Impact of hydroxyurea dose and adherence on hematologic outcomes for children with sickle cell anemia

Abstract

Background

Hydroxyurea is the primary treatment for sickle cell anemia (SCA), yet real-world implementation in high-income settings is suboptimal. Variation in prescribed hydroxyurea dose and patient adherence in these settings can both affect actual exposure to hydroxyurea. Quantifying the contributions of hydroxyurea dose and medication adherence to the relationship between hydroxyurea exposure and hematologic parameters could inform strategies to optimize exposure and improve outcomes.

Procedure

We evaluated the relationship between hydroxyurea exposure, defined by average prescribed dose and adherence, and hematologic parameters using data from children with SCA who were enrolled in two prospective hydroxyurea adherence studies. Hydroxyurea adherence was assessed by video directly observed therapy or electronic pill bottle and medication administration record. Average prescribed dose was abstracted from prescriptions in patients’ electronic medical record. Participants with a hydroxyurea exposure >20mg/kg/day and ≤20 mg/kg/day were included in the higher and lower exposure groups, respectively.

Results

Forty-five participants are included in the analysis (56% male; median age 12 years (range 2–19); 98% Black). Higher exposed participants (n=23) were prescribed a higher dose (27.2 vs. 24.4 mg/kg/day, P=0.002) and had better adherence (0.92 vs. 0.71, P=<0.001) compared to lower exposed participants (n=22). Higher exposure was associated with higher fetal hemoglobin (P=0.04) and mean corpuscular volume (P=0.02).

Conclusions

Higher hydroxyurea exposure is associated with improved hematologic parameters in the high-income setting and is affected by both prescribed dose and adherence. Future studies are needed to optimize both adherence and hydroxyurea prescribing and confirm that increasing exposure improves clinical outcomes in this setting.

Introduction

Hydroxyurea is the primary disease modifying therapy for individuals with sickle cell anemia (SCA). Studies demonstrate that hydroxyurea decreases SCA complications, including painful vaso-occlusive crises (VOC), acute chest syndrome (ACS), transfusions, hospitalizations, cost of care, and risk of early death^1–5. Despite increasing hydroxyurea use, a recent study suggests that rates of emergency department (ED) visits and hospitalizations remain unchanged among many children with SCA in clinical practice in high-income settings over the last decade⁶. While non-adherence likely limits the effect of hydroxyurea on clinical outcomes, there also remains uncertainty surrounding the most effective dosing strategy because there have been different dosing strategies have been used in different hydroxyurea clinical trials^1,2,7,8. While these trials showed that participants receiving hydroxyurea, regardless of the dosing regimen had improved clinical outcomes, there remain many important questions about appropriate dosing and associated adherence strategies to enhance hydroxyurea exposure and optimize outcomes in the real-world setting.

Recently, a trial of children with high adherence to hydroxyurea in Sub-Saharan Africa compared dose escalation to fixed dose hydroxyurea using more liberal dose escalation criteria than the 2014 National Heart Lung and Blood Institute’s Sickle Cell Disease (SCD) Guidelines⁹. The study found that dose escalation was safe and that those randomized to dose-escalation (mean hydroxyurea dose of 29.5 mg/kg/day) had improved hematologic parameters and fewer VOC and ACS episodes than those randomized to the fixed dose group (mean hydroxyurea dose of 19.2 mg/kg/day)¹⁰. These findings suggest that increasing hydroxyurea exposure is highly effective in improving outcomes in sub-Saharan Africa, but the tremendous disparities in resources, treatment paradigms, adherence behavior, and ability to access acute care for common SCA complications in high-income settings compared to low-income settings may limit the generalizability of these findings. Thus, additional studies are needed to better understand the impact of hydroxyurea exposure on outcomes in high-income settings where dosing and adherence are more variable^6,11–19.

To address this gap, rigorous quantitative methods must be utilized to measure hydroxyurea adherence. Directly observed therapy and electronic pill bottles are considered the gold standard methods to quantify medication adherence^20–24. We used these validated adherence measures to monitor adherence in two separate pediatric clinical trials evaluating the effect of electronic hydroxyurea adherence-promoting interventions at two SCA centers (NCT02578017 and NCT04675645) where hydroxyurea dose escalation to 30–35 mg/kg/day is common. The objective of the study was to evaluate the relationship of hydroxyurea exposure, as assessed by prescribed dose of hydroxyurea and adherence, to hematological parameters in children with SCD in the United States. Compared to most previously published studies that only report the prescribed dose of hydroxyurea, this analysis incorporated hydroxyurea adherence to better describe actual hydroxyurea exposure. We hypothesize that higher hydroxyurea exposure (>20 mg/kg/day) will be associated with improved hematologic outcomes.

Methods

Study Design and Participants

This was a secondary data analysis of data from two prospective Institutional Review Board approved studies at two Midwestern pediatric SCD centers. These studies sought to determine if electronic interventions, Mobile Directly Observed Therapy (Mobile DOT) and electronic pill bottles (AdhereTech devices), increased hydroxyurea adherence among pediatric patients with SCD. English-speaking patients with SCD at Nationwide Children’s Hospital (NCH) who were ≤19 years, were prescribed hydroxyurea for at least the previous 180 days, were not receiving chronic transfusion therapy, and had daily access to a smartphone that could do VDOT were eligible for the Mobile DOT study¹². Mobile DOT participants were instructed to self-record (or have their consenting caregiver record if <11 years of age) their daily hydroxyurea administrations with their smartphone and email these videos to the secure website for review. Mobile DOT participants’ hydroxyurea adherence for these analyses was measured by video directly observed therapy (VDOT) when the participant was at home and by the inpatient medical administration record (MAR), if hospitalized.

The electronic pill bottle study was a 12-month study at Ann & Robert H. Lurie Children’s Hospital of Chicago (LCH). The pill bottles record each bottle opening to electronically monitor hydroxyurea adherence. English-speaking patients 8–21 years old were eligible if they had SCD, were on a stable dose of hydroxyurea for ≥2 months at the time of enrollment and were not receiving chronic transfusion therapy. Participants also had to be prescribed the pill formulation of hydroxyurea, as the bottles are not compatible with liquid medication. Participants were instructed to use the bottles to store their hydroxyurea throughout the study period. The number of unique days the bottle was opened was recorded on a secure dashboard in real-time via the self-contained cellular connectivity within each bottle. Electronic pill bottle participants’ hydroxyurea adherence for these analyses was measured by the electronic pill bottles when the participant was at home and by the MAR, if hospitalized.

Only Mobile DOT and electronic pill bottle participants with hemoglobin SS or hemoglobin Sβ⁰-thalassemia were included in these analyses as patients with other genotypes may not have the same degree of hematologic response to hydroxyurea²⁵. Also, participants hospitalized for >15% of their intervention period were excluded to avoid including patients who were hospitalized for complications that hydroxyurea may not improve (e.g., chronic pain) (Figure 1). Hydroxyurea dosing in both trials was at providers’ discretion and both sites had hydroxyurea dose escalation guidelines available.

Figure 1.

Study Participant Diagram

Data Collection

Participants’ demographic, hydroxyurea prescribing, and clinical and laboratory data were collected from the EMR. Laboratory parameters, including hemoglobin (Hb), mean corpuscular volume (MCV), absolute neutrophil count (ANC), and fetal hemoglobin level (HbF) at study entry and within 50 days of the end of the intervention period were abstracted from the EMR..

Hydroxyurea Exposure

Days that hydroxyurea administration was confirmed by VDOT, electronic pill bottle, or MAR were considered hydroxyurea exposure days. Days without confirmation or when administration was suspended by a participants’ provider were recorded as hydroxyurea non-exposure days. Participants’ mathematical average of their prescribed hydroxyurea dose at study entry and the end of the intervention period (in mg/kg/day), and their adherence, defined by their number of hydroxyurea exposure days divided by their number of days in their intervention period, were used to calculate their average daily hydroxyurea exposure as follows:

$$=\left(\text{Mathematical average of the prescribed dose at entry and the end of the study}\right)\times \left(\frac{\text{Number of hydroxyurea exposure days }}{\text{Number of days in the intervention period}}\right)$$

Participants with an average daily hydroxyurea exposure of ≤20mg/kg/day during their intervention period were included in the “lower exposure” group, and those with an average daily exposure >20mg/kg/day were included in the “higher” exposure” group. The 20 mg/kg/day cutoff between higher and lower exposure groups was chosen since 20 mg/kg/day was the fixed dose prescribed the pediatric hydroxyurea efficacy trial but the 2014 SCD guidelines recommend dose escalation if tolerated.^2,9 Notably, it was possible for participants who were prescribed a relatively high hydroxyurea dose (i.e., >25 mg/kg/day) to be assigned to the lower exposure group if they had low adherence and for patients who were prescribed relatively low hydroxyurea dose (i.e., 21 mg/kg/day) to be assigned to the higher exposure group if they had high adherence.

Statistical Analysis

Descriptive statistics were used to summarize participant demographics, hydroxyurea prescribing and exposure data, as well as hematologic outcomes. Nonparametric statistical methods were used to compare differences between study sites and the higher and lower exposure groups – chi-square or Fisher’s exact tests for qualitative variables and Wilcoxon rank-sum tests for quantitative variables. P-values <0.05 were considered statistically significant and analyses were completed using SAS software, version 9.4 (Cary, NC).

Results

Patient Characteristics

Forty-five participants were included in the analyses, 29 (64.4%) from NCH and 16 from LCH (35.6%) (Table I). The reason and number of excluded participants from these trials are shown in Figure 1. Mobile DOT participants were significantly younger than electronic pill bottle participants (mean age 10 vs. 15 years, P<0.01).

Table 1.

Participant Demographic and Clinical Characteristics

Characteristics	All participants n=45	Lower hydroxyurea exposure n=22	Higher hydroxyurea exposure n=23	p-value
Study site, n (%)
NCH	29 (64.4)	13 (59.1)	16 (69.6)	0.46
LCH	16 (35.6)	9 (40.9)	7 (30.4)
Age, median (range)	12 (2 – 19)	12 (2 – 19)	11 (2 – 18)	0.20
Age group, years n (%)
<10	15 (33.3)	7 (31.8)	8 (34.8)
10–17	25 (55.6)	11 (50.0)	14 (60.9)	0.43
≥18	5 (11.1)	4 (18.2)	1 (4.3)
Males, n (%)	25 (55.6)	13 (59.1)	12 (52.2)	0.64
Race, n (%)
Black	44 (97.8)	22 (100)	22 (95.7)	0.99
Other	1 (2.2)	0 (0)	1 (4.3)
Ethnicity, n (%)
Non-Hispanic	44 (97.8)	22 (100)	22 (95.7)	0.99
Hispanic/Latino	1 (2.2)	0 (0)	1 (4.3)
Genotype, n (%)
Hemoglobin SS	45 (100)	22 (100)	23 (100)	n/a
Insurance type, n (%)
Private	12 (26.7)	7 (31.8)	5 (21.7)
Public/Medicaid	18 (40.0)	9 (40.9)	9 (39.1)	0.63
Other	15 (33.3)	6 (27.3)	9 (39.1)
Duration of hydroxyurea use prior to study entry, n (%)
6–12 months	13 (28.9)	7 (31.8)	6 (26.1)	0.68
>12 months	32 (71.1)	15 (68.2)	17 (73.9)

Nationwide Children’s Hospital (NCH); Ann & Robert H. Lurie Children’s Hospital (LCH)

Participants with lower hydroxyurea exposure had a calculated hydroxyurea exposure of ≤20mg/kg/day during their intervention period;

Participants with higher hydroxyurea exposure had a calculated exposure >20mg/kg/day during their intervention period.

P-value < 0.05 was statistically significant (highlighted in bold)

Hydroxyurea Exposure

Mobile DOT participants had a higher proportion of hydroxyurea exposure days compared to electronic pill bottle participants (0.88 vs. 0.71, P=0.02), but electronic pill bottle participants were prescribed an average hydroxyurea dose that was higher than the Mobile DOT participants (27.1 mg/kg/day vs. 25.8 mg/kg/day, P=0.05) and had a 2median change in their prescribed hydroxyurea from the start to the end of the intervention period that was significantly higher than Mobile DOT participants’ (2.95 mg/kg/day vs. 0 mg/kg/day, P=0.003). Ultimately, average daily hydroxyurea exposure did not significantly differ between Mobile DOT and electronic pill bottle participants (median average exposure 20.5 mg/kg/day vs. 19.4 mg/kg/day, p=0.63). Table II shows the hydroxyurea prescribing and adherence characteristics of the lower and higher exposure groups. Median change in prescribed hydroxyurea dose from the start to the end of the intervention period was 1.4 mg/kg/day (range: −2.7 to 6.4) for the higher exposed group and 0.9 mg/kg/day (range: −14.6 to 12.7) for the lower exposed group was not statistically different between groups (P=0.99). While hydroxyurea exposure was highest among those who were prescribed a higher hydroxyurea dose and who also had high adherence, prescribed dose and adherence both influenced average hydroxyurea exposure (Figure 2).

Table 2.

Hydroxyurea Prescribing, Adherence, Exposure, and Participants’ Hematologic Parameters

	All participants n=45	Lower hydroxyurea exposure n=22	Higher Hydroxyurea exposure n=23	P-value
Length of intervention period in days, median (IQR)	209 (184 – 365)	207 (189 – 365)	209 (182 – 341)	0.68
Suspended Doses
Participants with any suspended dose(s), n (%)	11 (24)	8 (36)	3 (13)	0.14
Number of suspended doses, median (IQR)	0 (0–5)	0 (0 – 13)	0 (0 – 0)	0.04
Prescribing Characteristics
Average prescribed dose in mg/kg/day, median (IQR)	26.3 (23.7 – 28.0)	24.4 (20.4 – 26.5)	27.2 (26.0 – 28.2)	0.002
Adherence Characteristics
Number of hydroxyurea doses administered during the intervention period, median (IQR)1	176 (154 – 207)	157 (134 – 200)	189 (170 – 264)	0.02
Proportion of days with hydroxyurea exposure, median (IQR)	0.80 (0.7 – 0.94)	0.71 (0.55 – 0.79)	0.92 (0.80 – 0.95)	<0.001
Exposure Characteristics
Average hydroxyurea exposure in mg/kg/day, median (IQR)	20.2 (16.3 – 23.6)	16.1 (13.4 – 19.1)	23.6 (21.8 – 26)	<0.001
Hematologic Parameters
Baseline, median (IQR)
Hb (g/dL)	9.2 (8.3 – 9.8)	9.1 (8.3 – 9.6)	9.4 (8.7 – 10.3)	0.46
HbF (%)	26.8 (15.6 – 30.8)	23.1 (14.4 – 28.7)	28.8 (21.2 – 31.6)	0.04
MCV (fl)	94.3 (85.7 – 104.3)	90.3 (83.1 – 100.4)	98.1 (87.6 – 106.2)	0.14
ANC (cells/μL)	3.3 (2.6 – 4.6)	3.6 (3.0 – 5.0)	3.1 (2.2 – 3.7)	0.13
End of intervention, median (IQR)
Hb (g/dL)	9.4 (8.5 – 10.2)	9.3 (8.2 – 9.7)	9.5 (8.6 – 10.3)	0.21
HbF (%)	26.2 (17.3 – 33.6)	23.4 (10.7 – 27.4)	28.9 (22.9 – 37.4)	0.04
MCV (fl)	95.9 (89.7 – 105.2)	93.8 (85.7 – 99.3)	101.0 (90.9 – 112.3)	0.02
ANC (cells/μL)	3.2 (2.4 – 6.3)	3.4 (2.8 – 6.4)	3.0 (1.8 – 5.8)	0.30

¹Measured by video observation or EM and MAR

Participants with lower exposure had a calculated exposure of ≤20mg/kg/day during their intervention period

Participants with higher exposure had a calculated exposure >20mg/kg/day during their intervention period.

P-value < 0.05 was statistically significant (highlighted in bold)

Figure 2.

Participants’ Average Hydroxyurea Exposure

Hematologic Outcomes

HbF was significantly higher in the higher exposure group compared to the lower exposure group, though the mean HbF was >20% in both groups. MCV was also significantly higher in the higher exposure group compared to the lower exposure group, and this difference remained significant when analyzing the Mobile DOT participants separately. Five participants, four of whom had been prescribed hydroxyurea for >12 months prior to the study, had HbF <20%, despite having higher exposure. Seventeen participants, including eight in the higher exposure group, had an ANC >5.0 cells/μL at the end of the study (Table II).

Discussion

While there are an increasing number of treatment options available to patients with SCA, hydroxyurea remains the primary treatment for most patients, with over 30 years of evidence demonstrating its safety and benefits^26–30. However, there remain many important unanswered questions about the appropriate dosing regimen and methods to optimize adherence that limit effective hydroxyurea treatment in high-income clinical settings where dosing practices and adherence vary considerably. This study used reliable and validated adherence measures as well as EMR documented prescribing data to describe actual hydroxyurea exposure. We identified that higher hydroxyurea exposure is associated with having a median HbF that is ≥5% higher and a MCV that is ≥7 fL higher than having lower exposure. These favorable outcomes resulted from both higher adherence, as represented by the higher number of days of confirmed hydroxyurea exposure and a higher prescribed hydroxyurea dose. These findings support the need for adherence-promoting interventions. Given the number of medication adherence barriers among children with SCA,³² technology-based adherence interventions have the potential to address many of these barriers and improve adherence, not only to hydroxyurea, but also to other disease modifying therapies for SCA^33–35 to optimize health outcomes in this vulnerable underserved population of children and adolescents with SCA³⁶. However, the findings also support the need to optimize hydroxyurea prescribing in the real-world clinical setting, such as evaluating the safety and effectiveness of prescribing higher doses of hydroxyurea to less adherent patients, especially for patients whose medication adherence behavior is challenging to change.

While our results suggest increasing exposure improves hematologic outcomes in the high-income setting, larger studies that are powered to analyze the effect of hydroxyurea exposure on clinical outcomes are needed. These studies should consider including patients who are frequently admitted since they still can require acute visits for complications that hydroxyurea may ameliorate and VOC episodes that do not result in an acute care visit since VOC episodes are frequently treated at home in high-income settings⁴³. Furthermore, our definition of higher hydroxyurea exposure (>20mg/kg/day) was chosen since guidelines recommend initiating 20 mg/kg/day and many children in clinical practice receive less than this because they are non-adherent^6,9,11–16. However, this may be too conservative for subsequent trials that seek to determine if higher exposure improves outcomes in high-income settings since average exposure in the superior arm in the dose-escalation trial in Sub-Saharan Africa was 29 mg/kg/day¹⁰.

Notably, MCV, a parameter that increases quickly with hydroxyurea treatment², increased among both the lower and higher exposure groups. This may be attributable to increased hydroxyurea adherence from baseline among all participants since all were receiving an adherence intervention and the median prescribed hydroxyurea dose in both groups remained stable. Median ANC remained ≥3.0 μL for both groups during the study and the number of suspended doses for both groups was low. Also, more than a third of the participants had an ANC ≥5.0 μL at the end of the study. Since the 2014 guidelines suggest aiming for an ANC of ≥2.0 μL⁹ and younger patients may tolerate an ANC of ≥1.25 μL, it is likely that patients in both groups may have been able to tolerate even higher hydroxyurea exposure. If larger subsequent studies confirm that higher exposure optimizes clinical outcomes in high-income settings, quality improvement initiatives to ensure that dose escalation occurs when warranted will likely be needed.

Overall, the end of study HbF was relatively high in both exposure groups (mean 23.4% and 28.9%). Half of participants in the lower exposure group achieved a HbF >20%, but five higher exposed participants did not achieve a HbF ≥20% suggesting that some participants may benefit with relatively low hydroxyurea exposure, while others may require much higher hydroxyurea exposure to achieve optimal benefit. This is consistent with the known inter-patient variability in hydroxyurea dose and response, possibly related to age, absorption patterns, or genetic factors^37–40. Careful attention to hydroxyurea exposure, using novel strategies, such as early hydroxyurea initiation and/or individualized pharmacokinetic-guided dosing, in combination with attention to medication adherence, may optimize clinical benefit for patients and requires further prospective clinical investigation^41,42. Furthermore, identifying patients who have limited hematologic response, despite high exposure is important as these individuals may be erroneously labelled as non-adherent but may need additional or alternative therapies to optimize their outcomes.

Our study has several limitations. First, we used gold-standard adherence methods, but the electronic pill bottles could not verify if participants actually ingested. Also, neither trial confirmed that the doses that were ingested were the doses that were prescribed nor could account for ingested doses that may not have been observed or contained within the bottles. Second, the method used for calculating hydroxyurea exposure does not account for variability in medication absorption and metabolism that can affect exposure³⁷ and does not fully account for dose modifications and the timing of these adjustments that may have occurred during the intervention period. However, we suspect these adjustments are unlikely to significantly impact the findings since the median change in prescribed hydroxyurea dose from the start of the intervention period to the end was small and not statistically different between groups. Finally, while both studies were designed to be pragmatic trials, both were adherence promoting trials and therefore, differences in the clinical care that participants received as part of these trials, rather than differences in hydroxyurea exposure, may have influenced their outcomes.

In conclusion, this study highlights the importance of evaluating both prescribed dose and medication adherence, as higher hydroxyurea exposure is associated with improved hematologic outcomes, and larger studies are warranted to determine if increasing hydroxyurea exposure through prescribing and adherence interventions have the potential to improve a range of clinical outcomes.

Abbreviation Full Term

SCA

Sickle Cell Anemia

VOC

Vaso-Occlusive Crises

ACS

Acute Chest Syndrome

ED

Emergency Department

MTD

Maximum Tolerated Dose

SCD

Sickle Cell Disease

Mobile DOT

Mobile Directly Observed Therapy

NCH

Nationwide Children’s Hospital

VDOT

Video Directly Observed Therapy

MAR

Medical Administration Record

EMR

Electronic Medical Record

LCH

Ann & Robert H. Lurie Children’s Hospital of Chicago

Hb

Hemoglobin

MCV

Mean Corpuscular Volume

ANC

Absolute Neutrophil Count

HbF

Fetal Hemoglobin Level