A meta‐analysis of toxicities related to hydroxycarbamide dosing strategies

Abstract Due to fear of short‐term toxicities, there is nonconsensus of hydroxycarbamide dosing strategy (escalated vs fixed‐dosing methods), which contributes to its suboptimal use. We performed a meta‐analysis to summarize the incidence rates of toxicities associated with both dosing methods. Summarized incidence rates could not be statistically compared between dosing methods due to sparse data. Summarized neutropenia and thrombocytopenia incidence rates were slightly higher when using escalated dosing than with fixed. Summarized reticulocytopenia was comparable. Summarized hepatic and renal toxicities’ incidence rates were slightly higher when using fixed doses than with escalated. We recommend diligent and transparent reporting of toxicities.


INTRODUCTION
Hydroxycarbamide (also known as hydroxyurea [HC]) is an approved treatment for adults and children (>2 years of age) with sickle cell anemia (SCA; HbSS and HbS 0 thalassemia) [7]. Based on the randomized, placebo-controlled, multicenter, BABY HUG trial (PMID: 21571150), HC is also commonly prescribed off-label by pediatric hematologists for infants (9-18 months old) (PMID: 20223921). Despite HC being front-line therapy for children and adults with SCA, there is a paucity of high-quality evidence regarding toxicity rates observed when employing two accepted dosing strategies: minimum effective dosing and maximum tolerated dosing [1,3]. Toxicity data are of critical importance as novel anti-sickling agents such as Oxbryta (voxelotor) and Adakveo (crizanlizumab)

METHODS
Studies for potential inclusion were identified by searching PubMed  Incidence rates (Y i /T i ) of each toxicity, along with 95% confidence intervals (CIs), were calculated by dividing the frequency of reported toxicities (Y i ) with the exposure patient years (PY) of follow-up (T i ) for each study. Using a random intercept Poisson regression model (Proc NLMIXED), incidence rates were averaged to calculate the summarized incidence rate for studies that employed fixed dosing and those that employed escalated dosing separately [11]. In post hoc sensitivity analyses, data from three studies deemed as outliers, that is, those with remarkably high frequency of toxicities were excluded and summarized incidence rates recalculated. All analyses were performed using the Statistical Analysis System (SAS) statistical software package version 9.4 (SAS Institute Inc., Cary, NC, USA).

RESULTS
Nine studies, described in Supplementary table 1, were included in the quantitative analysis. Figure 1 shows the Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart for study inclusion.

DISCUSSION
In this meta-analysis, incidence rates of neutropenia and thrombocytopenia in the post hoc analyses appear higher when escalating HC to a maximum tolerated dose compared with utilizing a lower fixed dose; however, reticulocytopenia was comparable between the two dosing strategies. The incidence rates of hepatic and renal toxicities were slightly higher when using fixed dosing when compared with escalated doses. However, the results of our analysis were limited by the small number of studies reporting toxicity-related information, imprecisely summarized incidence rates, and high heterogeneity between the studies, thus affecting our ability to draw conclusions about differences in toxicities related to the two HC dosing strategies.
For our meta-analysis, only nine studies had information related to the toxicities. No information on toxicities was reported in 37 potentially eligible articles. Even among the studies with information, details on many specific toxicities were missing. Due to the myelosuppressive nature of HC, standard baseline measurements before its administration should have included a full blood count, liver function tests, and urea and creatinine levels [15]. Therefore, information related to the toxicities may be available but not reported due to a lack of relevance for a particular publication. Some studies indicated the incidence of HC-related toxicities, but not the number of subjects with adverse  [12], which increased the 95% CI of the summarized neutropenia incidence. HC dosing of 20 mg/kg/day in children who were 9-18 months old at treatment initiation probably was associated with increased neutropenia because of the high frequency of viral infections in this very young age group [13]. Upon exclusion of these data, the 95% CI was relatively precise.
There were several sources of heterogeneity in our meta-analysis, such as age and genotype. We tried to address this issue by using random-effect modeling; however, heterogeneity persisted. We could not explore the sources of heterogeneity due to the small number of studies in escalated and fixed dosing groups and lack of information on toxicities by genotype and age group.
The lack of consensus regarding dosing strategy due to concern for short-term toxicities may lead to suboptimal use of HC. To address this critical gap in the literature, we attempted a meta-analysis comparing the toxicities associated with escalated and fixed dosing methods; however, we were unable to make conclusions on toxicities except neutropenia, mainly due to the lack of complete and systematic information on toxicities in many reports. Moving forward, we recommend diligent and standardized reporting of toxicities to help better understand the clinical effects of different dosing strategies.

AUTHOR CONTRIBUTIONS
JGM conducted the statistical analysis, interpreted the data, and