Chapter  165:  Hydroxyurea for the Treatment of Sickle Cell Disease

Literature Sources

We searched for articles using both electronic and hand searching. In March 2007, we searched the MEDLINE^® and EMBASE databases. We repeated the search in May 2007, adding a supplemental search targeting thrombocythemia. On June 30, 2007, the MEDLINE^® and EMBASE^® searches were updated and additional searches were executed using TOXLine and CINAHL. All searches were limited to English-language articles involving treatment of humans. Review articles were excluded from the searches. Searches were not limited by date of publication or subject age.

Eligibility Criteria

An article was included if it addressed one of the key questions. An article was excluded if it was (1) not written in English, (2) contained no original data, (3) involved animals only, (4) was solely a report of an in vitro experiment, or (5) was a case series. We excluded studies with fewer than 20 patients unless the article was primarily reporting on toxicities in sickle cell disease. We excluded trials involving other diseases if fewer than 20 patients received hydroxyurea. We allowed cohort studies of diseases other than sickle cell disease only if they described more than 100 patients treated with hydroxyurea. Although we excluded case series because they do not provide sufficient data about the effectiveness of a medication we included case reports if they had information regarding the dose of hydroxyurea and the duration of treatment that could be use to assess a causal relationship with potential toxic effects.

Quality Assessment

We graded the included studies on the basis of their quality with regard to reporting relevant data. For the RCTs, we used the scoring system developed by Jadad et al.^b For the observational studies (both cohort studies and controlled clinical trials), we created a quality form, based on those previously used by our EPC, that was aimed at capturing data elements most relevant to study design. We designed questions to evaluate the potential for selection bias (three items) and to assess the potential for confounding (five items). For our assessment of the quality of the qualitative studies we reviewed, we developed a nine-item form to identify key elements that should be reported when describing results from qualitative research, including a description of the population and subjects and transparency of the data collection procedures. Similarly, to assess the quality of the surveys we included, we created an eight-item form assessing information about the survey methods, population, and validity and reliability of the instruments used. A pair of reviewers each performed the quality assessment independently. In the case of the RCTs, a third reviewer reconciled the results of the first two reviewers; for the other study designs, the results of the two reviewers were averaged. The overall score was the percentage of the maximum possible score, ranging from 0 to 100 percent. The results for RCTs were reported as 0 to 5 points. We considered high-quality studies to be those with a Jadad score of 4 or 5, or those receiving 80 percent or more of available quality points. However, no study was excluded from review on the basis of its quality score.

Data Extraction

We used a sequential review process in which the primary reviewer abstracted all the relevant data into abstraction forms, and a second reviewer checked the first reviewer's forms for completeness and accuracy. Reviewer pairs were formed to include personnel with both clinical and methodological expertise. Differences were resolved by discussion. We then created detailed evidence tables containing information extracted from the eligible studies.

Grading of the Evidence

At the completion of our review, we graded the quantity, quality, and consistency of the best available evidence addressing Key Questions 1, 2, and 3 by adapting an evidence grading scheme recommended by the GRADE Working Group and the EPC guide that is was under development at the time of the review. We applied evidence grades to the bodies of evidence about the efficacy and/or effectiveness of hydroxyurea for the treatment of sickle cell disease in one assessment. In terms of the strength of the study designs, we considered RCTs best, followed by non-randomized controlled trials and observational studies. We assessed the quality and consistency of the best available evidence, including an assessment of limitations to individual study quality (using individual quality scores), certainty regarding the directness of the observed effects in studies, precision and strength of findings, and availability (or not) of data to answer the Key Question. We classified evidence bodies pertaining to each Key Question as shown in Table 1. The evidence from case reports was graded according to the criteria of the World Health Organization (WHO) Collaborating Center for Drug Monitoring.

Efficacy and Effectiveness of Hydroxyurea in Children

A single, small, placebo-controlled randomized trial of hydroxyurea for 6 months in Belgian children with sickle cell disease reported that the rate of hospitalization and number of days hospitalized per year were significantly lower in the hydroxyurea group (1.1 admissions, p=0.0016 and 7.1 days, p=0.0027) than in the placebo group (2.8 admissions and 23.4 days). Hb F% increased by an absolute 10.7 percent from baseline in the treated group (p<0.001).

Among the cohort studies, Hb F% was reported as an outcome in 17 studies. The mean pre-treatment Hb F% ranged from 5 to 10 percent, and the on-treatment values were in the range of 15 to 20 percent. The percentage of F cells was less frequently reported, but it increased from baseline in three of the four pediatric studies in which it was reported. Three of these cohort studies were retrospective; two reported increases in Hb F% comparable to those in the prospective studies. Hemoglobin concentrations increased modestly (roughly 1 gm/dL) but significantly across these studies.

The frequency of pain crises was reported as an outcome in five pediatric studies, with a reduction in frequency reported in three. In one retrospective cohort study in a resource-poor environment, the frequency of pain crises declined from a median of 3 per year to a median of 0.8 per year during treatment, with a median followup time of 24 months. Of note is the fact that these results were obtained using a fixed dose of hydroxyurea (15 mg/kg/day). A small, high-quality prospective study found a decrease in pain events, from 3.1 per year in the year prior to hydroxyurea therapy to 1.2 per year during the 18 months of therapy. Hospitalization rates decreased in all four studies describing this outcome. In the retrospective study described above, the hospitalization rates decreased to 0.5 per year during treatment, from a baseline rate of 4 per year. Within the Belgian Registry, hospitalization rates declined to 1.1 per patient-year during the third year of treatment, from 3.2 per patient-year.

One study assessed the impact of hydroxyurea on secondary stroke prevention by enrolling 35 children who needed to discontinue their chronic transfusion protocol. The average hydroxyurea dose was 27 mg/kg/day, and the children were treated for a mean of 42 months. The rate of recurrent ischemic events was 5.7 per 100 patient-years, which is better than was seen in another study in which children discontinued transfusions without starting hydroxyurea. One other study reported that brain images by magnetic resonance imaging (MRI) were stable during the course of treatment in 24 of 25 children. In the Belgian Registry, during 426 patient-years of hydroxyurea treatment, the rate of central nervous system events (stroke or transient ischemic attacks) was 1.3 per 100 patient-years, but no comparison rate was provided.

Based on one randomized trial in children and many observational studies, some of which were high-quality and most of which were consistent in their findings, we graded the evidence as shown in Table 1.

Efficacy and Effectiveness of Hydroxyurea in Adults

Only one randomized trial, the Multicenter Study of Hydroxyurea for Sickle Cell Anemia (MSH Study), tested the efficacy of hydroxyurea in adults with sickle cell anemia, with six additional analyses either based on this trial or on followup studies. The significant hematological effects of hydroxyurea after 2 years (as compared to the placebo arm) included a small mean increase of 0.6 g/dl in total hemoglobin and a moderate absolute increase in fetal hemoglobin of 3.2 percent. The median number of painful crises was 44 percent lower, and the time to the first painful crisis was 3 months, as compared to 1.5 months in the placebo arm. There were fewer episodes of acute chest syndrome and transfusions, but no significant differences in deaths, strokes, chronic transfusion, or hepatic sequestration. Use of hydroxyurea had no significant effect on annualized costs. It improved the quality of life, but only in those patients who experienced a substantial increase in Hb F%.

In all six prospective cohort studies in adults that reported hematological outcomes, Hb F% increased significantly. The mean baseline Hb F% ranged from 4 percent to 12 percent, and during hydroxyurea treatment, it ranged from 10 percent to 23 percent. As in the pediatric studies, there was a small increase in hemoglobin in most studies. The single retrospective study reported hematological outcomes comparable to those seen in the prospective studies. The number of pain crises was described in three studies. In a study of Sicilians with Hb Sβ thalassemia, the frequency of crises decreased significantly, from a mean of 7 (median of 9) per year to a mean of 1.1 (median 1.8) per year. In the non-randomized study comparing patients receiving hydroxyurea to those receiving cognitive behavioral therapy, those receiving hydroxyurea had fewer pain crises (1.4 per year compared to 4.3 per year, p≤0.05) but this was not a strong study design for assessing such an outcome. Similarly, hospitalization rates decreased consistently in adults treated with hydroxyurea. In the study of Sicilians, the number of hospitalized days per year declined from 22.4 days to 1.2 days (SD =2.3) (p<0.0001). In a retrospective effectiveness study, the rates of hospitalization declined from baseline in the group treated for longer than 24 months (2.1 per year from 3.1 per year, p=0.04). However, in the group treated for fewer than 24 months, the hospitalization rates were not significantly different from baseline values.

Based on one high-quality randomized trial in adults and many consistent observational studies, we graded the evidence as shown in Table 1.

Toxicities of Hydroxyurea in Children and Adults

Our assessment of the strength of the evidence regarding the toxicity of hydroxyurea, when used in children, was generally derived from our review of the report by the panel of experts that had been assembled by the National Toxicology Program (NTP)'s Center for the Evaluation of Risks to Human Reproduction (CERHR). The panel reviewed articles, published through January 2007, that pertained to the evaluation of adverse effects of hydroxyurea on development and reproduction in both humans and animals. Their review was not restricted to the use of hydroxyurea for sickle cell disease. The dosing of hydroxyurea for sickle cell disease is comparable to that in other diseases, although in the case of malignant disease, more drug is often given less frequently (such as 80 mg/kg every 3 days rather than 15–20 mg/kg daily).

The panel concluded that treatment of children aged 5 to15 years with hydroxyurea does not cause a growth delay. They felt there were insufficient data to allow them to evaluate the effects of hydroxyurea on pubertal development. The panel found no data regarding the effects on subsequent generations after exposure of germ cells to hydroxyurea, including exposure during fetal life, infancy, childhood, and adolescence. The CERHR report did not describe any studies on the long-term health effects, including carcinogenicity, of childhood exposure to hydroxyurea; we also found no such studies. The expert panel had concerns about the adverse effect of hydroxyurea on spermatogenesis in men receiving hydroxyurea at therapeutic doses; we also identified case reports of impaired spermatogenesis after hydroxyurea treatment in patients with sickle cell disease, as well as in patients with other illnesses. The CERHR report concluded that the use of hydroxyurea in pregnancy was not associated with adverse perinatal outcomes, but that there were no data on long-term outcomes in children who were exposed in utero. However, the panel expressed concern, based on minimal data from experimental studies, that hydroxyurea might increase the risk of congenital anomalies or abnormalities of fetal growth after exposure of pregnant women to the drug.

We found three cases of leukemia, described in observational studies, in patients with sickle cell disease who had been treated with hydroxyurea. We identified another three case reports of hydroxyurea-treated patients with sickle cell disease who developed leukemia, and one report of a child who developed Hodgkin's lymphoma. Toxicities in patients with sickle cell disease that are probably causally related to hydroxyurea are neutropenia, skin rashes, and nail changes.

We reviewed toxicity reports from hydroxyurea-treated patients with other illnesses and found many reports of leg ulcers and skin cancers. Among the randomized trials enrolling patients with other diseases, no trial demonstrated a greater number of cases of leukemia in the group treated with hydroxyurea. This parameter could not be assessed in the trials enrolling patients with chronic myelogenous leukemia (CML), as progression to acute leukemia was considered a poor response to the intervention and could not be considered a toxicity of treatment. We reviewed a case series of 26 patients with acute myelogenous leukemia (AML) with a unique t (3;21) chromosomal translocation. Among these 26 patients were 15 people with CML who had been treated with hydroxyurea. We found no other reports describing an association between this translocation and hydroxyurea.

We concluded that low-grade evidence suggested that hydroxyurea treatment in adults with sickle cell disease is not associated with an increased risk of leukemia. (Table 2)

High-grade evidence supported the assertion that hydroxyurea is not associated with leg ulcer development in patients with sickle cell disease, although high-grade evidence indicated that it is associated with leg ulcers in patients with other conditions. The evidence was insufficient in sickle cell disease to indicate whether hydroxyurea contributes to skin neoplasms, although high-grade evidence supported its involvement in patients with other illnesses. Similarly, there was insufficient evidence to establish whether hydroxyurea is associated with secondary malignancies in adults with sickle cell disease; the evidence in other diseases was only low-grade.

Barriers to the Use of Hydroxyurea and Other Treatments for Managing Sickle Cell Disease

Only two studies (one in patients and one in providers) investigated barriers to use of hydroxyurea; both used survey data. The study involving patients used a cross-sectional design and showed that the perceived efficacy and safety of hydroxyurea had the strongest association with patients' (or parents') choice of hydroxyurea therapy over other therapies. In the study of clinicians, the reported barriers to use of hydroxyurea for sickle cell disease included patient concerns about side effects and a variety of clinician concerns: the appropriateness of using hydroxyurea in older patients, patient compliance, a lack of contraception in premenopausal women, side effects and carcinogenic potential, doubts about effectiveness, and costs to patients.

We reviewed an additional 47 studies addressing barriers to the treatment of patients with sickle cell disease and interventions to overcome these barriers. In our review of barriers to adequate pain management, we found two factors that were identified as a barrier in more than two studies: negative provider attitudes and poor provider knowledge. Because of the quantity and consistency of these findings, we concluded that the evidence was high-grade that negative provider attitudes are barriers and moderate-grade that poor provider knowledge is a barrier to the use of pain medications in patients with sickle cell disease. The evidence for the remaining barriers to pain management was insufficient to allow us to draw any conclusions.

In our review of the barriers to other therapies for chronic sickle cell disease management, we concluded that the evidence was of a moderate grade that patient sex is not a barrier to use of therapies. Largely because of the paucity and inconsistency of the studies, we concluded that there was only low-grade evidence that patient/family knowledge, the number of hospital visits, and patient age are barriers to the use of therapies.

We identified three studies that tested interventions to improve patient adherence to established therapies for chronic disease management, but none of these three showed any effect on patient adherence. However, given the small sample sizes and the studies' diverse outcome measures, we concluded that there was only low-grade evidence that interventions did not improve patient adherence. In contrast, we identified nine studies that examined the impact of interventions to improve pain management during vaso-occlusive crises, and we concluded that there was moderate evidence that interventions can overcome barriers to the use of pain medications. We also identified one study that investigated the impact of an intervention to improve receipt of routine healthcare and, partly because of the strength of the effect found in the study, we concluded that there is moderate evidence to indicate that interventions can overcome barriers to the receipt of routine, scheduled healthcare for patients with sickle cell disease.

We found it informative that when researchers chose the barriers to investigate, they most often studied patient-related barriers. When patients were asked to identify barriers to the use of therapies, they most often cited provider-related barriers. The barrier to pain management that was most often identified by patients and providers was negative provider attitudes. However, only one of the nine pain management intervention studies addressed this issue directly through provider sensitivity training.

Clinical Characteristics

Patients with sickle cell disease experience both chronic and episodic pain and have a reduced quality of life. ³ Painful crisis is the most common reason for emergency department use by patients with sickle cell disease. ⁴ The pathophysiology of a painful crisis is not entirely clear, and its determinants are uncertain. Some patients have frequent crises and severe disability, whereas others are able to lead relatively normal lives. Much of what we have learned about the incidence of complications in people with sickle cell disease comes from the Cooperative Study of Sickle Cell Disease (CSSCD).⁵ (See list of acronyms.) This federally funded study, begun in 1979, was a large multi-institutional prospective study of the clinical course of sickle cell disease. In this study, the frequency of painful crises was variable: 0.8 episodes per person-year for sickle cell anemia, 1.0 episodes per person-year for Hb Sβ⁰ thalassemia, and 0.4 episodes per person-year for Hb SC disease. ⁶ In a study of 1,056 patients with Hb SS disease in California, 70 percent of patients were admitted for a crisis; the overall rate of hospitalizations for crisis was 57 admissions per 100 years of observation. ⁷

Acute chest syndrome is the most common cause of death and hospitalization in patients with sickle cell disease. ⁸ In a large multicenter study of acute chest syndrome, the working definition was a new pulmonary infiltrate in a patient with chest pain, with a temperature of more than 38.5°C and tachypnea, wheezing, or cough. ⁸ In the study in California, the incidence rate of acute chest syndrome was 14 per 100 years of observation. ⁷ In the CSSCD, acute chest syndrome occurred in nearly 30 percent of 3,751 patients. Its incidence was highest in patients with Hb SS disease (12.8 per 100 patient-years). ⁹

Stroke is another serious consequence of sickle cell disease and is seen more often in children than adults. In the CSSCD, the prevalence of stroke was 4 percent in those with Hb SS disease, with an incidence of 0.61 per 100 patient-years. ⁵ Investigators noted that stroke was associated with all the common genotypes. In the Powars⁷ study in California, 11 percent of patients had suffered a stroke. Children who have had a stroke or who are at risk for stroke (as determined by transcranial Doppler [TCD] flow velocity) are typically treated with a chronic prophylactic transfusion regimen.

Another complication of sickle cell disease that affects patients' quality of life is the development of leg ulcers. In the Powars⁷ study, 14 percent of the patients suffered from this complication. In the CSSCD, 25 percent of all patients had leg ulcers. ⁵ People with Hb SS disease or Hb Sβ⁰ thalassemia are at higher risk of developing leg ulcers than are those with other genotypes. ¹⁰ The ulcers usually occur between the ages of 10 and 50 years and are more common in men than in women. ⁵ Therapy is supportive, involving local care of the ulcer, but many of these ulcers become chronic.

Established Treatments

Most of the therapies offered to patients with sickle cell disease are supportive and do little to change the underlying pathophysiology of the disease. These supportive measures include the use of penicillin prophylaxis in children to prevent pneumococcal disease, routine immunizations, and hydration and narcotic therapy to treat painful events. Some treatments, such as penicillin therapy, have improved both quality of life and survival. ¹¹

Transfusions are often used to increase the oxygen-carrying capacity of the blood and to decrease the concentration of cells with abnormal hemoglobin. In patients with repeated, severe complications of sickle cell disease, simple transfusions or exchange transfusions are often used to preserve organ function and prolong life. In the multicenter study looking at the treatment of acute chest syndrome, 72 percent of the patients received red cell transfusions to treat this acute event. ⁸ As mentioned above, children with a stroke history are treated with chronic transfusion therapy. ¹² Despite the usefulness of chronic transfusion, its long-term effects include iron overload, which can damage the liver.

Currently, hydroxyurea is the only disease-modifying therapy approved for sickle cell disease. Hence, there is great interest in understanding more about its use in treating patients with this group of disorders.

Sources

Our comprehensive search included electronic and hand searching. On March 15, 2007, we ran searches of the MEDLINE^® and EMBASE^® databases. A supplemental search targeting essential thrombocythemia was added to the MEDLINE and EMBASE searches on May 7, 2007. On June 30, 2007, the MEDLINE and EMBASE searches were updated, and additional searches were executed using TOXLine and CINAHL. All searches were limited to English-language articles involving treatment of humans. Review articles were excluded from the searches. Searches were not limited by date of publication or by subject age.

Search Terms and Strategies

Search strategies specific to each database were designed to enable the team to focus the available resources on articles that were most likely to be relevant to the Key Questions. We developed a core strategy for MEDLINE, accessed via PubMed, on the basis of an analysis of the MeSH terms and text words of key articles identified a priori. The PubMed strategy formed the basis for the strategies developed for the other electronic databases (see Appendix A ^*).

Organization and Tracking of the Literature Search

The results of the searches were downloaded into ProCite^® version 5.0.3 (ISI ResearchSoft, Carlsbad, CA). Duplicate articles retrieved from the multiple databases were removed prior to initiating the review. From ProCite, the articles were uploaded to SRS 4.0 (TrialStat ^© 2003-2007). SRS is a secure, Web-based collaboration and management system designed to speed the review process and introduce better process control and scientific rigor. We used this database to store full articles in portable document format (PDF) and to track the search results at the title review, abstract review, article inclusion/exclusion, and data abstraction levels.

Description of Randomized Trials

We identified eight published reports describing results from two randomized controlled trials of hydroxyurea for the treatment of sickle cell disease (Appendix C, Evidence Table 1). These reports presented the results from the MSH, ^{21, 22, 39}–⁴² an extended followup of the participants in the MSH, ⁴³ and a pediatric study in Belgium. ⁴⁴ The MSH began enrollment at multiple centers in North America in 1992, and the results were published in 1995. This study enrolled 299 adults. The Belgian study began enrollment in 1992 at two centers in Europe, and the results of this study were published in 1996 after enrollment of 25 patients. The Belgian study had a crossover design; patients were randomized to receive hydroxyurea or placebo for the first 6 months and then to receive the other treatment for the next 6 months. Hematological outcomes were reported as the change from baseline after 6 months of hydroxyurea, and clinical outcomes were compared between the placebo and hydroxyurea arms. The methods used for dose escalation and monitoring for toxicity were similar in the two studies, except the maximum dose was 35 mg/kg/day in the MSH and 25 mg/kg/day in the Belgian study.

Description of the quality of the studies. Both trials had rigorous eligibility criteria designed to select patients with severe sickle cell anemia or sickle α -thalassemia and minimize the risk of known toxicities. The methodological quality of the MSH was excellent (Jadad score of 5, out of a maximum of 5) and that of the Belgium study was moderate (Jadad score of 3), because the method of masking was not described (Appendix C, Evidence Table 2). Both the MSH and the Belgian study provided a placebo control with masking of patients, but only the MSH study masked the providers and the endpoint adjudication panel. In the MSH, 53 patients (18 percent) had permanent or extended cessation of the study medication: 21 percent in the hydroxyurea group and 14 percent in the placebo group. Frequent reasons for discontinuation were pregnancy (n=16), inactivity (n=18), myelotoxicity at 2.5 mg/kg/day or “simulated toxicity” (n=5), and a need for long-term transfusion therapy (n=5). ³⁹ The Belgian investigators excluded three patients (14 percent) after 4 to 5 months for failing to make the required monthly visits.⁴⁴

Description of the included patients. A description of the patients is given in Appendix C, Evidence Table 3. The MSH included only adults (mean age, 30.5 years), and approximately half of the participants were male. The Belgian study included mostly children (median age, 9 years; range, 2 – 22 years), and approximately half were male. The majority of patients in both studies were African or African American and had sickle cell anemia. The β-globin haplotype was reported only in the MSH study; about 40 percent were homozygous for the Benin haplotype, 20 percent were heterozygotes with the Benin/Central African Republic (CAR) haplotype, and the remainder had other combinations.

Description of Observational Studies (Pre/Post Design or Non-Randomized Control Group)

Design. Our analyses included the results of 37 observational studies of hydroxyurea use in patients with sickle cell disease: 19 in North America, 11 in Europe, 2 in the Middle East, and 3 in Central or South America (Appendix C ^*, Evidence Table 4). The earliest studies we identified were published in 1992, ^{45, 46} and one-quarter of the studies were published in the past 2 years. The studies ranged in size from only 8 patients in a cytotoxicity study ⁴⁷ to 225 patients in the large French cohort. ⁴⁸ More studies were designed to enroll children only (n=20) than adults only (n=12).

The majority of the studies are best described as pre/post studies in which the patients' clinical parameters were described prior to starting hydroxyurea and again after they had been on the drug for a period of time. Nine of the studies were retrospective, ^{45, 48}–⁵⁵ two were cross-sectional, ^{56, 57} and the rest were prospective studies. Three studies described comparison groups of patients who were not treated^{58, 59} with hydroxyurea. ⁶⁰–⁶²

The study goals varied markedly (described in Evidence Table 4). The majority aimed to assess the long-term safety and efficacy of hydroxyurea. Some studies, however, were more specialized, including two that assessed the effect of the drug on transcranial Doppler (TCD) velocities, ^{62, 63} one that assessed the efficacy of a low dose of hydroxyurea, ⁵⁰ three that specifically assessed splenic function, ^{56, 64, 65} one that focused on cutaneous adverse effects, ⁶⁶ one that assessed albuminuria, ⁴⁹ one that assessed secondary stroke, ⁶⁷ and several that looked at malignancy and cytotoxic effects of the drug. ⁴⁷ Most studies reported both efficacy and toxicity data; however, 8 were primarily toxicity studies, ^{47, 48, 53, 57, 66 55, 64, 68}18 were primarily efficacy studies, ^{45, 62, 63, 69 46, 50, 59}–^{61, 70}–⁷⁸ and 11 were primarily effectiveness studies, ^{49, 51, 52, 56, 58, 65, 67, 79}–⁸² although the designs of the efficacy and effectiveness studies were often similar. Toxicity studies without efficacy data are not included in Evidence Table 8. ^{53, 57}

Patient Clusters. We identified four clusters of studies based on the patient populations examined; this approach was taken because of our concern that some patients might have been described in more than one publication. One cluster was comprised of manuscripts related to the Safety of Hydroxyurea in Children with Sickle Cell Anemia (HUG-KIDS) study. HUG-KIDS was a high-quality phase I/II study that began recruiting children in December, 1994 and continued through March, 1996. The first of these studies that we included in this cluster was primarily a toxicity study. ⁶⁸ The second included efficacy data from the children who reached the MTD during the study. ⁷³ The third study was a pre-post effectiveness study that included 15 of the children who had been enrolled in HUG-KIDS. ⁸¹

The second cluster was the Hydroxyurea Safety and Organ Toxicity (HUSOFT) cluster, which consisted of two studies. ^{60, 72} The first was the HUSOFT study itself, which was a cohort study of hydroxyurea use in young children with Hb SS. ⁶⁰ In this publication, the CSSCD cohort was described as a comparison group. The second study was an extension of HUSOFT study, in which investigators followed the patients for a longer time. ⁷²

In our third cluster, we grouped studies from a group of investigators in France. The patients were recruited from centers participating in the French Study Group on Sickle Cell Disease. The first study was a pre-post efficacy study; ⁷⁶ the second study was also an efficacy study that included some (or all) of the patients who had been described in the first study. ⁷⁵ The third publication was a survey of French physicians treating children with hydroxyurea to assess toxicity. ⁵⁵ We expect that those children who were enrolled in the first study were also described by their physicians in the survey. ⁷⁶ Finally, the most recent study was a cohort study⁴⁸ to explore toxicity, which involved recruiting patients who had been in the earlier studies and those identified by the survey. ⁵⁵

The fourth cluster consisted of two publications from the Belgian Sickle Cell Group. ^{58, 82} Both publications described the experiences of patients in their registry, with the latter study including more patients. ⁸² To our knowledge, the studies outside these four clusters included no overlapping patients.

Interventions. The initial dosage and titration schedule for hydroxyurea varied little across studies, with most starting at 15 or 20 mg/kg and titrating upward by 5 mg/kg at some interval (ranging from every 4 weeks to every 6 months) or according to clinical response. One study specifically tested the efficacy of the drug without up-titration, ⁵⁰ and the HUSOFT study did not involve titration. ⁶⁰ A number of the observational studies did not include any description of the dosages received by the patients.

Description of the quality of the studies. The studies that were evaluated with our 16-point scale for assessing the quality of observational studies received between 27 percent ⁴⁷ and 93 percent⁷⁷ of the possible points (Appendix C, Evidence Table 5). There were eight high-quality studies that received more than 80 percent of the quality points. ^{46, 64, 68, 70, 73, 76, 77, 81} There was a small but significant decline in quality score over the years (p=0.01). Most of the studies appropriately chose objective outcomes to report, and most described their interventions with adequate detail. Studies were less complete in their description of their source population, i.e., the population from which their participants were drawn; they were also less complete in their description of the inclusion criteria for their study and in their description of the characteristics of the included patients. Authors did a particularly poor job of describing how complete the adherence to the intervention was and of describing losses to followup. For cohort studies, inadequate description of losses to followup can be an important source of bias. There were no high-quality studies that directly addressed the effectiveness of treatment with hydroxyurea for sickle cell disease, an important issue for the implementation of a therapy. The two studies that used a survey design were not considered high-quality because they provided little detail about the surveyed patients and providers and did not use any validated instruments for collecting information; however, they were probably adequate for their purpose, which was to collect toxicity data from providers. ^{53, 55}

Description of the included patients. A description of the included patients is given in Appendix C, Evidence Table 6. Although not all studies reported the sex of the enrolled patients, for those that did, the percentage of males ranged from 40 percent⁷⁷ to 74 percent. ⁷⁶ The majority of the studies included patients with Hb Sβ thalassemia and with Hb SC disease in addition to Hb SS.

HUSOFT was a study of very young children (mean age of 1.3 years, with a range from 0.5 to 2.3 years). ⁶⁰ The 20 pediatric studies had mean or median ages ranging from 1.3 to 14 years, and the 12 adult studies had mean or median ages from approximately 21 to 33 years. Two studies included adults and children, ^{58, 70} and two did not describe the age range. ^{69, 79} Three of the studies from Europe were exclusively of Caucasian patients. ^{71, 74, 78} Few studies described the clinical activity of their cohorts upon entry, although this information could often be inferred from the inclusion and exclusion criteria. The duration of observation of the enrolled patients varied markedly both across studies and within studies. The studies with the longest median followup times were in the range of 36 to 45 months. Some individuals within each study were followed for much longer, although they were not necessarily treated with hydroxyurea for the duration of their followup.

Efficacy and Effectiveness of Hydroxyurea in Children

The Belgian RCT, a small placebo-controlled study, reported hematological outcomes as the change from baseline after 6 months of treatment with hydroxyurea (Appendix C, Evidence Table 7). In the hydroxyurea group hemoglobin increased by a mean of 0.4 g/dl, and the absolute mean increase in Hb F was 10.7 percent (p<0.001). ⁴⁴ For these children, the rate of hospitalization and number of days hospitalized per year were significantly lower for the hydroxyurea group (1.1 admissions, p=0.0016 and 7.1 days, p=0.0027) than for the placebo group (2.8 admissions and 23.4 days).

Among the observational studies, more reported hematological outcomes than reported clinical outcomes. Hb F% was reported as an outcome in 17 studies (Table 1; Appendix C, Evidence Table 8). In all the studies that reported Hb F% before and during treatment with hydroxyurea, the HbF% increased substantially while patients were being treated. The mean pre-treatment Hb F% ranged from 5 to 10 percent, and the post-treatment values were in the range of 15 to 20 percent. The percentage of F cells was less frequently reported, but it increased from baseline in three of the four pediatric studies in which it was reported. ^{68, 72, 75} In the fourth study, the maintenance of a stable percentage of F cells and Hb F% over 104 weeks of treatment, rather than the decline usually seen in young children, was attributed to hydroxyurea therapy. ⁶⁰ Three of these observational studies were retrospective. ⁴⁹–⁵¹ Of these, two reported increases in Hb F% that were comparable to those in the prospective studies. ^{49, 50} One brief report described 226 patients who had taken hydroxyurea and compared the 38 patients who died to the remainder who survived. ⁵¹ The Hb F% increase was greater in the patients who died than among those who survived, but these two groups were substantially different before treatment. The authors concluded that not all patients benefit from hydroxyurea. Hemoglobin concentration increased modestly (roughly 1 gm/dl) but significantly across these studies.

The frequency of pain crises was reported as an outcome in five pediatric studies. ^{50, 59, 64, 72, 82} In the retrospective study by Svarch et al., the frequency of pain crises declined from a median of 3 per year to a median of 0.8 per year while patients were on treatment, with a median followup time of 24 months. ⁵⁰ It is particularly important to note that these results were obtained in a resource-poor environment (Central America) and used a fixed dose of hydroxyurea of 15 mg/kg/day. Another study reporting pain outcomes was a small study by Hankins et al. ⁷² that prospectively followed 17 children who had been enrolled in HUSOFT during a 4-year extension study. The authors reported 33.8 pain events requiring hospitalization per 100 patient-years while their patients were on treatment, a rate that did not differ from that reported for untreated patients in the CSSCD cohort (32.4 per 100 patient-years, p=0.87). The authors felt that differences in the methods of collecting pain data in their study and in the CSSCD may have biased the results toward a finding of no difference. They observed fewer episodes of acute chest syndrome than were seen in the CSSCD. A small, high-quality study found a decrease in pain events from 3.1 per year in the year prior to hydroxyurea therapy to 1.2 per year during 18 months of therapy. ⁶⁴ Similarly, in a 12-month study by Santos et.al, pain frequency decreased from a median of four episodes per year to two per year during the year of therapy (p=0.0009). ⁵⁹ In the most recent study from the Belgian Sickle Cell Registry, patients had 2.2 pain crises per year that required hospitalization while on treatment, although it is not clear what the baseline rate was for their population. ⁸²

Hospitalization rates were reported in four studies. Again, in the retrospective Central American study, the hospitalization rates decreased to 0.5 per year while on treatment from a baseline rate of 4 per year. ⁵⁰ In the study by Oliveri, rates declined by 75 percent, to 1.7 per year from 6.7 per year; ⁶⁴ compliance with medication use was very high in this study. Similarly, in a small study of severely ill children, the hospitalization rates dropped to 3 per year from 7 per year. ⁷⁷ In the Belgian Registry, hospitalization rates declined to 1.1 per patient-year from 3.2 per patient-year after 3 years. ⁸²

Two studies reported TCD velocities. ^{62, 63} In the study by Kratovil et al., the mean maximum velocity decreased on treatment to 111 cm/sec from a mean maximum of 125 cm/sec. ⁶² A control group that was not treated with hydroxyurea had an increase in velocity over the same time period of 4.7 cm/sec. In the recent prospective study by Zimmerman et al., 37 children had TCD measurements prior to starting hydroxyurea. ⁶³ The children (n=36) reached a stable MTD of 27.0 mg/kg/day and had repeat Doppler studies after a mean of 10 months. Velocities decreased significantly in the right and left middle cerebral arteries, right and left anterior cerebral arteries, and left posterior cerebral artery. In 14 of 15 children with conditional baseline TCD velocities, the values improved; in 5 of 6 with abnormal velocities, whose families refused transfusions, the velocities decreased to less than 200 cm/sec.

One study assessed the impact of hydroxyurea on secondary stroke prevention by enrolling 35 children who needed to discontinue their chronic transfusion protocol. ⁶⁷ The average dose of the drug was 27 mg/kg/day, and the children were treated for a mean of 42 months. Seven children had recurrent ischemic events, for a rate of 5.7 per 100 patient-years. We noted, for comparison, that this rate was higher than the 2.2 per 100 person-years reported in a retrospective cohort study of children who received ongoing transfusions, ⁸³ but it was better than the 70 percent prevalence of recurrent stroke seen in the first year after discontinuing transfusion without alternative treatments. ⁸⁴ One other study reported that brain images obtained by MRI were stable during the course of treatment in 24 of 25 children. ⁶⁵ In the Belgian Registry, during 426 patient-years of hydroxyurea treatment, the rate of central nervous system events (stroke or transient ischemic attacks) was 1.3 per 100 patient-years, but no comparison rate was provided. ⁸²

Four studies assessed splenic function during hydroxyurea therapy. ^{56, 59, 64, 65} One found no difference in the 12 children in whom the number of pitted red blood cells was counted. ⁶⁴ A cross-sectional study used Howell-Jolly bodies as the outcome: ⁵⁶ In the group of patients with spleens, patients on hydroxyurea therapy had a greater number of Howell-Jolly bodies than did those in the group not taking the drug. This relationship was true as well for the patients without spleens, suggesting that Howell-Jolly bodies are not simply a measure of splenic function. In a prospective study of 52 children, of whom 43 had had spleen function measured with scintigraphy both at baseline and on therapy, 6 patients (14 percent) completely recovered splenic function, and 2 (5 percent) had preserved splenic function after a median of 2.6 years of hydroxyurea at the MTD. ⁶⁵ In the study by Santos et al., splenic function as measured by scintigraphy improved in 10 of 21 children was stable in 8, and worsened in 3. ⁵⁹ A retrospective study reviewed the efficacy of hydroxyurea in reducing the progression of microalbuminuria⁴⁹: Of the 17 treated patients without microalbuminuria at baseline, 16 remained free from microalbuminuria; 4 of 9 patients with baseline microalbuminuria normalized their urinalysis during treatment. The study by Santos and coworkers was the only one to describe transfusion use in children, reporting that the transfusion rate decreased from 3.9 per year to 0.43 per year in their 21 treated patients. ⁵⁹

Efficacy and Effectiveness of Hydroxyurea in Adults

In the MSH RCT, significant hematological effects of hydroxyurea after 2 years (relative to the placebo arm) included a small mean increase in total hemoglobin (0.6 g/dl) and moderate increases in fetal hemoglobin (3.2 percent) and mean corpuscular volume (MCV; 10.1 fl) (Appendix C, Evidence Table 7). ^{21, 39, 40} The median number of painful crises was 2.5 per year for those receiving hydroxyurea, as compared to 4.5 per year (a 44 percent decrease, p<0.001), and the time to first painful crisis was 3 vs 1.5 months (p<0.01). There were fewer episodes of acute chest syndrome (25 vs 51, p<0.001), transfusions (55 vs 79 patients, p=0.002, and 423 vs 670 units, p=0.002), but no significant differences in deaths (2 vs 6), strokes/chronic transfusion (2 vs 3), or hepatic sequestration (1 vs 3). ³⁹ In the long-term followup study, which was observational after the initial period of randomization, the mortality rate was 40 percent lower (p=0.04) when patients were taking hydroxyurea (1.5 per 3-month period) than when taking placebo or no treatment (2.6 deaths per 3-month period). However, the long-term mortality, when analyzed according to the initial treatment assignment in the 2-year randomized trial, was similar for the hydroxyurea (3.1 per 100 person-years) and placebo (3.6 per 100-person-years) groups. The rates of stroke, sepsis, and renal and hepatic failure were also similar between the two groups. ⁴³

The MSH trial also included an evaluation of costs and quality of life. Annualized total costs were $16,810 for the hydroxyurea group and $22,270 for the placebo group (p=0.21), with significantly lower costs for hospitalization for pain in the hydroxyurea group ($12,160, p<0.05) than the placebo group ($17,290). ²² The hydroxyurea and placebo groups were similar in terms of all the quality of life measures, but participants with the greatest increase in Hb F (upper half of the change in Hb F) had significantly better “general health now” (p<0.001), decreased pain by 4-week recall (p=0.004), and better general health (p=0.001). ⁴² A sub-study of the MSH, completed at a single institution, evaluated exercise capacity; which increased in the hydroxyurea group when compared to the placebo group. This improvement was accompanied by an increase in weight and decrease in the resting heart rate in the hydroxyurea group (3.2 kg and -14 beats/min, as compared to the placebo group's 1.8 kg and -4 beats/min). ⁴¹

In the six prospective, observational studies of adults that reported hematological outcomes, Hb F% increased significantly in all six studies (Table 2; Appendix C ^*, Evidence Table 8). ^{45, 46, 71, 74, 78, 80} The mean baseline Hb F% ranged from 4 percent to 12 percent, and during hydroxyurea treatment it ranged from 10 percent to 23 percent. The greatest increase was seen in the study by Voskaridou et al., ⁷⁸ which was a prospective study of Caucasian patients with Hb Sβ⁺ thalassemia and Hb Sβ⁰ thalassemia who were treated with high doses of hydroxyurea, up to 2.5 g/day. These 14 patients had an increase from a mean of 3.6 percent (standard deviation [SD]=2.1) to 23 percent (SD=7.7). The smallest increase in Hb F% was seen in the study from Brazil, which reported outcomes by haplotype. This was a study of 22 patients; the greatest increase was among patients who had a homozygous Bantu haplotype (n=9 patients), from 4 percent to 9 percent (p=0.003). ⁸⁰ As in the pediatric studies, there was a small increase in hemoglobin in most studies. The retrospective study by Loukopoulos reported hematological outcomes very comparable to those seen in the prospective studies. ⁷⁴

The number of pain crises was described in three studies. ^{46, 61, 71} The frequency of crises experienced by the 32 patients who completed the study by Charache et al. decreased from 4 per 6 months (range 0 to 20) to 1.3 per 6 months (range 0 to 9), although this difference was not statistically significant. In a study of Sicilians with Hb Sβ⁺ thalassemia and Hb Sβ⁰ thalassemia, the frequency of crises decreased from a mean of 7 (median of 9) per year to a mean of 1.1 (median 1.8) per year (p<0.0001). ⁷¹ These results included all crises, not just pain crises. In a non-randomized study comparing patients receiving hydroxyurea to those receiving cognitive behavioral therapy, those receiving the drug had fewer pain crises (1.4 per year compared to 4.3 per year, p≤0.05), although this was not a strong design on which to base such an outcome. ⁶¹

Hospitalization rates also decreased for adults treated with hydroxyurea. In the study of Sicilians, the number of hospitalized days in a year declined from 22.4 days to 1.2 days (SD =2.3; p<0.0001). In a retrospective effectiveness study by Ferguson et al., the rates of hospitalization declined from baseline in the group that was treated for longer than 24 months (to 2.1 per year from 3.1 per year, p=0.04). ⁵² For the group treated for fewer than 24 months, however, the investigators did not find a significant difference in hospitalization rates from baseline. In the study comparing hydroxyurea to cognitive behavioral therapy, the patients receiving the drug and those receiving behavioral therapy had similar hospitalization rates (1.1 per year [SD= 2.4] versus 0.9 per year [SD=1.2]). ⁶¹

Several of the studies reported additional efficacy outcomes. In the study by Loukopoulus et al., a mean clinical severity score was calculated based on an arbitrary scale that quantified pain and duration of pain. ⁷⁴ Over 12,018 patient-weeks of treatment, the severity score declined to 81.7 from 1182. In comparing the group receiving hydroxyurea to the group receiving cognitive behavioral therapy, the investigators reported a significant improvement in General Health Perception from the SF36 Health Survey among those receiving the drug. ⁶¹

Ferster et al. enrolled children and young adults in their study, ⁵⁸ while Al-Jam'a et al. enrolled adults and children older than 5 years. ⁷⁰ Ferster et al. demonstrated hematological benefit, with an increase in Hb F% from a mean of 7.3 to 16.7 percent (p<0.01) and an increase in hemoglobin from 8.2 g/dl to 8.8 g/dl (p<0.01). ⁵⁸ Also, none of the patients had a stroke during the study, despite the fact that this was a sick population, with 9 of the 93 having a history of prior stroke and 19 having a history of acute chest syndrome. The rate of hospitalization for those receiving hydroxyurea was 1.1 per patient-year, and the rate of acute chest syndrome was 3.5/100 patient-years. In the study of Saudi Arabian patients, Al-Jam'a et al. demonstrated good hematological outcomes, with an increase in Hb F% from 12.6 percent to 25.7 percent (p<0.05), as well as clinical benefit, with a decrease in hospitalization to a mean of 0.93 per year (p<0.0001) and a decrease in hospital days to 5.1 per year from 34 per year (p<0.05).

Predictors of benefit from hydroxyurea treatment. Many studies have explored predictors of benefit from hydroxyurea (Table 3); one was designed specifically to address this question. ⁷³ Predictors of the fetal hemoglobin response to hydroxyurea were most frequently reported. In the MSH study, Hb F% increased to a greater extent in participants with a lower rate of painful events and, at baseline, reticulocytes greater than 300,000/μl, F reticulocytes greater than 12 percent, absolute neutrophil count (ANC) greater than 7500/μl, and fetal hemoglobin greater than 7.5 percent. Hb F% increased less in men, to a lesser degree in those with the CAR haplotype, with less than 80 percent adherence to therapy, or with fewer than two episodes of hematological toxicity during treatment. ²¹ In a Phase II study from the same group, Hb F was associated with the most recent plasma level of hydroxyurea, as were Hb F and white blood cell counts at baseline. ⁴⁶ Similarly, in children, an increase in Hb F% was associated between higher F reticulocyte counts at baseline⁷⁵ and adherence to therapy, ⁵²as well as increases in hemoglobin and MCV from baseline, decreases in reticulocytes and white blood count from baseline, and lower reticulocyte and white blood counts at the MTD. ⁷³ A pediatric study also identified an association with a higher baseline hemoglobin and greater response. ⁷³ Neither adult nor pediatric studies found that Hb F% was predicted by age; ^{40, 46, 73, 75}; the pediatric studies did not find that gender, hematological toxicities, ⁷³ or haplotype⁷⁵ predicted response.

Clinical responses to hydroxyurea, i.e., a decreased rate of painful episodes, were associated with the baseline rate of painful episodes, decreases in the absolute neutrophil count and absolute reticulocyte count, and increases in MCV. ²¹ Men treated with the drug had greater absolute increases in aerobic power (p<0.05) than did women. ⁴¹ Two pediatric studies from Belgium did not identify any predictors of clinical response; ^{44, 58} however, other observational studies did do so. Hospital admissions were significantly decreased in adults with at least 2 years of hydroxyurea treatment with no interruptions exceeding 2 weeks, when compared to those with a shorter duration of therapy or interruptions. ⁵² In children, an increasing dose of hydroxyurea was associated with a decrease in cerebral blood flow velocity. ⁶² Recurrent stroke in children receiving hydroxyurea for secondary stroke prevention was associated with older age, initiation of hydroxyurea after chronic transfusion had been stopped, and a higher ANC during treatment.⁶⁷

Response to hydroxyurea treatment according to genotype. As described above, studies were generally stratified by age (adult or pediatric); however, other subgroups were less thoroughly investigated. While the majority of patients studied were homozygous for Hb S, most of the studies had some patients with other genotypes. Five studies reported outcomes stratified by genotype (Table 4). The most detailed study was that by Loukopoulous et al., which described all of the hematological outcomes in their small patient sample, stratifying them by sex and genotype. ⁷⁴ Only a single study reported specific outcomes for patients with Hb SC disease; this study demonstrated less improvement with hydroxyurea treatment in this subgroup, and these seven children were less able to tolerate dose increases. ⁸¹ In general, patients with Hb Sβ⁰ thalassemia or Hb Sβ⁺ thalassemia responded to treatment in a manner comparable to that of patients with Hb SS.

Studies of biomarkers suggesting a drug effect. We identified eight studies that compared potential surrogate markers of disease severity or response in patients treated with or without hydroxyurea. ⁸⁵–⁹² These studies were all cross-sectional or cohort studies with at least one comparison group (Appendix C ^*, Evidence Tables 9–11). The studies enrolled patients from North America (5), Europe (2), or Central and Latin America (1). The description of the eligibility criteria was often quite limited, and the majority of studies were of moderate to poor quality. There was only limited information regarding patient characteristics and the starting dose, monitoring, and titration of hydroxyurea.

Four studies included a report of hemoglobin and Hb F levels among groups and reported increases in total and fetal hemoglobin that were comparable to those of the other observational studies in sickle cell disease after treatment with hydroxyurea. ^{86, 87, 90, 92} In three studies, treatment with hydroxyurea was associated with significantly increased levels of nitric oxide metabolites, ^{87, 91} cyclic guanosine monophosphate, ⁸⁷ and nitric oxide synthase and with reduced levels of arginase. ⁸⁶ Another study identified lower levels of endothelin-1, a potent vasoconstrictor, in children treated with hydroxyurea for more than 12 months, when compared to untreated patients. ⁸⁸ These molecules may be biomarkers of abnormal vasoreactivity in sickle cell disease that may contribute to vaso-occlusive complications. Other potential biomarkers of vaso-occlusion were the significant decreases in rigidity and rates of elastic shear in patients with sickle cell disease who had been treated with hydroxyurea, when compared to those in untreated sickle cell disease patients. ⁸⁵ However, these values were still significantly higher than those in controls without sickle cell disease. A small study failed to show differences in tumor necrosis factor-α in adults with Hb SS treated with and without hydroxyurea, ⁸⁹while a study with similar design (comparing adults with Hb SS with and without vaso-occlusive complications and treated with and without hydroxyurea) described a higher percentage of oxyhemoglobin and a lower percentage of reduced hemoglobin in patients on hydroxyurea, with or without vaso-occlusive complications. ⁹¹ The higher percentage of oxyhemoglobin may reflect decreased adhesion and more rapid transit of the cells through the capillary beds. A final pediatric study demonstrated significant decreases in total bilirubin (most likely secondary to decreased hemolysis and release of heme) after treatment with hydroxyurea. ⁹⁰ The baseline level and absolute decrease in bilirubin were strongly correlated with promoter polymorphisms of uridine diphosphoglucuronate glucuronosyltransferase 1A (UGT1A). The lowest levels of bilirubin, both before and during treatment, were seen in children with the UGT1A 6/6 genotype; intermediate levels were seen in the heterozygotes (6/7), and the highest levels were seen in those with the UGT1A 7/7 genotype. These diverse studies of biomarkers suggest possible mechanisms for hydroxyurea's clinical benefits in addition to its ability to increase Hb F and reduce hemoglobin polymerization.

Strength of the evidence regarding the efficacy and effectiveness of hydroxyurea. Based on one RCT in children and many observational studies, some of which were of high quality and most of which were consistent in their findings, we graded the evidence as follows: We concluded that there was a high grade of evidence to support the contention that hydroxyurea raises Hb F in children, and subsequent research is unlikely to change our estimate of that effect, except perhaps in unique populations (such as infants or patients with Hb SC). There was moderate evidence to support the claim that hydroxyurea reduces the frequency of pain crises, and a high grade of evidence to support the contention that treatment reduces the frequency and/or duration of hospitalization in children. There was only a low grade of evidence to support the claim that hydroxyurea reduces neurological events in children and insufficient evidence to allow any conclusions regarding transfusion frequency.

Based on one high-quality RCT in adults and many observational studies, we concluded that there was a high grade of evidence to support the conclusion that hydroxyurea raises Hb F in adults with sickle cell disease, and future research is unlikely to substantially alter these conclusions. There was also high-grade evidence that the drug reduces the frequency of pain crises and that it reduces the frequency and/or duration of hospitalization in adults. There was also high-grade evidence that it reduces transfusions, but only low-grade evidence, from the observational followup of patients in MSH, that it reduces mortality. The evidence base was insufficient to allow us to comment on neurological events in adults (Table 5).

Report by The Center for the Evaluation of Risks to Human Reproduction (CERHR)

The National Toxicology Program (NTP) and the National Institute of Environmental Health Sciences (NIEHS) established the NTP's CERHR in June 1998. The stated purpose of the CERHR was “to provide an unbiased, scientific evaluation of human and experimental evidence of the adverse effects on reproduction and development caused by agents to whom humans may be exposed.” Hydroxyurea was selected for study by the Center in 2006; we briefly review their findings here, as they are relevant to our review of the toxicities of hydroxyurea. CERHR researchers searched databases that included REPROTOX, HSDB, IRIS, DART, PUBMED, and Toxline, through January 2007, to identify articles pertinent to the evaluation of adverse effects on development and reproduction in both humans and animals. The articles were reviewed in advance by an expert panel, which then prepared a document describing the strength of the evidence that hydroxyurea is a reproductive or developmental toxicant. ²⁶ The 13-member expert panel discussed the data and finalized their opinions about the toxicity ofhydroxyurea, identified areas of knowledge gaps, and identified future research priorities. Given this detailed report and acceptable methodology, this EPC opted not to duplicate their effort, and we instead report here a summary of their findings regarding the developmental and reproductive toxicities of hydroxyurea.

Results of Randomized Trials in Sickle Cell Disease

The toxicity data reported from randomized trials of hydroxyurea in sickle cell disease have mainly been limited to short-term toxicities. Only four publications from the MSH study reported toxicities in adults. ^{21, 39, 40, 43} The investigators described lower absolute neutrophil counts (4900/μl vs 6400/μl) in the hydroxyurea group than in the placebo group, but both groups had similar numbers of other adverse events, including thrombocytopenia, thrombocytosis, malignancy, aplastic crisis, aseptic necrosis, lymphadenoathy, and bleeding tendency. The proportion of patients with hair loss, fever, rash and/or nail changes, or gastrointestinal disturbance reported at three or more followup visits was similar for both groups (Table 6; Appendix C ^*, Evidence Table 12). ²¹ The one publication describing long-term followup of the MSH participants described only two malignancies, one in each group. ⁴³ In the study of Belgian children, white blood cell counts decreased from baseline by 3570/μl (p<0.001), but changes in the absolute neutrophil count were not reported. ⁴⁴

Results of Observational Studies in Sickle Cell Disease Including Case Reports

Observational Studies. Although studies that lack comparison arms are not optimal for attributing causality to an observed event, observational studies are useful for describing events in people exposed to a drug outside of a randomized trial (Table 6; Appendix C, Evidence Table 13). We first describe the observational studies reporting cases of leukemia.

In the observational studies we reviewed, three cases of leukemia were reported in people with sickle cell disease who were treated with hydroxyurea. In a study from the French group, a 10-year-old girl was treated with the drug for 18 months. She was admitted with pain and fever; bone marrow examination disclosed ALL, with the Philadelphia chromosome. ^{48, 55} Thus, this group noted a single case among 225 treated patients in this well-characterized cohort whose investigators paid careful attention to losses to followup. The Belgian group reported on a 21-year-old woman who developed acute promyelocytic leukemia after 8 years of hydroxyurea therapy. ⁸² Researchers for the International Association of Sickle Cell Nurses and Physician Assistants collected data about cancer development in 16,613 patients with sickle cell disease. Cancer was diagnosed in 49 patients, including seven cases of leukemia. Three of the 49 had been using hydroxyurea, including one 14-year-old who developed ALL 3 months after initiation of the drug. There were no data on the prevalence of hydroxyurea use among this population of 16,613 people (Appendix C, Evidence Table 13). ⁵³

Another study described a related toxicity, acquired DNA mutations, in hydroxyurea-treated patients with sickle cell disease. ⁵⁷ In this study, two assays were used to quantify acquired somatic DNA mutations in peripheral blood mononuclear cells (PBMCs) after in vivo exposure to hydroxyurea: The HPRT assay measures hypoxanthine phosphoribosyl transferase (HPRT) mutations, while the VDJ assay identifies “illegitimate” T-cell receptor Vγ-Jβ interlocus recombination events. The authors looked at PMBCs from three groups: patients with sickle cell disease who were treated with hydroxyurea, patients with myeloproliferative diseases who had been exposed to hydroxyurea, and normal controls. They found that adults with sickle cell disease and adults with myeloproliferative diseases had a comparable number of mutations when compared to controls; however, children with sickle cell disease having 30 months of hydroxyurea exposure had more VDJ mutations (1.82 +/- 1.2) than did children with 7 months of exposure (1.58 +/- .87) or no exposure at all (1.06 +/- 0.45, p=0.04). HPRT mutations were similar in the two exposed groups. The authors interpreted this result as a slight increase in recombination events and suggested that this increase does not directly portend the development of leukemia. Similarly, 26 adult patients with sickle cell disease who had been exposed to hydroxyurea for at least 5 years at the MTD were found to have no increase in illegitimate VDJ rearrangements, when assessed using similar methodology. ⁸¹ Karotypic analysis in one study revealed no difference in the percentage of abnormal chromosomes before and after treatment with hydroxyurea. ⁴⁶ In a similar, although smaller study, there was no significant difference in chromosomal aberrations (p>0.05) pre- and post-treatment and no difference in the mitotic index (Appendix C ^*, Evidence Table 13). ⁴⁷

The more frequent toxicities are described in Appendix C, Evidence Table 13. Toxicities were described in 22 of the 35 observational studies; 8 of these were studies designed to primarily report toxicities from hydroxyurea. ^{47, 48, 53, 55, 57, 64, 66, 68} Additional articles described moderate decreases in platelet counts on therapy; this observation is not included in the table, since this is an known effect of the drug and is generally not considered to be an adverse event. Rare deaths were reported. In one study with 455 patient-years of followup, one child died of pneumococcal sepsis despite a normal absolute neutrophil count, and another child died from an acute transfusion reaction. ⁷⁸ Neither death was thought to be related to hydroxyurea. There were single deaths reported in five other studies^{48, 60, 72, 77, 82}; all of the deaths were from expected complications of sickle cell disease, and none were thought to be due to myelosuppression (Appendix C, Evidence Table 13).

Neutropenia was a frequently reported adverse event. In the HUSOFT study, 17 of the 28 children had an absolute neutrophil count of less than 1500/μl, including 6 with an absolute neutrophil count of less than 500/μl. ⁶⁰ In the extension of the HUSOFT study, there were 21 episodes of neutropenia in 10 children in the third treatment year and 21 episodes in 9 patients in the fourth year. ⁷² In the HUG-KIDS study, 56 of 84 patients had an absolute neutrophil count of less than 2000/μl. Thrombocytopenia was less frequently reported.

Leg ulcers were only reported as occurring in three studies and were infrequent. ^{55, 66, 74} Prior leg ulcer was associated with the development of leg ulcer during hydroxyurea treatment in the study that reported the highest incidence of ulcers (5 of 17 treated patients). ⁶⁶ Rash and nail changes were moderately common.

The other even rarer toxicities are described in Appendix C, Evidence Table 14. The number of individuals experiencing these toxicities was low in all studies. No study reported any secondary malignancies.

Case Reports. We identified 19 published case reports about toxicities associated with hydroxyurea use in patients with sickle cell disease (Appendix C, Evidence Table 15). Two of these reports described the same Greek child who developed Hodgkin's lymphoma. ^{96, 97} The 18 unique case reports included four reports of low sperm count or decreased sperm motility, two cases of avascular necrosis, two cases of skin hyperpigmentation, and one case each of leg ulcer, cytopenia, splenomegaly, cryptosporidium infection, intracerebral hemorrhage, acute myocardial infarction, and the case of Hodgkin's lymphoma previously mentioned. All of these toxicities were described in adults, except for the case of Hodgkin's.

In addition, leukemia was reported in three young women with sickle cell anemia who had been treated with hydroxyurea. We describe these three cases in detail here: One was the 21-year-old woman mentioned above who was treated as part of the Belgian Registry of Sickle Cell Disease. ⁹⁸ She had been taking hydroxyurea for 8 years but stopped for 2 years while pregnant and nursing. She resumed hydroxyurea therapy, and 8 months later was diagnosed with AML (M3v). Another report was of a 25-year-old Saudi Arabian woman who was treated with hydroxyurea for 2 years with good response. She was subsequently diagnosed with AML ( FAB M1); cytogenetic studies revealed no abnormal clone. ⁹⁹ Interestingly, this patient had splenomegaly, without explanation, at the time that she began hydroxyurea therapy and also had hepatitis C infection. The final case report described a 42-year-old woman with Hb SS who was treated for 6 years with hydroxyurea. She was diagnosed withAML; she had no cytogenetic analysis. ¹⁰⁰ We are aware of one other case report of leukemia in a patient with sickle cell anemia treated with hydroxyurea. This case was reported in abstract form and described a 27-year-old woman who developed an acute non-lymphocytic leukemia after 8 years of hydroxyurea therapy. Her bone marrow aspirate suggested that the leukemia developed in the setting of myelodysplasia. ¹⁰¹

Each of these toxicities had only Level 3 evidence for causality (at least one “possible” report of an adverse event, but no “certain” or “probable” case report), with the exception of cytopenia, which was considered to have Level 2 evidence (at least one “probable” report of an adverse event, but no “certain” report) because there was one case report that demonstrated probable causality. Reports of leukemia are difficult to score with the WHO causality scale because there is no possibility for regression of disease with removal of the putative causal agent (leukemia cannot spontaneously remit), so the case reports of leukemia cannot be described as showing probable or certain causality.

Results of Studies of Other Diseases

Given that the number of patients with sickle cell disease who were treated for long durations with hydroxyurea is few, we opted to review toxicities in patients with diseases other than sickle cell disease in order to gather additional evidence regarding the potential toxicities of this drug.

Randomized trials and large observational studies. We found 39 publications (20 randomized and 19 observational studies) that examined the toxicity of hydroxyurea in diseases other than sickle cell disease. (Appendix C ^*, Evidence Tables 16, 17, 18, 19, 20, and 21). Included among these were studies of the addition of hydroxyurea to other often-used therapies, enabling us to describe the additive toxicity attributable to hydroxyurea. These publications included studies examining the effects of adding hydroxyurea to standard antiretroviral therapy in patients with HIV/AIDS and the addition of hydroxyurea to interferon, as these drugs may be used in patients with sickle cell disease and comorbid illnesses. There were eight publications about hydroxyurea use in patients with HIV, nine in patients with CML, five in patients with a variety of myeloproliferative diseases, two in patients with solid tumors, eight in patients with polycythemia vera, and six in patients with essential thrombocythemia. We present the results for the RCTs, followed by those for the observational studies.

Description of the quality of the studies. The quality of the RCTs was evaluated using the Jadad criteria. The study scores ranged from 1^{102, 103} to 4¹⁰⁴ (Appendix C, Evidence Table 22), with most of the studies scoring a 2 or 3. The studies were all randomized, but most did not describe the method of randomization, and they also lost points for not describing the blinding of the participants. A majority of the studies also provided at least some information about the subjects that were withdrawn from the study.

The observational studies were evaluated with our 16-point scale for assessing the quality of these studies. These studies received between 28 percent ¹⁰⁵ and 73 percent¹⁰⁶ of the available points (Appendix C, Evidence Table 23). Thus, none of these studies reached our cutoff of more than 80 percent, which we judged to indicate high quality. Only one of these studies reported on adherence. ¹⁰⁷ The scores were also diminished because most of them did not describe the subjects that were lost to followup.

Characteristics of Studies Addressing Barriers to the Use of Therapies for Sickle Cell Disease

Of the studies that addressed this key question, 15 employed a cross-sectional study design to test an association between a patient, provider, or system factor and the use of therapy; ¹³⁸–¹⁵² 18 described patient and/or provider-reported barriers to therapy; ^{93, 153, 153}–¹⁶⁸ and 3 used both methods. ^{155, 169, 170} Studies employing cross-sectional designs to assess the association between a patient, provider, or system factor and the use of therapies are shown in Appendix C ^*, Evidence Table 25, and studies describing patient and/or provider-reported barriers to therapy are shown in Appendix C, Evidence Table 26.

Cross-sectional studies testing associations. Of the 18 studies using a cross-sectional design (Appendix C, Evidence Table 25), only 1 addressed barriers to the use of therapies to increase Hb F, ¹³⁸ 8 addressed barriers to the use of established therapies for management of sickle cell disease, ¹³⁹–^{144, 169, 170} 3 addressed barriers to the use of appropriate pain medication during vaso-occlusive crises, ^{145, 146, 155} and 6 addressed barriers to the use of routine, scheduled care for sickle cell disease. ¹⁴⁷–¹⁵²

The one study that dealt with therapies to increase Hb F examined factors associated with patient (or parent) decisions to initiate therapy. ¹³⁸ Of the eight studies that addressed barriers to the use of established therapies, one focused on patient adherence to chelation therapy, ¹³⁹ while the remainder focused on patient adherence to antibiotic prophylaxis. ¹⁴⁰–^{144, 169, 170} Of the three studies that that dealt with the use of appropriate pain medications during vaso-occlusive crises, two addressed providers' provision of pain medications to patients, ^{145, 155} and one addressed patients' use of pain medications. ¹⁴⁶ Of the six studies that dealt with routine, scheduled care, three directly addressed the use of routine health services, ^{147, 148, 150} one addressed the transition to adult care, ¹⁴⁹ one addressed appointment-keeping, ¹⁵² and one addressed general adherence. ¹⁵¹

Two-thirds of the cross-sectional studies were published in the past 5 years (2002-2007), while the remainder were published in the previous decade (1992-2001). Three of the 18 cross-sectional studies focused on health professionals as study subjects, ^{140 145, 155}while the remainder studied patients. Adult patients with sickle cell disease were the targeted patient population of interest in only 1 of the 18 cross-sectional studies. ¹⁵² In four of the cross-sectional studies, children and adults were the targeted patient population of interest.^{147, 149, 150, 155} The remainder focused on children (or parents/caregivers). The majority were conducted in the United States, while one was conducted in Saudi Arabia¹⁷⁰and five did not specify a location.

Descriptive studies of reported barriers. Of the 20 descriptive studies of patient- and provider-reported barriers to the use of therapies for sickle cell disease (Appendix C, Evidence Table 26), 1 addressed barriers to the receipt of treatment to increase Hb F, ⁹³ 2 addressed barriers to patient adherence to established therapies for disease management, ^{169, 170} 13 addressed barriers to the receipt of pain medications, ¹⁵³–^{155, 159}–¹⁶⁸ 1 addressed barriers to bone marrow transplantation, ¹⁵⁸ and 3 addressed barriers to generic healthcare quality. ^{156, 157, 171}

About half of the descriptive studies were published in the past decade (1998-2007), with the remainder published in the preceding decade (1988-1997). Most (n=13) descriptive studies occurred in the United States; however, 6 were conducted in the United Kingdom, ^{154, 159, 161, 163, 166, 168} and one was conducted in Saudi Arabia. ¹⁷⁰

Of the 20 descriptive studies, 9 used primarily quantitative descriptive methods (e.g., questionnaires), ^{93, 153}–^{158, 169, 170} 10 used primarily qualitative methods (e.g., focus groups and in-depth individual interviews), ^{153, 159}–¹⁶⁷ and 1 used mixed quantitative and qualitative methods. ¹⁶⁸ The majority of the descriptive studies included patients as a study population, ^{154, 158, 159, 161, 162, 164}–^{167, 169, 170} while others included providers, ^{93, 155}–¹⁵⁸ and some included both patients and providers. ^{160, 163, 168, 171} Of the 15 descriptive studies that included patients as the study population, 12 were focused on adults, ^{154, 159}–^{168, 171} and 3 were focused on children. ^{158, 169, 170}

Results of Studies Addressing Barriers to the Use of Therapies for Sickle Cell Disease

Results of cross-sectional studies testing associations. The results of the cross-sectional studies are summarized in Table 8. Each of the potential barriers and facilitators below was identified in only one study. The factors in each category that were examined but not associated with use of therapies are included in Table 8 but not detailed below.

The one study that addressed barriers to the use of therapies to increase Hb F (specifically, hydroxyurea) found that the perceived efficacy and perceived safety of hydroxyurea had the largest influence on patients' (or parents') choice of hydroxyurea therapy over other therapies. ¹³⁸

The eight studies that addressed potential barriers to the use of established therapies for disease management found two potential patient-related barriers (family stress and having more children in the home), and one potential system-related barrier (being seen in an academic medical center). ¹³⁹–^{144, 169, 170} These eight studies also identified 11 potential patient-related facilitators of the use of established therapies for disease management (private insurance, sharing of responsibilities between parent and child, more hospital visits, more adults in the home, having a car, no child prior history of transfusion, younger patient age, more caregiver knowledge, greater intent to adhere, greater perceived benefits, and family employment) and two potential provider-related facilitators (provider female gender and pediatric specialty).

The three studies that addressed barriers to the use of appropriate pain medication during vaso-occlusive crisis found one patient-related barrier (an increased number of hospital visits was associated with less optimal pain management) and one provider-related barrier (negative provider attitudes). ^{145, 146, 155} These studies also found one potential patient-related facilitator (dispositional optimism being associated with better patient use of pain medications) and two potential provider-related facilitators (provider female sex and fewer years in practice).

The six studies that addressed barriers to use of routine, scheduled care for sickle cell disease¹⁴⁷–¹⁵² found one potential patient-related barrier (greater community socio-economic distress) and eight potential patient-related facilitators (greater parental knowledge, rural geographic region, higher self-efficacy, female patient sex, higher family problem-solving effort, higher family income, greater illness-related stress, and greater social support). Of note, the studies that found rural location to be a potential facilitator controlled for distance to the clinic, which may have eliminated the typical reason for decreased access by rural patients.

Results of descriptive studies of reported barriers. The results of studies employing descriptive methodologies to identify patient and provider-reported barriers to the use of therapies are summarized in Table 9.

The one study that explored barriers to the use of treatments to increase Hb F (specifically, hydroxyurea) for patients with sickle cell disease found that providers reported the barriers to be patients' concerns about side effects and the providers' own concerns about the use of hydroxyurea in older patients, about patient compliance, about a lack of contraception, about side effects and carcinogenic potential, doubts about effectiveness, and concern about the costs to patients. ⁹³

The two studies that addressed barriers to the use of established therapies for disease management both examined patient (caregiver)-reported reasons for missing doses of prophylactic antibiotic medication and found that caregivers reported missing doses as a result of forgetting, being too busy, running out of medication, having the child fall asleep, and the child not liking the taste of the medication. ^{169, 170}

All of the 13 studies that addressed barriers to the receipt of pain medications during vaso-occlusive crises found that both patients and providers reported that some type of negative provider attitude affected the quality of the pain management for patients with vaso-occlusive crisis. ¹⁵³–^{155, 159}–¹⁶⁸ Other barriers identified by patients and providers included poor provider knowledge of sickle cell disease (mentioned in five studies), lack of time (mentioned in two studies), inadequate pain assessment tools (mentioned in two studies), and race (mentioned in one study).

The one study that addressed barriers to bone marrow transplantation found that providers from bone marrow transplant centers reported that the major barriers to bone marrow transplantation for patients with sickle cell disease were lack of a donor, lack of psychosocial or financial support, a history of patient noncompliance, parental refusal, and physician refusal. ¹⁵⁸

The three studies that addressed barriers to general healthcare quality found that patients and providers reported that three patient-related factors (patient race, older patient age, and patient male sex) may affect the quality of care provided to patients with sickle cell disease. ^{156, 157, 171}

Strength of the evidence of the existence of barriers to the use of therapies in sickle cell disease. There was insufficient evidence to allow us to identify barriers to the use of hydroxyurea. Regarding barriers to the use of established therapies for sickle cell disease, four items were identified as either barriers, facilitators, or neither in more than two studies and thus were eligible for evidence grading. These were patient/family knowledge, number of hospital visits, patient age, and patient sex.

We concluded that the evidence that sex is not a barrier to the use of therapies was of a moderate grade. Largely due to the relative paucity of studies and their inconsistency, we concluded that there was only low-grade evidence that patient/family knowledge, the number of hospital visits, and patient age are barriers. The evidence for the remaining barriers to the use of established therapies was insufficient to allow us to draw any conclusions.

Regarding barriers to pain management, we identified two factors that were identified as a barrier in more than two studies and were thus eligible for evidence grading. These were negative provider attitudes and poor provider knowledge. Because of the quantity and consistency of these findings, we concluded that the evidence was high-grade that negative provider attitudes are barriers and moderate-grade that poor provider knowledge is a barrier to the use of pain medications for patients with sickle cell disease. The evidence for the remaining barriers to pain management was insufficient to allow us to draw any conclusions. There was insufficient evidence to allow us to identify barriers to the use of routine health services and bone marrow transplantation for sickle cell disease (Table 5).

Characteristics of Studies Addressing Interventions to Overcome Barriers to the Appropriate Use of Therapies

Thirteen studies addressed interventions to increase the appropriate use of therapies. ¹⁷²–¹⁸⁴ None of these studies focused on interventions to increase the appropriate use of hydroxyurea. Nine focused on provider interventions to increase the appropriate provision of pain medications for patients with vaso-occlusive crisis, ¹⁷²–¹⁸⁰ three focused on patient interventions to improve adherence to therapies (desferoxime, ¹⁸¹ antibiotics, ¹⁸² and general adherence to health-promoting activities¹⁸³), and one focused on a patient intervention to improve the receipt of routine, scheduled health care for sickle cell disease (Appendix C, Evidence Tables 27 and 28). ¹⁸⁴

The majority of these interventions were assessed using observational (e.g., pre/post) study designs, but there were two RCTs^{182, 183} and three studies that had a concurrent control group. ^{173, 175, 176} The majority of the studies were conducted in the United States, with two taking place in the United Kingdom^{177 179} and one in Canada. ¹⁸²

The majority of the provider interventions to improve pain management involved clinical protocols/pathways, ¹⁷²–^{175, 178, 180} while one primarily involved audit and feedback, ¹⁷⁷ and two involved changing the structure of care through the use of a day hospital¹⁷⁶ or a fast-track admission process. ¹⁷⁹ In addition, one of the clinical protocol/pathway interventions included staff sensitivity training. ¹⁷⁴

The effect of interventions to improve pain management was measured directly in terms of its impact on pain management quality, as assessed by medical record review in six studies^{173, 175}–^{177, 179, 180} and patient ratings in three studies; ^{174, 179, 180} the effect of such interventions was also assessed indirectly through healthcare utilization in five studies^{175 172, 174, 176, 178} and healthcare costs in two studies. ^{174, 175} No study examined the impact of an intervention directly on patient-reported levels of pain.

Results of Studies Addressing Interventions to Overcome Barriers to the Appropriate Use of Therapies

A summary of the results of studies addressing interventions to increase the appropriate use of therapies is provided in Table 10 (see also Appendix C ^*, Evidence Tables 27 and 28).

Four of the studies that measured the impact of an intervention to improve the quality of pain management during vaso-occlusive crisis showed improvement in one or more direct outcomes, ^{173, 179, 180, 185} while the remaining five studies showed potential improvement either through the suggestion of an improvement on a direct outcome (without a statistical test) or a statistically significant improvement in one or more indirect outcomes.

Two of the three studies that focused on patient interventions to improve adherence to therapies showed no effect of the intervention on patient adherence to desferoxime¹⁸¹ or to antibiotic prophylactic therapy. ¹⁸² One of the studies that focused on patient interventions to improve adherence to therapies showed no increase in health-promoting activities as a result of the intervention but did show some improvements in child health-related quality of life and child-parent relationships. ¹⁸³

The one study that evaluated a patient intervention to improve receipt of routine, scheduled health care for sickle cell disease demonstrated a substantial and significant reduction in the percent of patients who had not attended clinic over the past 2 years. ¹⁸⁴

The strength of the evidence addressing interventions to overcome barriers to the use of therapies. The evidence was insufficient to allow us to identify interventions to overcome barriers to the use of hydroxyurea and bone marrow transplantation.

None of the three studies testing interventions to improve patient adherence to established therapies for chronic disease management showed any effect on patient adherence. However, due to the small sample sizes and diverse outcome measures, we concluded that there was only low-grade evidence that interventions cannot improve patient adherence.

We concluded that there was moderate evidence that interventions can overcome barriers to the use of pain medications and moderate evidence to support the contention that interventions can overcome barriers to the receipt of routine, scheduled healthcare for patients with sickle cell disease (Table 5).

Efficacy and Effectiveness of Hydroxyurea in Children

A single, small RCT investigated the efficacy of hydroxyurea in children. In this Belgian study, the rate of hospitalization and number of days hospitalized per year were significantly lower in the hydroxyurea group than in the placebo group. ⁴⁴ The small size of this study and the short duration of treatment with the drug (6 months) did not provide adequate data to permit assessment of the long-term responses to hydroxyurea. The results of this trial are supported by data from 20 observational studies in children. Interpretation of many of these observational studies is complicated by their incomplete description of losses to followup. The HUSOFT was the only study of very young children, with a mean age of 1.3 years; ⁶⁰ this Phase I/II study also had a published followup study. ⁷² Hb F% was reported as an outcome in 16 of these observational studies. In the studies that reported Hb F% before and during treatment with hydroxyurea, the Hb F% increased substantially while patients were on treatment, with results comparable to those reported in the RCT. The mean pre-treatment Hb F% ranged from 5 to 10 percent, and the post-treatment values were in the range of 15 to 20 percent. In the study of infants, hydroxyurea therapy prevented the expected decline in Hb F% that is usually seen in this age group. Hemoglobin concentration increased only modestly (roughly 1 gm/dl) but significantly across studies.

The frequency of pain crises decreased in three of the five studies in which this variable was assessed; in one study without a comparison group, it was unclear how the rate differed from an untreated group. Hospitalization rates declined in all studies in which this was assessed. Two studies reported TCD velocities and demonstrated decreased velocity while the children were being treated. ^{62, 63} One study included a control group that showed an increase in velocity over the treatment period. ⁶² One other study demonstrated rates of recurrent stroke in patients receiving hydroxyurea to be comparable to the rate typically seen in children on chronic transfusion therapy. ⁶⁷

Based on one RCT in children and on many observational studies, some of which were of high quality and most of which were consistent in their findings, we concluded that there was a high grade of evidence to support the claim that hydroxyurea raises Hb F in children. There was moderate evidence to support the contention that hydroxyurea reduces the frequency of pain crises, and a high grade of evidence that it reduces the frequency and/or duration of hospitalization in children. There was a low grade of evidence to support that claim that hydroxyurea reduces neurological events in children, and insufficient evidence to allow us to draw any conclusions regarding transfusion frequency.

Efficacy and Effectiveness of Hydroxyurea in Adults

Only one randomized trial tested efficacy in adults with sickle cell disease, the MSH Study. ³⁹ Six additional analyses based on this trial or on followup studies have also been published. ^{21, 22, 40}–⁴³ In the MSH, 18 percent of patients had permanent or complete cessation of the study medication during the trial. The significant hematological effects of hydroxyurea after 2 years (when compared to the placebo arm) included a small increase in total hemoglobin of 0.6 g/dl and a moderate absolute increase in fetal hemoglobin of 3.2 percent. ⁴⁴ The results of the MSH study included significant differences in several critical clinical outcomes between the hydroxyurea arm and placebo. The median number of painful crises was 44 percent lower in the hydroxyurea arm, and the time to the first painful crisis was 3 months, as compared to 1.5 months for those in the placebo arm. There were fewer episodes of acute chest syndrome and transfusions, but no significant differences in deaths, strokes/chronic transfusion, or hepatic sequestration. The MSH study also included an evaluation of costs and quality of life, both of which favored the use of hydroxyurea.

The results of the 12 observational studies enrolling only adults supported the findings in the MSH study. In all six studies of adults that reported hematological outcomes, Hb F% was significantly higher for those receiving hydroxyurea. ^{45 46, 71, 74, 78, 80} The number of pain crises was given in three studies, all of which demonstrated a significant decline in frequency with drug treatment. ^{46, 61, 71} Similarly, hospitalization rates decreased for adults treated with the drug. In a group of patients treated for fewer than 24 months, however, the investigators did not find a significant difference in hospitalization rates from baseline, ⁵² although patients who discontinue before 24 months may represent a different population than those who are able to tolerate a longer duration of therapy.

There was only one study that assessed responses in patients with Hb SC, although many studies included at least some patients with Hb Sβ^0/+ thalassemia. The results appeared to be comparable for patients with Hb SS and Hb Sβ^0/+ thalassemia. Sex and age had little influence on outcomes.

Several interesting studies have described potential biomarkers of the response to hydroxyurea. One study identified significantly decreased rigidity and rates of elastic shear in patients with sickle cell disease treated with hydroxyurea, when compared to untreated sickle cell disease patients, but the values were still significantly higher than those for controls without sickle cell disease. ⁸⁵ Another study described lower rates of arginase activity in those treated with hydroxyurea, but again the rates were lower in controls without sickle cell disease. Similarly, increased nitric oxide metabolites and cyclic GMP were reported in patients treated with hydroxyurea. ⁸⁷ Levels of endothelin-1 were significantly lower in children with Hb SS on hydroxyurea at steady-state when compared to an untreated group, ⁸⁸ but the levels of tumor necrosis factor-1 were similar for untreated patients with Hb SS and those receiving a single dose of hydroxyurea. ⁸⁹ These studies suggest that other mechanisms may contribute to the benefit resulting from hydroxyurea in addition to the anti-sickling effect produced by an increase in Hb F concentration in red blood cells.

Based on one high-quality RCT in adults and many observational studies, we concluded that there was a high grade of evidence to support the claim that hydroxyurea raises fetal hemoglobin in adults with sickle cell disease and that future research is unlikely to substantially alter these conclusions. There was also high-grade evidence that hydroxyurea reduces the frequency of pain crises and that the drug reduces the frequency and/or duration of hospitalization in adults. There was also high-grade evidence that hydroxyurea reduces transfusions, but only a low grade of evidence that it reduces mortality. The evidence base was insufficient to allow us to comment on neurological events in adults.

Toxicities of Hydroxyurea in Children and Adults

Our assessment of the strength of the evidence regarding the toxicity of hydroxyurea when used in children came largely from our review of the report by the panel of experts assembled by the CERHR. This panel reviewed articles, published through January, 2007, that were pertinent to the evaluation of adverse effects of hydroxyurea on development and reproduction in both humans and animals.²⁶

The panel concluded that treatment of children aged 5 to15 years with hydroxyurea does not cause a growth delay. They felt that there were insufficient data to allow them to evaluate the effects of the drug on pubertal development. The expert panel also concluded that there were insufficient data regarding the effects on subsequent generations following exposure of germ cells to hydroxyurea, including exposure during fetal life, infancy, childhood, and adolescence. The CERHR report did not describe any studies on the long-term health effects, including carcinogenicity, of childhood exposure to hydroxyurea; we were also unable to find any such studies. The expert panel had concerns about the adverse effect of hydroxyurea on spermatogenesis in men receiving the drug at therapeutic doses, and we also identified relevant case reports in both patients with sickle cell disease and patients with other illnesses who had been treated with hydroxyurea. The CERHR report concluded that the use of the drug in pregnancy was not commonly associated with adverse perinatal outcomes and that there were no data on long-term outcomes in children who were exposed in utero.

Three cases of leukemia were described in the observational studies we reviewed that dealt with patients with sickle cell disease who had been treated with hydroxyurea; ^{53, 55, 82} we also identified another three case reports of hydroxyurea-treated patients with sickle cell disease who developed leukemia, and one report of a child developing Hodgkin's lymphoma. Without knowing how many patients had been treated with hydroxyurea, it was impossible to calculate whether this rate of leukemia was higher than the baseline rate for young adults with sickle cell disease. In one study, a higher rate of VDJ recombination events was reported in patients treated for 30 months than in those treated for 7 months; ⁵⁷ another study found this not to be the case. ⁸¹ Two studies found no increase in chromosomal abnormalities in the treated patients. ^{46, 47} Other toxicities associated with hydroxyurea in patients with sickle cell disease that appeared likely to be causally related to its use were neutropenia, skin rashes, and nail changes. The other toxicities reported were rare.

We reviewed toxicity reports from patients with other diseases who were treated with hydroxyurea and found additional toxicities, including a high number of leg ulcers and skin cancers. Of the RCTs enrolling patients with polycythemia vera, MPD, HIV, and essential thrombocytosis, none demonstrated a greater number of cases of leukemia in the group treated with hydroxyurea as a single agent. This parameter could not be assessed in the trials enrolling patients with CML, since acute leukemia was evaluated as an outcome rather than as a medication-related toxicity. To further address this potential relationship between hydroxyurea and the risk of malignancy, we reviewed a case series consisting of 26 patients with AML who had a unique t(3;21) chromosomal translocation. ¹¹⁷ Among these 26 patients were 15 with CML who had been treated with hydroxyurea. We found no other reports describing an association between this translocation and the use of this drug.

In our analysis of patients with diseases other than sickle cell disease, we also found evidence from case reports that hydroxyurea could cause fever, hepatitis, and interstitial pneumonitis. The hematological toxicities (cytopenias) seen were often intensified when patients were receiving other myelotoxic drugs such as antiretroviral therapies.

We concluded, based on our review of toxicities in both patients with sickle cell disease and patients with other diseases, that there was low-grade evidence that hydroxyurea treatment in adults with sickle cell disease is not associated with leukemia. High-grade evidence supported the claim that hydroxyurea is not associated with leg ulcers in patients with sickle cell disease, although high-grade evidence also indicated that hydroxyurea is associated with leg ulcers in patients with other conditions. We hypothesized that the improvement in rheology offsets any increase in leg ulcer risk associated with the drug.

The evidence was insufficient with regard to sickle cell disease to allow us to determine whether hydroxyurea contributes to skin neoplasms, although high-grade evidence in other conditions suggested that it does. The other populations studied were largely light-skinned populations, so we were not surprised that the skin cancer risk was notably different across populations. Likewise, there was insufficient evidence to indicate whether hydroxyurea is associated with secondary malignancies in adults with sickle cell disease, and the evidence in other diseases was only low-grade.

Barriers to the Use of Hydroxyurea and Other Treatments for Managing Sickle Cell Disease

We anticipated finding few data that specifically addressed barriers to the use of hydroxyurea. Indeed, only two studies explored barriers to use of this drug, ^{93, 138} and no study tested interventions to overcome such barriers. Given the scarcity of the data, we sought information on barriers to the use of other therapies for the treatment of sickle cell disease, including the receipt of routine, scheduled care; adherence to medications; and receipt of therapies, including pain control and prescriptions.

As expected, we found insufficient evidence to allow us to directly identify barriers to the use of hydroxyurea. Of the 18 cross-sectional studies we reviewed that tested whether hypothesized barriers affected the use of therapies, only one investigated barriers to hydroxyurea use. ¹³⁸ This study found that the perceived efficacy and perceived safety of the drug had the largest influence on patients' (or parents') choice of hydroxyurea therapy over other therapies.

The one descriptive study that explored barriers to the use of hydroxyurea in patients with sickle cell disease found that providers reported such barriers to be patients' concerns about side effects, as well as the providers' own concerns about the use of the drug in older patients, about patient compliance, about a lack of contraception, about side effects and carcinogenic potential, doubts about effectiveness, and concern about the costs to patients. ⁹³

Largely because of the relative paucity of relevant studies and their inconsistency, we concluded that there was only low-grade evidence that patient or family knowledge, the number of hospital visits, and patient age are barriers. We concluded that the evidence was of a moderate grade that sex is not a barrier to use of therapies. The evidence about the remaining barriers to the use of established therapies was insufficient to yield any firm conclusions.

Regarding barriers to adequate pain management that were identified in both cross-sectional studies and descriptive studies, we identified two barriers that were cited in more than two studies: negative provider attitudes and poor provider knowledge. Because of the quantity and consistency of these findings, we concluded that the evidence was high-grade that negative provider attitudes are barriers and moderate-grade that poor provider knowledge is a barrier to the use of pain medications for patients with sickle cell disease. The evidence for the remaining barriers to pain management was insufficient to allow us to draw any conclusions.

None of the three studies testing interventions to improve patient adherence to established therapies for chronic disease management showed any effect on patient adherence. ¹⁸¹–¹⁸³ However, because of the small sample sizes and diverse outcome measures in these studies, we concluded that there was only low-grade evidence that interventions cannot improve patient adherence. We concluded that there was moderate evidence that interventions can overcome barriers to the use of pain medications, and moderate evidence supported the possibility that interventions can overcome barriers to the receipt of routine, scheduled healthcare for patients with sickle cell disease.

We found it informative that when researchers chose barriers to investigate, they most often studied patient-related barriers. When patients were asked to identify barriers to use of therapies, they most often identified provider-related barriers. The barrier to pain management that was most often identified by patients and providers was negative provider attitudes. However, only one of the nine pain management intervention studies addressed this issue directly through provider sensitivity training. Although the barriers related to the use of pain medications during vaso-occlusive crisis may not seem immediately relevant to the use of hydroxyurea, we concluded that it is likely that patients who have bad experiences when seeking healthcare may lose trust in the healthcare system and be less willing to take recommended medications, including hydroxyurea.