Predicting Adverse Outcomes in Sickle Cell Disease

Abstract

Source: Miller ST, Sleeper LA, Pegelow CH, et al. Prediction of adverse outcomes in children with sickle cell disease. N Engl J Med. 2000;342:83–89.Knowing which children with sickle cell anemia are most likely to develop complications might help pediatricians to tailor therapy, counsel families, and design research studies for new treatments. Miller et al sought to determine early predictors of complications of sickle cell disease in a prospective study of 392 children followed from early infancy to about age 10. In multiple sites across the United States, 414 infants with homozygous sickle cell anemia or sickle cell–ß°-thalassemia were enrolled before age 6 months. Twenty-two (5%) were lost to follow-up. Of 392 remaining children, 70 (18%) had an adverse outcome, defined as death (n=18), stroke (n=25), frequent pain (average of 2 events/year, n=17), or recurrent acute chest syndrome (average of 1 episode/year, n=10). After multivariate analysis, 3 statistically significant predictors of future adverse outcome were: dactylitis before age 1 (relative risk, 2.55; 95% CI, 1.39 to 4.67); hemoglobin < 7 g/dL (relative risk, 2.47; 95% CI, 1.14 to 5.33); and leukocytosis in the absence of infection during well child visits (relative risk, 1.80; 95% CI, 1.05–3.09; median leukocyte count 13.7 ± 4.5 x 103). Elevated pocked-red-cell count before one year was predictive of severe disease, but was only measured in 40% of the cohort. In a comparison cohort, 20% of patients with severe disease were identified using a predictive model generated from these data, with a 1% false positive rate. The authors conclude that the risk of severe complications later in life can be predicted by features of sickle cell disease that can be readily detected in the first 2 years of life: dactylitis, severe anemia, and leukocytosis.Research on the pathophysiology of sickle cell anemia has been translated into clinical applications that have increased survival and improved quality of life, including prophylactic penicillin and vaccination to prevent sepsis and hydroxyurea to treat crises.1 New therapies such as stem cell transplantation offer hope for cure, but carry significant risks.2 These new therapies are most beneficial if employed early in life prior to the onset of significant organ damage. Therefore, we need ways to identify, early in life, those patients at risk for frequent pain, stroke, acute chest syndrome, or death. This study attempts to construct a predictive model to identify such children. The authors determined that 3 easily measured parameters—dactylitis, hemoglobin <7 g/dL and leukocytosis >13,000/cu.mm in the absence of infection—were predictive of severe disease. Elevated pocked-red-blood cell count was also predictive of severity in the 40% of patients in whom it was measured, but this test is not readily available. Leukocytosis may predict severity by reflecting damage to the vascular endothelium, related in part to abnormal leukocyte adhesions and/or release of cytotoxic enzymes, and by increasing platelet adhesion, which increases the risk of stroke.3–4To minimize the likelihood that a patient who would not develop severe disease would be referred for transplantation or undergo therapies that carry a significant risk, the authors adopted a classification that was associated with a low false positive rate. Using a predicted probability of severe disease of 36% derived from a receiver-operating-characteristic curve, their model has a sensitivity of 20% and a specificity of 99%. In other words, only about one-fifth of patients with severe disease would be identified, but only about 1% of patients identified as high risk would be misclassified. This model is potentially useful to clinicians managing patients for whom risky therapeutic interventions are contemplated.