (SF), we applied mixture modeling to the distribution of TS in data from the third National Health and Nutrition Examination Survey. As shown in the Table, after adjustment for diurnal variation in TS and age effects in SF, three TS subpopulations were identified with progressively increasing means. These subpopulations had corresponding increases in mean SF for each sex and ethnic grouping (P<0.03 for all). pulation 2 24.9% 37.4 + 6.38 202(188,216) 247(173,331) Caucasian Men Subpopulation 1 81.4% 28.9 ±7.52 149(144,154) N-2322 Subpopulauon 2 17.6% 45.3 + 8.55 164(151,174) n;11.1 243(159,346) African-American Women Subpopulauon 1 81.9% 21.6 + 6.66 88(83,92) N'1047 Subpopulation 2 17.2% 35.6 + 6.85 104(92,117) n;7.29159(88,250) Caucasian Women Subpopulation 1 86.0% 25.9 ±7.66 66 ( 62, 67) N-2475 Subpopulation 2 13.5% 45.7 + 8.73 92(83,104) ;9.23 110(81,144) The TS findings are consistent with population genetics for single major loci affecting the distributions: satisfying Hardy-Weinberg conditions, the sum of the square roots of the proportions with the lowest and highest mean TS is approximately 1 in African Americans (1.062 for males and 1.000 for females) and in Caucasians (1.000, 0.961). In Caucasians the major locus is presumably the HFE gene. The possible major iron-loading defect in African Americans could be different, as Subpopulations 2 and 3 had mean TS values that were from 7% (95% C.I.: 5.4%, 7.6%) to 19% (15.4%, 22.4%) lower than those for Caucasians of the same gender (P<0.0001 for all) and mean SF values that tended to be substantially higher. The relationship between TS and SF in this study suggests that initial screening for hemochromatosis and iron overload using both TS and SF should be considered.

Duke Scholars

Author Victor Hasselblad Biostatistics & Bioinformatics, Division of Biostatistics