Abstract 886

The pathophysiology and etiology of ethnic neutropenia (NP) have not been well elucidated. Several small studies have described leukopenia, and lower hemoglobin (Hb) levels among certain ethnic groups, including African Americans (AA). Some studies have shown that use of steroids resulted in a sub-optimal response, but the cause of this phenomenon has not been adequately explored. Therefore it is important to characterize racial differences in distribution of absolute neutrophil count (ANC) and to explore contributing factors. We searched the National Health and Nutrition Examination Survey III (NHANES III) for linkage between white count and other laboratory measurements which may be associated with NP. All subjects in the community-based NHANES III selection process were stratified by age and over sampled for AA. Data were available on 2709 non-Hispanic whites (W) and 3327 AA, >1 and <20 years of age. Case evaluation included a laboratory panel, which provides the basis for this investigation. NP was defined as an ANC <1500. For each combination of race and sex, we characterized the skewed distribution of ANC, using a combination of two normal curves with different means and variances. This suggests that there may be two populations contributing to the overall distribution. The mean ANC of the population with the lower ANC was 2400 and 3000 for AA and W respectively; the % of the total population contributing to this sub-group was 53% and 37% for AA and W. This indicated that there is a larger subgroup with a lower ANC in AA than in W. For purposes of analysis, cases were categorized as neutropenic (ANC <1500, all subjects) AA=241, W=31; and as low ANC (LANC), ANC <2400 and <3000 for AA and W respectively, AA=1038, W=720. Univariate analysis was used to study differences between neutropenic and LANC cases by race. Both AA and W subjects with LANC had significantly lower Hb, red blood cell (RBC) and platelet count, and higher iron, total iron binding capacity (TIBC), and transferrin saturation levels, suggestive of disordered marrow function and iron utilization. MCV and vitamin B12, were not significantly different for either race and RBC folate was only significantly lower for white LANC subjects. Within race, comparisons of subjects with NP vs normal ANC showed neutropenic AA had significantly lower Hb, higher transferrin saturation and normal MCV, and neutropenic W subjects had a significantly lower MCV, not seen in AA. This study confirms ethnic differences in Hb levels and ANC may be related. While these differences are statistically different they may not be large enough to be clinically meaningful. By using a mixture modeling approach, we characterized the skewed distributions of ANC and find supporting evidence that these racial differences may be due to a genetic polymorphism which needs further investigation.