Abstract

Sickle cell disease (SCD) is highly clinically variable. Because prediction of adverse outcomes is imprecise, it is difficult to determine which young children would benefit most from early, specific therapies, such as hydroxyurea, chronic transfusion, or stem cell transplantation. Therefore, identification of prognostic features in early childhood would be clinically important. Birth weight is a risk factor for cerebral and coronary vascular disease in the general population, so it might also be predictive of morbidity or mortality due to sickle vasculopathy. Polymorphisms of red blood cell transfusion-related antigens may also modify the clinical severity of SCD. These antigens are integral components of the erythrocyte membrane, polymorphisms of which could alter red cell physiology. For example, Lewis (Le) blood group phenotypes have been associated with clinical severity of SCD (

Blood
2002
:
100
,
449
–50a
). Therefore, we aimed to determine whether birth weight and blood group antigens predicted adverse outcomes in children with SCD. We studied a subset of the Dallas Newborn Cohort (
Blood
2004
:
103
,
4023
–7
) who had either SS or sickle-β°-thalassemia (Sβ°) and ≥5 years of follow-up (birth through last clinical encounter). We defined four adverse outcomes: death, overt stroke, and a lifetime hospitalization rate for painful crisis or acute chest syndrome (ACS) of ≥0.5 episodes/year. Logistic regression analysis was performed using 5 predictor variables: birth weight, ABO blood group, and the Lewis (Le), Duffy (Fy), and Kidd (Jk) antigens. The Le, Fy, and Jk antigens have 4 phenotypes each: (a+b+), (a+b−), (a−b+), or (a−b−). Of 301 eligible subjects, 178 (173 SS, 5 Sβ°; 88 female, 90 male) had complete predictor data (both birth weight and red cell phenotype) and were included in this analysis. Mean age and follow-up were 12.0 and 10.8 years. Mean birth weight was 3168 g (SD 575 g). The most frequent blood groups were O 51%; Le(a−b+) 51%; Fy(a−b−) 70%; and Jk(a+b−) 57%. Ninety-two (51.7%) subjects experienced at least one adverse outcome: 1 (0.6%) died, 5 (2.8%) had stroke, 31 (17.4%) had frequent pain, and 21 (11.8%) had frequent ACS. The multivariate model predicted 72.5% of outcomes accurately. Although the results of the model χ² test were not statistically significant (P=0.238), adjusted odds ratios (OR) showed that subjects with either phenotype Jk(a+b−) [OR=2.30, 95% CI 1.04 – 5.10, P=0.041] or Jk(a−b+) [OR=3.56, 95% CI 0.91 – 13.94, P=0.068] were more likely to have adverse outcomes than those with Jk(a+b+). None of the remaining explanatory variables significantly predicted adverse outcomes. In summary, we show that birth weight does not predict the severity of SCD, and we do not confirm prior studies that demonstrated a prognostic significance of the Le blood group. However, we do show that certain polymorphisms of Jk (an erythrocyte urea transporter) appear to increase the risk of adverse outcomes in children with SCD. The exclusion of potentially eligible subjects because of missing predictor data may have biased the findings of this analysis. Thus, further study is necessary to confirm these findings and to identify other novel clinical and laboratory predictors of outcome in SCD.

Author notes

Corresponding author