Hydroxyurea dosing in sickle cell anemia: a systematic review, dose-response meta-analysis, and population-level benefit-risk assessment

Hydroxyurea is an established disease-modifying treatment for sickle cell anemia (SCA), but dose-response relations for clinical benefit and hematologic toxicity remain uncertain across study populations. We examined associations between hydroxyurea dose, recurrent vaso-occlusive crises (VOC), and cytopenias and used these estimates to evaluate a prespecified population dosing framework. We conducted a systematic review and one-stage multilevel dose-response meta-analysis of randomized trials, nonrandomized comparative studies, and longitudinal cohorts including children or adults with sickle cell disease (SCD). The main quantitative synthesis was based principally on SCA, defined as HbSS or HbSβ0-thalassemia. Hydroxyurea dose was expressed in mg/kg/day, with achieved or maintenance dose preferred when available. Recurrent VOC rate was the primary benefit outcome, and annual participant risks of severe or treatment-limiting neutropenia and thrombocytopenia were the primary safety outcomes. Emax models were used for the primary analyses, with linear and dose-independent models examined in sensitivity analyses. The decision rule required predicted annual neutropenia and thrombocytopenia risks of no more than 5% and 3%, respectively, with at least an 80% simulation-derived probability that both criteria were met. Eligible doses also had to retain near-optimal VOC control. The final reference dose was selected from conventional 2.5-mg/kg/day increments. Certainty of evidence was assessed using GRADE. Thirty-nine studies or linked study programmes met the eligibility criteria. Common empirical support for the primary benefit-risk analysis ranged from 20 to 28 mg/kg/day. The primary model predicted recurrent VOC rates of 39.8 events per 100 person-years at 20 mg/kg/day and 32.5 at 28 mg/kg/day. The Emax ED50 reached the upper limit of the prespecified search grid, and a linear model had a slightly lower AICc (ΔAICc = 0.86), so the available data did not identify an efficacy plateau within this dose range. Predicted annual severe or treatment-limiting neutropenia risk increased from 1.98% at 20 mg/kg/day to 3.74% at 21.5 mg/kg/day and 5.63% at 22.5 mg/kg/day. Thrombocytopenia remained uncommon and showed little evidence of a dose-dependent increase. The probability that both hematologic safety criteria were met was 87.1% at 20 mg/kg/day, 85.6% at 21 mg/kg/day, and 77.8% at 21.5 mg/kg/day. Because the safety evidence was sparse and the exact joint classification depended partly on the relation between the two safety models, the low-20-mg/kg/day range was interpreted as an approximate safety transition rather than a precise toxicity threshold. The lowest modeled recurrent-VOC rate within empirical support occurred at 28 mg/kg/day, and the unconstrained composite-benefit maximum occurred at 25.5 mg/kg/day. Under the primary model, 21 mg/kg/day was the highest dose meeting the analytical safety rule, and 20 mg/kg/day was selected as the conventional population reference dose. Certainty in the dose-response evidence was very low for recurrent VOC, neutropenia, and thrombocytopenia. Recurrent-VOC rates continued to decline above 20 mg/kg/day, and the available evidence did not show an efficacy plateau between 20 and 28 mg/kg/day. Dose selection was limited instead by increasing neutropenia risk and uncertainty in the low-20-mg/kg/day range. Under the prespecified benefit-risk framework, 20 mg/kg/day was selected as the conventional population reference dose. This dose is not a maximum tolerated dose or a ceiling for treatment. Higher doses may be appropriate for individual patients when further clinical benefit is needed and hematologic tolerance, adherence, and laboratory monitoring allow dose escalation. The review was prospectively registered with PROSPERO ( https://www.crd.york.ac.uk/PROSPERO/view/CRD420251238763 ).