Robust Clinical and Laboratory Response to Hydroxyurea Using Pharmacokinetically Guided Dosing for Young Children with Sickle Cell Anemia

Abstract

Hydroxyurea is FDA-approved and now increasingly used for children with sickle cell anemia (SCA), but dosing strategies, pharmacokinetic (PK) profiles, and treatment responses for individual patients are highly variable. Typical weight-based dosing with step-wise escalation to maximum tolerated dose (MTD) leads to predictable laboratory and clinical benefits, but often takes 6–12 months to achieve. The Therapeutic Response Evaluation and Adherence Trial (TREAT, NCT02286154) was a single-center study designed to prospectively validate a novel personalized PK-guided hydroxyurea dosing strategy with a primary endpoint of time to MTD. Enrolled participants received a single oral 20 mg/kg dose of hydroxyurea, followed by a sparse PK sampling approach with 3 samples collected over three hours; analysis of individual PK data into a population PK model generated a starting dose that targets the MTD. The TREAT cohort (n=50) was young, starting hydroxyurea at a median age of 11 months (IQR 9–26 months), and PK-guided starting doses were high (27.7 ± 4.9 mg/kg/day). Time to MTD was 4.8 months (IQR 3.3–9.3), significantly shorter than comparison studies, thus meeting the primary endpoint. More remarkably, the laboratory response for participants starting with a PK-guided dose were quite robust, achieving higher hemoglobin (10.1 ± 1.3 g/dL) and HbF (33.3 ± 9.1%) levels than traditional dosing. Though higher than traditional dosing, PK-guided doses were safe without excess hematologic toxicities. Our data suggest early initiation of hydroxyurea using a personalized dosing strategy for children with SCA provides laboratory and clinical response beyond what has been seen historically with traditional, weight-based dosing.

Introduction

Hydroxyurea is FDA approved and is the only widely utilized disease-modifying therapy for the treatment of sickle cell anemia (SCA), with proven benefits for the reduction of morbidity and mortality in both adults and children.^1–5 Untreated SCA results in significant progressive damage to nearly all organs, but most notably the spleen,⁶ kidneys,⁷ heart,^8,9 lungs,^10,11 and brain,^12,13 beginning as early as the first year of life. With decades of clinical research involving patients with SCA, there is no signal to suggest that hydroxyurea has short- or long-term risks that outweigh its benefits, with improvement noted in nearly all clinical outcomes and even reduced mortality.^2–4,14,15

Based on this large body of evidence including the placebo-controlled infant hydroxyurea (BABY HUG) trial,^1,2 the 2014 NHLBI evidence-based guidelines strongly recommended offering hydroxyurea to children with SCA as young as 9 months of age, regardless of clinical severity.¹⁶ Unfortunately, many children with SCA still are not prescribed hydroxyurea until they manifest overt or serious clinical complications.⁵ Hydroxyurea therapy for SCA thus represents a challenging clinical paradigm in which an affordable, available, safe, and effective once-daily oral medication for a potentially life-threatening disease is not universally prescribed.¹⁷

In addition to the underuse and delayed initiation of hydroxyurea, there is also lack of standardization regarding the optimal dosing, monitoring, and dose escalation processes, especially regarding treatment targets and thresholds for toxicity.¹⁸ There is well-described variability in hydroxyurea pharmacokinetics (PK) including absorption profiles, distribution, and clearance;¹⁹ variability in pharmacodynamics (PD), including hemoglobin (Hb) and fetal hemoglobin (HbF) responses; and differences in the maximum tolerated dose (MTD) of hydroxyurea for adults and children with SCA.²⁰

Despite this inter-patient PK and PD variability, initial hydroxyurea dosing is weight-based with a starting dose of 10–20 mg/kg/day.^16,21 Dose escalation, if performed, typically involves a trial-and-error approach that targets mild myelosuppression and requires considerable effort by both patients and providers, with frequent monitoring of blood counts, assessment of medication adherence, and dosing adjustments. Even with experienced providers, dose escalation to MTD usually takes 6–12 months,^15,22,23 resulting in sub-optimal hydroxyurea exposure and ongoing risk for short- and long-term disease complications.

Following the release of the NHLBI guidelines that recommended early hydroxyurea initiation as early as the first year of life, we sought to improve hydroxyurea dosing and treatment responses by prospectively evaluating a novel PK-guided dosing model.²² The Therapeutic Response Evaluation and Adherence Trial (TREAT, ClinicalTrials.gov identifier NCT02286154) uses an individualized PK-based hydroxyurea dosing strategy to identify a starting dose that approximates the eventual MTD, without the need for multiple dose escalations or adjustments. The primary objective of TREAT was to simplify and shorten hydroxyurea dose escalation so children receive an optimal dose based on their individual PK parameters, representing a more efficient strategy than traditional weight-based dosing and stepwise escalation to MTD. We now describe the primary TREAT results for 50 children with SCA.

Methods

Study Overview.

TREAT was a prospective, open label study of hydroxyurea with a primary objective of developing and evaluating a personalized PK-guided dosing strategy for children with SCA at hydroxyurea initiation. Children (age 6 months to 21 years) were eligible to enroll after their parents or legal guardians decided to initiate hydroxyurea therapy; in this PK-guided dosing study, the primary study endpoint was time to MTD.

PK Sampling and Hydroxyurea Measurement.

We developed a Bayesian adaptive control strategy to accurately estimate individual systemic hydroxyurea exposure using a sparse sampling schedule, to allow the model to be clinically practical outside of this clinical trial.²² Whole blood was collected at three time points (15–20 minutes, 50–60 minutes, and 180 minutes) following a single oral 20 mg/kg dose of liquid hydroxyurea. Quantitative measurement of the plasma hydroxyurea concentration was performed by high performance liquid chromatography (HPLC),^24,25 using 150–200 μL of plasma. Blood collection was performed most commonly from a finger, toe, or heel stick, but often, one of the three PK samples were collected by venipuncture at the time of additional baseline laboratory studies. PK parameters were calculated using MW/Pharm (Mediware, Prague, Czech Republic).

Initial Dose Selection.

Hydroxyurea concentrations were incorporated into the previously described population PK-model using MW/Pharm.²⁶ This model was generated using PK data from the Hydroxyurea Study of Long-term Effects (HUSTLE, ClinicalTrials.gov NCT00305175 ), in which the average hydroxyurea area under the concentration-time curve (AUC) increased from 92.9 ± 23.4 mg*L/h at baseline to 115.7 ± 34.0 mg*L/h at MTD.^19,22 To use the PK-guided dosing strategy to initiate treatment at the optimal (MTD) dose, individual PK-data from the 20 mg/kg test dose were used to calculate a starting dose projected to yield an AUC of 115 mg*L/h. Figure 1 illustrates how this software incorporates individual PK data to select a starting dose with this AUC target.

Figure 1.

Prediction of hydroxyurea starting dose using individual PK parameters.

The PK-guided dose was determined and initiated for each TREAT participant typically within 7–10 days of the test dose, even if the daily dose (mg/kg) was much higher than the traditional weight-based dosing algorithms. However, if the PK-based recommended starting dose was >35 mg/kg/day, the selected starting dose was 30–35 mg/kg/day, recognizing that the 35 mg/kg maximum dose limit is somewhat arbitrary and not evidence-based for SCA.^16,20 If the PK sampling process was insufficient to generate a predicted starting dose, the participant started at 20 mg/kg/day.

Dose Escalation and Monitoring.

With the goal of practical patient care following initial dose selection, TREAT did not mandate the frequency of patient follow-up, laboratory monitoring, or dose adjustments; these were performed by each primary hematology provider according to our local institutional Hydroxyurea Clinical Guidelines (see Supplement). Briefly, hydroxyurea is escalated by 2.5–5.0 mg/kg/day every 4–6 weeks until laboratory values fall within a target range, typically the absolute neutrophil count (ANC) ≤3.0×10⁹/L. These guidelines recommend holding hydroxyurea for one week if laboratory toxicity occurs, defined as ANC <1.0 ×10⁹/L, platelet count <80×10⁹/L, absolute reticulocyte count (ARC) <80 ×10⁹/L with hemoglobin <9.0 g/dL, or hemoglobin <5.0 g/dL regardless of ARC.

Following the initial PK visit, there were no study specific visits and both laboratory and clinical data were collected as the participants presented for routine scheduled follow-up. In an effort to obtain a sense of medication adherence, at each clinical visit, the study team asked the family how many doses they missed and attempts were made to obtain records of pharmacy refills.

Determination of Maximum Tolerated Dose.

As time to MTD was the primary study endpoint, we standardized the definition of MTD for this study. MTD was declared when two criteria were met: 1) No change in dose over the previous eight weeks; and 2) adequate myelosuppression with ANC between 1.5–3.0 × 10⁹/L on two consecutive blood counts. In some cases, MTD was defined by ARC suppression. Two independent pediatric hematologists retrospectively reviewed all de-identified dosing and laboratory data for each patient, and reached consensus on time to MTD.

Three previous NHLBI-funded clinical trials were used for comparison for the primary TREAT endpoint of time to MTD. HUSTLE was a prospective, observational cohort study of children with SCA that evaluated hydroxyurea PK/PD and long-term effects of hydroxyurea; as a single institution study it is comparable to TREAT in terms of routine clinical monitoring and adherence patterns.^19,27 Stroke With Transfusions Changing to Hydroxyurea (SWiTCH, NCT00122980) was a multicenter phase 3 randomized trial comparing blood transfusions and iron chelation to hydroxyurea and phlebotomy for children with SCA, stroke, and iron overload.²³ TCD With Transfusions Changing to Hydroxyurea (TWiTCH, NCT01425307) was a multicenter, phase 3 randomized trial comparing hydroxyurea to blood transfusions for children with abnormal transcranial Doppler (TCD) velocities.¹⁵ SWiTCH and TWiTCH were both rigorous clinical trials with monthly visits mandated per study protocol. These three comparison studies (HUSTLE, SWiTCH, and TWiTCH) used ANC<4.0×10⁹/L as the target range to define MTD, which was less myelosuppression than the TREAT MTD criteria (<3.0 ×10⁹/L).

Documentation of Clinical Events:

TREAT participants were carefully monitored with documentation of all clinical events requiring access to care, including ED visits, sick visits to the sickle cell clinic and hospitalization. Definitions of vaso-occlusive pain events included any event requiring oral pain medication without another explanation (e.g. injury). Acute chest syndrome was defined using the criteria described in BABY HUG and independently reviewed by an independent, masked pediatric hematologist.

Statistical Analysis.

For sample size calculations, we averaged several published studies to estimate the average time to MTD as 9 ± 4 months.^{15,23,27–29} We hypothesized that the PK-based dosing strategy would reduce time to MTD by 2 months; to detect a statistically significant difference with 80% power, a minimum sample size of 25 was required. As TREAT was designed to evaluate the applicability of this PK-based dosing strategy, all participants were analyzed, including those with insufficient PK sampling who started at 20 mg/kg/day. Time to MTD was estimated using the Kaplan-Meier method with censoring at 24 months. Differences in time to MTD were compared between TREAT and the three comparison studies (HUSTLE, SWiTCH, TWiTCH) using a two-sided log-rank test. Categorical and continuous data were described by frequency (percent) and mean ± SD, respectively. Univariate and multivariate linear regression was used to determine associations with continuous outcomes with relevant covariates. For comparisons between the TREAT and BABY HUG, continuous outcomes were compared using two sided Welch’s t-tests, and incidence rate differences were compared using the method outlined in Sahai and Khurshid (1996).³⁰ Of note, the primary BABY HUG manuscript reported only mean laboratory values without standard deviations,¹ which precluded complete statistical comparisons with participants enrolled in TREAT.

Results

Demographics and Baseline Laboratory Characteristics.

As of November 2018, a total of 51 children with SCA (50 HbSS, 1 HbSβ⁰-thalassemia) were enrolled in TREAT and contributed 105.6 patient-years on treatment. Acceptance of the study was high with enrollment of 51 of the 56 (91%) families approached. Of the 5 families that declined, 2 were not interested in research at all, 2 were not interested in the PK portion of the study and 1 was still skeptical of starting hydroxyurea at all. None had deletional hereditary persistence of fetal hemoglobin (HPFH) by specific genetic testing. One family decided to not initiate hydroxyurea after enrollment of their child, so the analysis included only 50 participants who started hydroxyurea. The median age at hydroxyurea initiation was 11 months (IQR 9–26 months), with slightly more males (58%) than females (42%). Most children (36/50=72%) started before 2 years of age and many (27/50=54%) started before their first birthday.

Baseline laboratory parameters (Table 1) varied with age, although anemia and reticulocytosis were present even within the first year of life. The average HbF at enrollment was 25.1 ± 11.6%, consistent with their young age. In contrast, the average baseline ANC was not elevated (3.6 ± 1.9 × 10⁹/L), and 48% of the children had ANC <3.0 × 10⁹/L and 22% had ANC <2.0 × 10⁹/L, levels that would normally preclude dose escalation. A total of 48 of 50 participants had alpha-globin gene analysis; 3 (6%) were homozygous for the common single 3.7kb deletion of the α-globin gene (-α^−3.7/ -α^−3.7), 10 (21%) were heterozygous for the single α-globin gene deletion (-α^−3.7/ αα), and two children (4%) had five α-globin genes, with the ααα^{anti 3.7} α-globin gene triplication.

Table 1.

Laboratory values and trends with PK-guided initial dosing of hydroxyurea.

	Age at Hydroxyurea Initiation			All Participants
	< 12 months N = 27	12–24 months N = 9	>24 months N = 14
Hemoglobin, g/dL
Baseline	9.1 ± 1.2	9.5 ± 1.3	8.9 ± 1.4	9.1 ± 1.3
6 months	10.1 ± 1.4	10.2 ± 1.1	10.9 ± 1.0	10.3 ± 1.3
12 Months	10.2 ± 1.3	10.0 ± 1.1	10.2 ± 1.5	10.1 ± 1.3
Baseline to 12 M	+1.0 ± 1.2*	+0.7 ± 1.5	+1.1 ± 1.4*	+1.0 ± 1.3*
HbF, %
Baseline	31.4 ± 8.1	24.1 ± 9.0	14.0 ± 8.0	25.1 ± 11.6
6 months	37.9 ± 8.5	31.5 ± 9.2	30.4 ±9.9	34.6 ± 9.5
12 Months	35.8 ± 9.1	31.9 ± 9.5	28.4 ± 7.0	33.3 ± 9.1
Baseline to 12 M	+4.4 ± 7.9*	+9.2 ± 10.1*	+14.4 ± 6.3*	+7.8 ± 8.9*
MCV, fL
Baseline	77.8 ± 9.7	75.0 ± 8.9	83.9 ± 7.4	79.0 ± 9.4
6 Months	86.4 ± 10.0	87.4 ± 16.1	102.2 ± 9.0	90.7 ± 13.0
12 Months	88.1 ± 8.6	86.6 ± 13.5	105.5 ± 21.5	91.9 ± 15.3
Baseline to 12 M	+10.6 ± 5.0*	+13.0 ± 6.8*	+19.2 ± 21.0*	+13.2 ± 11.4*
ARC, ×109/L
Baseline	273 ± 150	269 ± 95	345 ± 91	292 ± 129
6 Months	143 ± 76	179 ± 56	153 ± 75	153 ± 72
12 Months	136 ± 75	214 ± 92	186 ± 63	164 ± 81
Baseline to 12 M	−139 ± 133*	−62 ± 129	−172 ± 70*	−130 ± 123*
WBC, ×109/L
Baseline	12.7 ± 5.8	13.8 ± 6.1	13.2 ± 4.0	13.0 ± 5.3
6 Months	9.8 ± 4.0	10.9 ± 4.9	7.4 ± 2.6	9.4 ± 4.0
12 Months	9.5 ± 4.2	13.2 ± 5.8	8.4 ± 3.0	10.0 ± 4.6
Baseline to 12 M	−2.2 ±6.9*	−0.9 ± 4.7	−6.3 ± 4.4*	−2.9 ± 6.2*
ANC, ×109/L
Baseline	2.8 ± 1.3	4.0 ± 2.6	4.7 ± 1.6	3.6 ± 1.9
6 Months	2.7 ± 1.8	3.4 ± 2.6	2.9 ± 1.6	2.9 ± 2.0
12 Months	3.0 ± 2.5	3.7 ± 2.0	3.7 ± 2.1	3.3 ± 2.3
Baseline to 12 M	+0.4 ± 2.7	−0.5 ± 2.6	−1.6 ± 3.1	−0.3 ± 2.8
Platelets, ×109/L
Baseline	325 ± 114	327 ± 78	365 ± 162	337 ± 123
6 Months	333 ± 216	322 ± 142	303 ± 186	323 ± 190
12 Months	297 ± 83	282 ± 117	327 ± 186	301 ± 118
Baseline to 12 M	−30 ± 106	−39 ± 72	−67 ± 179	−41 ± 119*

^*indicates p<0.05 for baseline to month 12 comparisons.

Hydroxyurea Concentrations and PK Profiles.

Figure S1 (Supplementary Information) summarizes PK curves for each participant, demonstrating significant variability. Most children (68%) had their peak plasma concentration at the 1-hour timepoint, although some absorbed hydroxyurea quickly with peak concentration at 15-minutes and others had 3-hour concentrations higher than the 1-hour values. PK profiles showed some differences with age, likely secondary to renal hyperfiltration in infants. Interpatient variability in the initial hydroxyurea exposure was observed with average AUC = 80.4 ± 20.0 mg*h/L (IQR 67.4−-92). Complete PK parameters for the TREAT cohort, including AUC, half-life, clearance, and volume of distribution are summarized in Table S1 (Supporting Information).

PK-Based Starting Doses.

The PK collection with dose-modeling strategy was unsuccessful for only three participants, who began treatment at 20 mg/kg/day. These all occurred early in the study with the last 30 participants all having sufficient PK data. Two of the three insufficient PK collections were were performed while patients were hospitalized, and PK samples were drawn from a peripheral IV. We suspect that the samples were diluted despite attempts to waste prior to PK sample collection. With these failures, we no longer draw PK samples from peripheral IVs. The third failure was due to insufficient blood collection in a young infant. For the remaining 47 participants, the PK-guided predicted starting dose needed to achieve the target AUC of 115 mg*h/L was 30.6 ± 8.3 mg/kg/day. However, due to the decision to start all children at <35 mg/kg/day, the actual average starting dose for the TREAT cohort was 27.7 ± 4.9 mg/kg/day. Using the PK model, 21/50 (42%) starting doses were ≥30.0 mg/kg/day, 16/50 (32%) were 25.0–29.9 mg/kg/day, 10/50 (20%) were 20.0–24.9 mg/kg/day, and only 3 (6%) started at <20.0 mg/kg/day.

Dose Progression and Time to MTD.

The mean duration of treatment at data analysis was 26 ± 15 months; 39 participants have had MTD declared, with median time to MTD of 4.8 (IQR 3.3–9.3) months, which is significantly shorter than time to MTD in the three comparison studies with median 7.6, IQR 5.9–11.7 months (Figure 2, p<0.0001). The average dose at MTD declaration was 26.7 ± 4.8 mg/kg/day, and most children did not require adjustments from their original PK-guided starting dose (64%) or had only adjustments for weight gain (18%) before reaching MTD. Only 2 children required ≥5 mg/kg/day dosing decreases from the recommended starting dose, both due to asymptomatic neutropenia; similarly, only 3 children required dose escalations ≥5 mg/kg/day from their starting dose. To date, 39 of 44 participants treated for >6 months (89%) have achieved MTD, comparable to SWiTCH (60/67=90%), TWiTCH (57/60=95%), and HUSTLE (74/87=85%). Multivariate analysis, including age and baseline laboratory values, did not identify any significant predictors of the dose or time to MTD.

Figure 2.

Time to MTD comparing TREAT and three other prospective studies. A total of 50 participants received hydroxyurea in TREAT, compared to 87 in HUSTLE, 60 in SWiTCH, and 67 in TWiTCH. The time to MTD was significantly shorter (p<0.0001) for participants enrolled in TREAT (median 4.8, IQR 3.3–9.3 months) compared to the three comparison studies (7.6, IQR 5.9–11.7 months).

Fetal Hemoglobin Response and Laboratory Trends.

Despite the early age of hydroxyurea initiation with relatively high baseline Hb (9.1 ± 1.3 g/dL) and HbF (25.1 ± 11.6%), there were significant improvements in all hematological parameters with a robust effect noted within 6 months of treatment initiation (Table 1). The average Hb concentration increased by 1.0 g/dL with 58% of participants achieving Hb ≥10.0 g/dL and 29% reaching Hb >11.0 g/dL after 12 months of therapy. Despite their high starting HbF levels, 87% of the children had further increases, with an average HbF value of 33.3 ± 9.1% after 12 months of therapy and nearly one-third of participants achieving HbF values >40%. Fetal hemoglobin responses were robust across all ages, but especially for those who started ≤1 year of age, with 48% of these infants achieving HbF >40% after 12 months of therapy. Univariate analysis identified age and baseline HbF as predictors of the HbF response, but only baseline HbF remained a significant predictor of ultimate MTD after multivariate analysis (p=0.0001).

Hematological Toxicities and Medication Holds.

Hydroxyurea was very well tolerated despite the relatively high average starting dose of nearly 28 mg/kg/day. For 30 study participants, there were no treatment interruptions, while 20 children had a total of 29 temporary hydroxyurea holds (range 1–3). There were no associations between the starting dose and hematological toxicities; most (22/29=76%) were transient neutropenia (ANC <1.5×10⁹/L), typically discovered during hospitalization or evaluation for viral-like febrile illnesses. Only 7 episodes of severe neutropenia (ANC <0.5×10⁹/L) occurred in 5 participants, but none had prolonged cytopenia or documented infection. Other cytopenias included thrombocytopenia in two children (one with splenic sequestration), anemia with reticulocytopenia in three children and severe anemia with a normal/high reticulocyte count in two children.

Clinical Events.

Consistent with the robust laboratory response, acute sickle cell complications were infrequent in this cohort, particularly among children initiating hydroxyurea within the first 2 years of life. As expected in a young cohort, febrile viral-like illnesses frequently led to medical evaluation and hospital admission. Among 50 participants, there were 90 hospitalizations and 156 additional evaluations (ED, sickle cell clinic) that did not result in hospitalization. Most hospital admissions were for febrile episodes, and only 3 participants developed acute chest syndrome.

Comparison to BABY HUG.

Most participants in the TREAT cohort started hydroxyurea at a similar age to children enrolled in the BABY HUG study, which initiated hydroxyurea between 9–18 months of age and maintained a fixed dose of 20 mg/kg/day.¹ Table 2 compares laboratory benefits and toxicities, as well as acute clinical complications, for TREAT participants compared to published data from BABY HUG.^1,2 The laboratory benefits in TREAT were notably better despite the similar starting age, with more robust hemoglobin and HbF responses using the individualized, PK-guided increased dosing strategy. There were no significant differences in the frequency of laboratory toxicities between TREAT and BABY HUG, but a significantly lower frequency of sickle cell complications in the TREAT cohort.

Table 2.

Comparison of PK-Guided hydroxyurea dosing to fixed dosing in young infants with SCA. Laboratory changes BABY HUG are from baseline to 24 months, for the children who randomized to hydroxyurea and had paired laboratory data, while changes for TREAT are from baseline to 12 months and for the participants who initiated treatment before 2 years of age. The BABY HUG manuscript did not report standard deviations for laboratory changes and therefore direct statistical comparisons were not possible. Clinical events are presented as events per 100 patient-years.

	Fixed Dose (BABY HUG)	PK-Guided Starting Dose (TREAT)	p-value
N	96	36
Starting Age, months	13.6 ± 2.6	12.1 ± 4.8	0.0824
Patient-Years	189	77
Starting Dose (mg/kg/day)	20.0	28.0
HbF Change, %	25.6 → 22.4	29.0 → 34.8
Hb Change, g/dL	8.9 → 9.1	9.2 → 10.1
MCV Change, fL	80 → 92	76 → 88
ANC Change (x109/L)	4.9 → 4.5	3.1 → 3.2
ARC Change (x109/L)	286 → 227	275 → 157
Platelet Count Change (x109/L)	374 → 351	326 → 293
Dactylitis, per 100 patient-years	12.6	0	0.0017
All vasoocclusive pain events, per 100 patient-years	93.6	6.5	<0.0001
Splenic Sequestration, per 100 patient-years	6.3	2.6	0.44
Hospitalization, per 100 patient-years	123	93.5	0.041
Transfusion, per 100 patient-years	18.5	19.4	0.88

Discussion

The introduction of hydroxyurea has transformed the clinical treatment paradigm of SCA, particularly for children who begin treatment before organ damage and chronic sequelae have manifested. Our prospective study demonstrates that early, individualized, PK-guided dosing of hydroxyurea initiated at the projected MTD is associated with a robust and sustained HbF response, with shorter time to MTD and improved clinical benefits compared to traditional weight-based dosing. While we believe that personalized dosing and targeted hydroxyurea exposure is optimal, these results illustrate several important lessons about the age of hydroxyurea initiation and the tolerability and response to higher starting doses of hydroxyurea. Though there is not wide accessibility to PK-guided dosing at this time, these data provide evidence to support the early initiation of hydroxyurea with aggressive dose escalation is both safe and highly effective, compared to traditional dosing strategies.

To move from weight-based dosing to an individualized dosing model, we utilized the known PK data of hydroxyurea for children with SCA and developed a novel, PK-guided Bayesian approach to select individualized starting hydroxyurea doses intended to approximate the MTD and optimize the HbF response. TREAT was designed to develop a personalized, PK-guided dosing strategy in a manner that could potentially become clinically practical, if measurement of hydroxyurea becomes more widely available. Our data demonstrate that a sparse PK sampling approach around an initial test dose was able to accurately estimate hydroxyurea exposure and allow calculation of a patient-specific dose requiring only a 3-hour clinical visit, considered a feasible timeframe for patients and families. The PK-guided starting doses in the TREAT cohort were significantly higher than traditional dosing, but this individualized dosing was safe and significantly reduced the time to achieve MTD. The high acceptance rate by families (91% of families approached enrolled in the study) and the high success rate of PK sampling (94%) was encouraging and provides support for our intention to make this PK-sampling approach amenable to both families and providers. The follow-up, randomized, multicenter Hydroxyurea Optimization through Precision Study (HOPS, ClinicalTrials.gov NCT03789591) will further evaluate the feasibility, safety, and benefits of providing PK-guided dosing to children across the country. If PK-guided dosing continues to demonstrate benefit compared to standard dosing, we intend to increase the availability of PK-guided dosing as an option for patients and providers outside of the context of a clinical trial.

Prior to prospective evaluation of this personalized dosing model, we hypothesized that PK-guided dosing would achieve MTD more quickly, but did not predict such robust laboratory responses with HbF levels routinely >30–40%. The reason for this robust hematologic effect is likely multi-factorial, including the young age and higher starting dose compared to traditional dosing strategies. Starting at the optimal dose was critical to achieving these high HbF values, since the normal baseline WBC and ANC laboratory values for these infants would not have permitted dose escalation using step-wise algorithms that target an ANC <3–4 × 10⁹/L. Despite our high starting doses and relatively low baseline ANC values, there were very few dose modifications and no unexpected laboratory or clinical toxicities.

Without treatment, infants with SCA have a gradual reduction in HbF with <20% levels in most children by 2 years of age.³¹ In BABY HUG, hydroxyurea was initiated at a mean age of 13.6 months, but average HbF levels actually declined from a baseline of 25.6% to 22.4% after 24-months of treatment (Table 2), with fixed-dose drug benefits offset by the expected natural physiological silencing of HbF.^32–34 In the TREAT cohort, HbF increased in nearly all participants from a baseline of 25.1% to 33.3% after 12 months of hydroxyurea. As expected given more extreme starting values due to more severe SCA, the absolute changes in hematologic parameters (Table 1) were expectedly the greatest for the oldest patients, but the true “response” as defined by the hemoglobin and HbF achieved appeared highest for the youngest cohort initiating hydroxyurea before the first year of life. In addition to HbF response, additional hematological parameters were more pronounced in the TREAT cohort, including increase in hemoglobin and decrease in absolute reticulocyte count. Interestingly, the MCV in the TREAT participants less than 2 years of age was not as high as older children or the BABY HUG cohort. Alpha-thalassemia trait was not a factor, as there was decreased prevalence of alpha-thalassemia trait in TREAT (27%) compared to the hydroxyurea arm of BABY HUG (40%). It is possible that the younger starting age of the youngest TREAT cohort and the lower ultimate reticulocyte count may contribute to the lower MCV in these young children. Together, these data suggest that early initiation of hydroxyurea or individualized dosing (or more likely, a combination of both) are capable of further inducing HbF, despite the expected physiologic decline, to levels sufficient to inhibit sickling in the majority of erythrocytes and minimize clinical complications. Future studies of the TREAT cohort will investigate the contribution of genetic variation on hydroxyurea pharmacokinetics and HbF response.

Based on our data, we hypothesize that early hydroxyurea therapy, before the genetic and epigenetic silencing of HbF, can effectively induce HbF expression, especially when given at MTD. Realizing that TREAT is a single arm and single institution study, we will further explore and attempt to validate these results in the upcoming Hydroxyurea Optimization through Precision Study (HOPS, ClinicalTrials.gov NCT03789591). TREAT had three participants with insufficient PK samples who started at 20 mg/kg/day, but these three participants do not provide sufficient data to allow for comparison of standard (20 mg/kg/day) vs. PK-guided starting dosing. HOPS is a prospective, multicenter, randomized trial of PK-guided (TREAT approach) versus weight-based dosing of hydroxyurea for children ages 6 months to 21 years with SCA. This prospective and multicenter study will allow validation as to the feasibility of PK sampling outside of our single center and will determine whether PK-guided dosing results in more robust hematologic and clinical response compared to weight-based design with an aggressive dose escalation strategy. In addition, the study will allow more broad exploration of the benefits of PK-guided dosing across ages, as TREAT enrolled mostly young children. Ancillary aims will include studies of gene expression, gene silencing, both before and after initiation of hydroxyurea, to further investigate the benefits of this dosing model and to improve our understanding of how hydroxyurea maintains robust HbF expression.

Given historical concerns about the myelosuppressive effects of hydroxyurea, older dosing strategies were conservative, aiming to achieve some benefits while avoiding toxicities. As evidence and clinical experience grew, however, dosing strategies became more flexible to allow some toxicities, to lower toxicity thresholds, and to target a goal of HbF levels of 20%, often considered a clinically meaningful threshold.^27,35 Hydroxyurea dosing has remained weight-based, typically starting at 15–20 mg/kg/day with subsequent trial-and-error based dose escalation to achieve MTD.³⁶ Although some providers may feel comfortable starting at higher doses, there have been no prospective cohorts examining the benefits of this approach. The Phase I/II HUG-KIDS pediatric trial featured dose escalation to MTD, but held the hydroxyurea dose for ANC <2.0×10⁹/L, and did not achieve an average 20% HbF.²⁸ More recently, by lowering the ANC toxicity threshold to 1.5 or even 1.0×10⁹/L,^21,37 children treated with hydroxyurea reach MTD at ~25 mg/kg/day and can routinely achieve 25% HbF levels.^{15,18,23,27,38} But relatively few patients treated at MTD will consistent maintain HbF ≥30%, a level that approximates deletional HbS/HPFH, a benign form of SCA in which erythrocytes are protected from sickling due to high levels of pancellular HbF distribution.^39,40 In TREAT, the early initiation of PK-guided dosing led to most children achieving and maintain 30–40% HbF levels. A model-based publication in Blood in 2014 suggested that HbF >30% was a target “to achieve a pharmacologic cure of most disease manifestations.”⁴¹ Our TREAT data suggest that instead of a 20% clinically meaningful threshold, we should now target a higher and potentially curative HbF threshold of 30–40%.

The young age of our TREAT cohort limits the generalizability of this dosing strategy for older children and adolescents with SCA, though clinical and laboratory responses for older children were also robust. Due to the practice patterns and wide use of hydroxyurea of our institution, the eligible patients enrolling in TREAT were mostly young. There were only 14 participants starting hydroxyurea after 2 years of age, and their HbF responses were excellent (average HbF = 28.4 ± 7.0%) but not as high as for the young infants and toddlers who started with higher baseline HbF levels. The previously mentioned HOPS trial will prospectively evaluate this PK-guided dosing approach in a multi-center study that will likely enroll patients across a greater age range, though infants and toddlers are still likely to be the predominant age group initiating hydroxyurea due to current practice patterns and NHLBI guidelines. Evaluation of a PK-guided dosing approach in older adolescents and adults is a future direction, which is important to evaluate given the underlying chronic renal damage and limited bone marrow reserve in the older sickle cell population. Despite the age limitations of the TREAT cohort, these data support the safety and benefits of individualized dosing for children of all ages with the goal of pharmacodynamic (HbF) optimization.

In summary, our data document that a PK-guided hydroxyurea dosing strategy shortens the time to MTD, which should improve the treatment paradigm and patient outcomes. The clinical and laboratory benefits of personalized hydroxyurea dosing starting within the first 1–2 years of life make a more compelling case for early treatment targeting MTD at the initiation of therapy. Though the primary intention of this study was to evaluate the benefits of a personalized, PK-guided treatment model, this study provides strong evidence that early initiation of hydroxyurea with an aggressive dosing strategy to optimize HbF response is indicated even in the absence of PK studies. Traditional starting doses of 15–20 mg/kg/day certainly provide both laboratory and clinical benefit but do not achieve the target of minimizing all disease complications. While future studies will hopefully allow for more widespread availability of PK-guided dosing, we recommend a starting dose a starting dose of 25 mg/kg/day for all patients with subsequent dose escalation to optimize HbF response. In this era, we should be confident that hydroxyurea is safe, as has been proven over several decades. Dosing strategies should be confident and aggressive with a focus should upon optimizing HbF response in order to improve the quality and quantity of life for individuals affected with SCA.