Impact of hydroxyurea on clinical events in the BABY HUG trial

The Pediatric Hydroxyurea Phase 3 Clinical Trial (BABY HUG; clinicaltrials.gov no., NCT00006400) was an National Heart, Lung, and Blood Institute (NHLBI)– and National Institute of Child Health and Human Development (NICHD)–sponsored multicenter, randomized, double-blind placebo-controlled study of daily oral hydroxyurea in children with sickle cell anemia (SCA; HbSS, and Sβ⁰ thalassemia) 9 to 18 months of age. The primary aim of the trial was to determine whether daily hydroxyurea would reduce by ≥ 50% spleen and renal damage as measured by radionuclide scanning. Although failing to achieve this aim, there were significantly fewer sickle cell disease–related clinical events in the hydroxyurea group including painful events, acute chest syndrome (ACS), hospitalizations, and transfusions.¹ 

Based on the results of the BABY HUG trial, the authors concluded that hydroxyurea can now be considered for all patients starting at a young age.¹  However, there are barriers to implementation into clinical practice including patient and provider misconceptions about toxicity.^2,3  We now report detailed analysis of sickle cell disease–related clinical events and other adverse events (AEs) from the BABY HUG trial, which further documents the beneficial effects and lack of toxicity of hydroxyurea for very young patients with SCA.

Subjects with HbSS or HbSβ⁰-thalassemia 9 to 18 months of age were recruited without regard to clinical severity and enrolled into the BABY HUG study.⁴  The trial began in October 2003 and was completed in September 2009. All children were taking prophylactic penicillin and received age-appropriate immunizations. Each participating site obtained local institutional review board approval to conduct the study. Consent was obtained before any study procedures, and all studies were conducted in accordance with the Declaration of Helsinki.

Data regarding AEs, including sickle cell–related clinical events, were collected as part of the AE reporting process. AEs were determined as required by 21 CFR 312 and in accordance with 45 CFR 46.⁵  For this trial, serious adverse events (SAEs) included the sickle cell–related clinical events of splenic sequestration, stroke, transient ischemic attack (TIA), and ACS. SAEs also included any AE that resulted in death, was life-threatening, or required hospitalization > 7 days or intensive care unit admission. Because of the unique manifestations of SCA and expected admissions for the primary disease process, the standard SAE reporting requirement was changed from any inpatient hospitalization or prolongation of existing hospitalization to hospitalization > 7 days or requiring intensive care unit admission. Any AE that was both treatment-related and unexpected was classified as an SAE.

AEs were captured by parent report at routine visits every 2 weeks until a stable dose of hydroxyurea was established, then every 4 weeks, and by review of emergency department, hospital, and clinic records. The Clinical Center staff determined the degree of severity. SAEs were reported to the Medical Coordinating Center (MCC) at Clinical Trials and Surveys Corporation within 24 hours of the Clinical Center becoming aware of the event and were reviewed by an independent classification committee. In addition, AEs were reviewed by the NHLBI project officer and the MCC medical consultants who were aware of treatment assignment. These individuals determined whether an AE should be reclassified as an SAE.

Parents and caregivers were educated about dactylitis and painful events including home management with fluids, ibuprofen and acetaminophen with codeine, and were questioned about occurrence of episodes at each visit. Initially, painful events were defined as pain lasting 2 hours or more without obvious cause and requiring the use of one or more doses of nonsteroidal or narcotic pain medications, while dactylitis was defined as pain and tenderness with or without swelling of the hands and/or feet. In 2005, at the request of the National Institute of Child Health and Human Development (NICHD) to help support Food and Drug Administration (FDA) approval of a pediatric indication for hydroxyurea, criteria for reporting pain were expanded to include any mention of pain, whether or not it was treated or required a medical visit.

Parents and caregivers were educated about signs and symptoms of ACS at regular clinic visits and advised to seek medical attention if these occurred. ACS was defined by a new pulmonary infiltrate and at least 3 of the following findings: chest pain, temperature elevation > 38.5°C, tachypnea, wheezing, or cough.

Fever was defined by a temperature > 38.5°C. Parents were advised to seek urgent medical attention for their child's fever, which was managed as per local institutional protocol.

Splenic sequestration was characterized by an increase in spleen size by ≥ 2 cm below the costal margin compared with previous measurement and a reduction of hemoglobin level by ≥ 2.0 g/dL or 20% from recent steady-state values, with or without a drop in platelet or leukocyte counts. Parents and caregivers were instructed in techniques of spleen palpation at each clinic visit and asked to feel for the child's spleen daily. Parents were asked to immediately report a newly palpable spleen or one larger than previously noted.

Cholelithiasis was defined by the presence of biliary stones in the gallbladder or common bile duct, with minimal or no symptoms.

Gastroenteritis was defined as at least 2 episodes of nausea, vomiting (excluded if it occurred with cough or medication dosing), or diarrhea (excluded if it occurred with laxative use). Constipation was defined as decreased stool frequency as determined by the caretaker or healthcare provider. There were no protocol definitions of eczema or asthma.

Mild to moderate neutropenia was defined as absolute neutrophil count (ANC) 0.5-1.249 × 10⁹/L and severe neutropenia was defined as ANC < 0.5 × 10⁹/L. Thrombocytopenia was defined as platelet count < 80 × 10⁹/L. Severe anemia was defined as hemoglobin < 7 g/dL and absolute reticulocyte count (ARC) < 80 × 10⁹/L.

Asymptomatic subjects were defined as those who had no history of pain, dactylitis, ACS, or splenic sequestration before enrollment.

χ² or Fisher exact tests were applied to compare the frequency of AEs between the hydroxyurea and placebo groups. The Fisher exact test and the Wilcoxon rank-sum test were used to compare associated symptoms, complications, and treatments of ACS. The Fisher exact test was used to evaluate the proportion of neutropenic events associated with viral illness. Logistic regression was used to evaluate predictors of thrombocytopenia. Cox models were used to compare febrile and infectious events between treatment groups, as well as clinical events of children who were asymptomatic at enrollment in both treatment groups. The log-rank life test was used to compare time to first event. The Wilcoxon rank-sum test was used to compare the number of hospitalizations per subject associated with pain, dactylitis, and ACS and length of hospitalizations in the 2 groups. A Cox model was used to compare the rate of prolonged hospitalizations between the 2 groups. A generalized estimating equation (GEE) model using Poisson regression to model serial outcome counts and a logarithm link function for the model parameterization was performed for multivariate analyses of baseline predictors of multiple pain events. A GEE model was also performed for multivariate analysis of gastroenteritis events. This model included the use of time-dependent variables including age at event, hemoglobin, white blood cell count (WBC), ANC, ARC, and absolute HbF (all calculated as the moving average of the 3 steady-state values before the event time). Absolute HbF was calculated using the percentage of HbF and hemoglobin concentration.

All analyses used SAS Version 9.2 (SAS Institute). Analyses were conducted by Z.L. and B.T. All authors had access to primary data on request.

One hundred ninety-three subjects were randomized and 192 began study medication. Subjects received placebo or hydroxyurea for a planned duration of 2 years; 86% of enrolled subjects completed 24 months of study treatment.⁶ 

Baseline characteristics of the study population have previously been reported with no significant differences in age, sex, genotype, clinical severity, laboratory values, or physical findings between the 2 groups.¹  In addition, there were no differences in baseline rates of hospitalization, dactylitis, pain, ACS, splenic sequestration, or transfusion. At baseline, 52 (54%) of 96 children in the hydroxyurea group and 49 (51%) of 97 children in the placebo group were clinically asymptomatic.

Overall, there were 2560 AEs and 107 SAEs reported during a total of 374 patient-years of on-study observation. None of the AEs were classified by the investigators, NHLBI project officer or the MCC medical consultants as both unexpected and treatment related.

The incidence of pain was 94 events per 100 patient-years in the hydroxyurea group compared with 203 per 100 patient-years in the placebo group (hazard ratio [HR] 0.59; P = .002).¹  Data regarding frequencies of painful events are shown in Table 1. In the hydroxyurea group, there were no baseline parameters that predicted multiple pain events.

The rate of dactylitis was 12.7 per 100 patient-years in the hydroxyurea group compared with 66.5 per 100 patient-years in the placebo group (HR 0.27; P < .001).¹  Data regarding the frequency of dactylitis are shown in Table 1.

The incidence of ACS was 4.2 per 100 patient-years in the hydroxyurea group compared with 14.6 per 100 patient-years in the placebo group (HR 0.36; P = .02),¹  and a higher proportion of subjects in the placebo group had multiple ACS events (P = .03; Table 1). Both rates were lower than that seen in similarly aged children in the Cooperative Study of Sickle Cell Disease (CSSCD); the rate of ACS in the CSSCD was 27-34 events per 100 person-years for children ages 1-3 years.⁷  Characteristics of subjects with ACS are summarized in Table 2.

A 3-year-old male child in the placebo group was diagnosed with acute ischemic stroke after 102 weeks on study. He was subsequently started on chronic blood transfusion therapy. There were no reported TIAs in the BABY HUG trial. There were 4 children (3 receiving placebo) with abnormal TCD results at study exit (ages 35, 36, 39, and 42 months). The impact of hydroxyurea on TCD velocity was previously reported; at baseline, TCD velocity was negatively correlated with hemoglobin level.^8,9  The average increase in TCD velocity during the study was significantly lower in the hydroxyurea group.^1,8,9 

On abdominal ultrasound at study exit, 8 children in the hydroxyurea group and 6 children in the placebo group had sludge and 5 children in the hydroxyurea group and 5 children in the placebo group had cholelithiasis.

There were 12 splenic sequestration events in 8 subjects in the hydroxyurea group (median age 28 months; range 12-36 months) and 12 splenic sequestration events in 9 patients in the placebo group (median age 22 months; range 11-34 months).

Sixty-nine subjects in the hydroxyurea group had a total of 232 hospitalizations (123 hospitalizations per 100 patient-years) compared with 84 subjects in the placebo group with 324 hospitalizations (175 per 100 patient-years; HR 0.73; P = .05). There was no statistically significant difference in the rate of prolonged hospitalization (> 7 days) between the groups (2.7 per 100 patient-years in the hydroxyurea group and 4.3 per 100 patient-years in the placebo group; P = .40). When the primary reason for hospitalization was evaluated, hydroxyurea therapy was associated with fewer hospital admissions for fever (181 vs 240, P = .01), pain (73 vs 128, P < .001), and ACS (8 vs 27, P = .02). There was no difference in length of hospital stay between the 2 groups for pain (hydroxyurea 4.0 days, placebo 3.6 days), dactylitis (hydroxyurea 3.9 days, placebo 3.4 days), and ACS (hydroxyurea 5.5 days, placebo 4.7 days).

Twenty (21%) of 96 patients in the hydroxyurea group received a total of 35 transfusions compared with 33 (34%) of 97 patients in the placebo group (63 transfusions; P = .04). Seven of 20 transfused patients (35%) in the hydroxyurea group and 17 (52%) of 33 patients in the placebo group received 2 or more transfusions (P = .06). The most common reasons for transfusion were splenic sequestration/splenomegaly (18 in each group) and ACS (2 in the hydroxyurea group and 12 in the placebo group).

There were no deaths during the treatment period.

A summary of infectious complications is shown in Table 3. There were a total of 754 febrile events with 47 (24%) of 193 children having 6 or more episodes of fever over the 2-year study period. Fever was the main reason for 421 hospitalizations during the trial. The rate of gastroenteritis was 13.8 per 100 patient-years in the hydroxyurea group and 37.9 per 100 patient-years in the placebo group (HR 0.35; P < .001). In multivariate analysis, ANC (P = .006) and placebo group (P = .002) positively correlated and WBC (P = .006) negatively correlated with gastroenteritis.

There were 8 episodes of bacteremia/sepsis (hydroxyurea group 3, placebo group 5), accounting for 1.1% of febrile episodes. Three children had blood cultures positive for Streptococcus pneumoniae. All 3 S pneumoniae infections occurred in children not taking hydroxyurea (2 in the placebo group and one assigned to, but not yet exposed to hydroxyurea). These children were 17, 22, and 33 months old at the time of bacteremia/sepsis and all were up-to-date with age appropriate immunizations including protein-conjugated pneumococcal vaccines. Overall, the rate of pneumococcal bacteremia/sepsis for the entire cohort was 0.8 cases per 100 patient-years. Two children had blood cultures positive for Salmonella. One child had bacteremia with Hemophilus parainfluenzae in the setting of acute chest syndrome and hemoglobin of 4.9 g/dL. Two children had blood cultures positive for Streptococcus viridans; these were probably skin contaminants because neither child had a central venous access device.

In addition, a 19-month-old child in the placebo group was treated for presumed bacterial meningitis. The child was given ceftriaxone in the setting of fever before presenting one day later with fever, irritability, and neck stiffness. Twenty-four hours after the parenteral antibiotics cerebrospinal fluid showed one red blood cell and 52 white blood cells with 66% neutrophils and 34% mononuclear cells; bacterial gram stain, culture, and antigen testing were negative. Two children in the placebo group were diagnosed with and treated for acute culture-negative osteomyelitis. No patients required vasopressor support, mechanical ventilation, or dialysis.

The numbers of subjects with neutropenia, thrombocytopenia, and severe anemia were previously reported.¹  In the hydroxyurea group, 3 of 5 neutropenic events were associated with viral illness compared with one of 2 in the placebo group (P = 1.0). Neutropenia was never associated with invasive bacterial infection. Severe anemia occurred in 3 subjects with aplastic crises: a 19-month-old child in the hydroxyurea group with a hemoglobin of 3.8 g/dL and ARC of 16 × 10⁹/L; a 30-month-old child in the placebo group with a hemoglobin of 3.3 g/dL and ARC of 71 × 10⁹/L, and a 36 month old child in the placebo group with a hemoglobin of 3.3 g/dL and ARC of 4 × 10⁹/L. Thrombocytopenia occurred in 11 children in the hydroxyurea group and 7 children in the placebo group. Thrombocytopenia was associated with lower hemoglobin (P < .001). Of 11 subjects treated with hydroxyurea who also had thrombocytopenia, none had concomitant neutropenia, whereas one of 7 placebo-treated subjects with thrombocytopenia had concomitant neutropenia.

When only the children who were asymptomatic at baseline were compared, those children in the hydroxyurea group had significantly decreased episodes of dactylitis, hospitalizations, and transfusions (Table 4).

Twelve (13%) children in the hydroxyurea group and 11 (11%) on placebo had eczema (P = .80). Overall, 15 (8%) children had a diagnosis of asthma by the end of the study, 9 (9%) children in the hydroxyurea group, and 6 (6%) children in the placebo group (P = .44). No child developed a malignancy.

Further analysis of data from the BABY HUG trial and continued follow-up of participants provide critical information for clinicians considering hydroxyurea prescription for young children with SCA. Overall, young children taking hydroxyurea in the BABY HUG trial had significantly fewer acute sickle cell–related complications than those taking placebo. The lower rates of complications were associated with fewer hospitalizations and transfusions. There were no clinically important AEs or SAEs that were related to hydroxyurea.

Hydroxyurea treatment was associated with a 2.2-fold lower rate of pain and 5.2-fold lower rate of dactylitis. Children taking hydroxyurea were less likely to have multiple episodes of pain or dactylitis, and all patients with ≥ 4 episodes were in the placebo group. This reduction in pain rate with hydroxyurea was similar to that seen in the Multicenter Study of Hydroxyurea (MSH), indicating that the salutary effects of hydroxyurea on painful vaso-occlusive events can be observed at all ages.¹⁰  The proportion of patients hospitalized for pain was substantially lower in those taking hydroxyurea than those taking placebo; there was no difference in their length of stay, which was approximately 4 days and similar to other reports.^11-13  In the hydroxyurea group, none of the baseline parameters was associated with the occurrence of multiple pain events, perhaps because there were so few children in the hydroxyurea group with multiple pain events. Subjects who were asymptomatic at study enrollment had lower rates of dactylitis, as well as fewer transfusions and hospitalizations (Table 4). Further study is needed to determine whether hydroxyurea impacts pain severity, intensity of outpatient and inpatient analgesic treatment, and time missed from school, and if clinical response can be accurately predicted in clinical practice. Lowered rates of pain and dactylitis should have a substantial long-term impact with regard to lowering the medical costs of outpatient and inpatient management¹⁴  and improving quality of life.¹⁵ 

Hydroxyurea was also associated with a 3.5-fold lower rate of ACS and a lower rate of recurrent ACS. A smaller percentage of children in the hydroxyurea group were transfused for ACS compared with the placebo group; however, this difference was not statistically significant. ACS in children receiving placebo was associated with lower HbF. The overall rate of ACS in BABY HUG was lower than reported in the CSSCD. We believe this was partially attributable to improved immunization against S pneumoniae and Hemophilus influenzae, and uniform use of penicillin prophylaxis; in one CSSCD study, 14% of infants, age 0-2 years, with ACS were bacteremic, mostly with S pneumoniae.¹⁶  Prevention of ACS has prognostic importance because early ACS is associated with recurrent episodes later in childhood¹⁷  and pulmonary disease and mortality later in life.¹⁸ 

Despite the beneficial hematologic effects of hydroxyurea, we did not identify a significant difference in gallbladder disease during our 2 years of follow-up; although hemolysis was reduced in the subjects in the hydroxyurea group, it was not completely eliminated. Rates of splenic sequestration and transfusion for these events were similar between the 2 groups. The median time to the first event was 39 weeks for 8 hydroxyurea-treated subjects, and 53 weeks for 9 placebo-treated subjects (P = .79).¹  In the subjects who were asymptomatic at baseline, there was a trend toward a lower rate of splenic sequestration in the hydroxyurea group, suggesting that children may need to start hydroxyurea before symptoms to realize some of its protective effects.

Fever was the most common AE during the study, accounting for approximately one-third of all events. While over 50% of children with fever were admitted to the hospital only 5% of the febrile episodes were associated with bacteremia/sepsis, meningitis, osteomyelitis, or ACS. The small percentage of children with invasive infections is likely attributable to the introduction of penicillin prophylaxis and pneumococcal and meningococcal vaccines.^19,20  Unfortunately, risk of infection still exists because of the emergence of nontypeable and nonvaccine strains of pneumococcus.^21,22 

In the BABY HUG trial, all 3 S pneumoniae infections occurred in children not taking hydroxyurea. Follow-up studies of children on hydroxyurea will be needed to determine whether indeed hydroxyurea has any protective effects against pneumococcal bacteremia/sepsis in children with SCA. Recent evidence from a mouse model indicates that hydroxyurea could have a protective effect against invasive pneumococcal disease.²³ 

Unexpectedly, gastroenteritis was significantly less frequent in children taking hydroxyurea; this effect has not been previously reported in other adult or pediatric hydroxyurea studies. It is tempting to speculate that hydroxyurea's ability to replenish nitric oxide (NO)^24,25  may be related to this finding. NO improves gut motility²⁶ ; however, as described for priapism,²⁷  chronic NO depletion may change the sensitivity of smooth muscle to NO and disrupt normal protective feedback mechanisms, resulting in a robust response to a modest increase in NO because of infection/inflammation. In addition, NO is bacteriostatic for several organisms, including Escherichia coli,²⁸  and bactericidal for some.²⁹  It may be that alteration in bowel flora by hydroxyurea somehow reduced the incidence of gastroenteritis.

As previously reported, mild to moderate myelosuppression occurred more frequently in the hydroxyurea group; however, mild to moderate neutropenia is an expected treatment effect and severe neutropenia was rare.¹  All subjects initially received hydroxyurea at 20 mg/kg/d, but 9 (9%) had dose reductions (typically to 17.5 mg/kg/d) during the treatment phase because of excessive myelosuppression. Currently, the standard clinical practice is to hold or reduce the dose of hydroxyurea in the setting of moderate hematologic toxicities, particularly neutropenia below 1.0-1.5 × 10⁹/L.³⁰  However, this may not be necessary because myelosuppression was not associated with infections or other cytopenias. Although the BABY HUG trial protocol did not include dose escalation to maximum tolerated dose, based on previously published studies^31-34  and results from the BABY HUG Follow-up Study I (NCT00890396), we expect that these subjects will tolerate dose escalation to MTD as they get older.

Our study was potentially limited by identification of many clinical events by self-report. However, particularly for painful events, self-report is required to identify the extent of disease.^35,36  Study coordinators were diligent in documenting clinical events at each monthly study visit, records were obtained from pediatricians and outside hospitals/emergency departments to document events, and the MCC reviewed source data for 100% of events.

In conclusion, hydroxyurea significantly reduced the frequency of painful events, ACS, and associated transfusions and hospitalizations, without significant hydroxyurea-related AEs. Importantly, hydroxyurea was not associated with an increased rate of infection in this very young population. Reduction of sickle cell–related complications in early childhood should translate to long-term benefits. As previously reported, disease severity was not a criterion for enrollment into the trial, and family perception of disease severity was not a primary reason for enrollment in the trial.³⁷  Fifty-four percent of children in the hydroxyurea group were clinically asymptomatic at enrollment, and they also had significant reductions in sickle cell–related events. Therefore, beneficial effects of hydroxyurea occur in both asymptomatic and symptomatic young children with SCA.

Follow-up studies analogous to the recently published adult MSH follow-up¹⁸  will evaluate the long-term safety and effectiveness of hydroxyurea and its potential for improved longevity and quality of life. Continued collection of clinical event and safety data are being performed in the current BABY HUG Follow-up Study (NCT00890396) that has been extended through 2016. Because of the benefit of hydroxyurea in decreasing acute sickle cell–related complications, clinicians should consider shifting their practice to prescribe hydroxyurea therapy for all very young children with SCA, rather than treating only those most severely affected.³⁸ 

The authors acknowledge the efforts of the BABY HUG subjects and their families, the contributions of all who participated in BABY HUG (http://www.c-tasc.com/cms/StudySites/babyhug.htm). The authors especially thank Dr Bea Files and Dr Michael Jeng, the medical consultants assigned to reviewing AEs.

This work was supported by NHLBI/National Institutes of Health Contracts N01-HB-07150 through N01-HB-07160, with additional support from the Best Pharmaceuticals for Children Act and the NICHD.

National Institutes of Health

Contribution: C.D.T., B.A.F., Z.L., S.T.M., B.T., R.E.W., and W.C.W. designed the analysis; C.D.T. and Z.L. analyzed the data; and C.D.T., B.A.F., Z.L., S.T.M., R.K., R.I., P.S., J.L., O.A., R.E.W., and W.C.W. wrote the manuscript; and all authors approved the manuscript submission.

Conflict-of-interest disclosure: The authors declare no competing financial interests.

A complete list of the BABY HUG Investigators appears online at http://www.c-tasc.com/sites/default/files/StudySites/babyhug.html.

Correspondence: Courtney D. Thornburg, MD, MS, Duke University Medical Center, DUMC Box 102382 Durham, NC 27710; e-mail: courtney.thornburg@dm.duke.edu.