Risk Stratification in Pediatric Sickle Cell Disease Using Absolute Reticulocyte Counts

Abstract 1014

All sickle cell anemia patients (HbSS, SCA) have the same genetic mutation, but the clinical phenotype is highly variable and difficult to predict prior to the onset of disease complications. If severe SCA could be predicted early in life, then disease modifying therapies could be instituted prior to the onset of organ damage.

To determine if reticulocyte levels in SCA patients are useful in disease severity prediction, a convenience sample of 50 children with SCA was enrolled in an observational study. After consent and assent were obtained, discarded peripheral blood obtained during routine clinic visits was collected and analyzed within 48 hours of collection and storage at 4°C. Hematologic data, including absolute reticulocyte counts (ARC), was collected using a Sysmex Hematology Analyzer. Clinical events were examined prospectively from the time of enrollment and retrospectively if the patient had events prior to study enrollment. Clinical events included: painful crises requiring hospital admission (VOC), acute chest syndrome, and splenic sequestration that occurred prior to the onset of chronic therapy. ARC and hematologic data were collected over time and analyzed using Cox regression analysis to determine the relationship between ARC levels and time to the first event. To evaluate the utility of ARC in risk stratification, patients were divided into two groups: ARC less than 200K/uL (ARC<200) and ARC greater than or equal to 200K/uL (ARC≥200).

Initial analyses were performed using steady state ARCs prior to the first clinical event (pre-event baseline ARC). Steady state was defined as a sample collected at least 30 days from an acute illness and at least 60 days since the patient received a blood transfusion. Patients were followed an average of 6.7 years (range 0.82–16.8 years), which provides 332 person years of follow-up. A time dependent Cox regression analysis of pre-event baseline ARC≥200 compared with ARC<200 over the first 3 years of life generated a hazard ratio of having a first clinical event of 4.7 [95%CI 1.83–12.29 (p=0.0013)].

Maximum ARC before age 6 months (defined as the infant baseline ARC) was utilized for additional analyses. Cox regression analysis revealed that those subjects with an infant baseline ARC≥200 had 3.2 times the risk of having an event within the first 3 years of life than the group with an infant baseline ARC<200 (HR 3.15, 95%CI 1.54–6.45, p=0.0017). Forty-eight percent (12/25) of patients with an infant baseline ARC <200 had an event by age 3 years compared to 88% (22/25) of patients in the ARC≥200 group (p=0.001). The number of events per patient years was higher in the infant baseline ARC≥200 group (0.74 events/patient years of follow up vs. infant baseline ARC<200, 0.29 events/patient years of follow up, p=0.0004). The median time to first event in the infant baseline ARC≥200 group was shorter [1.39 years (95% CI 0.87–1.93)] than the baseline ARC<200 group [3.06 years (95%CI 1.71–3.80)].

These data suggest that both pre-event and infant baseline ARCs assist with risk stratification in infants and young children with SCA. Further studies are needed to determine if ARC risk stratification assessments are sufficiently robust for the guidance of treatment decisions for pediatric SCA.