Hydroxyurea Initiation and Occurrence of Cerebrovascular Outcomes in Sickle Cell Disease

Background: Cerebrovascular disease (CVD) is a common cause of morbidity in sickle cell disease (SCD). Silent cerebral infarct (SCI) impacts ~37% of children by age 14, leading to increased stroke risk and neurocognitive impairment. Hydroxyurea (HU) protects against stroke in patients who have developed abnormal transcranial Doppler (TCD) ultrasound velocities. However, the neurologic impact of starting HU before onset of SCD-CVD is not well defined. We hypothesize that HU protects against initial development of CVD.

Methods: We performed a single center, retrospective cohort study of the impact of HU initiation on incident CVD, defined as first occurrence of: abnormal TCD, brain magnetic resonance imaging (MRI) with SCI, MR angiography (MRA) with moderate or severe cerebral vasculopathy, or overt ischemic stroke. Incident HU was defined as the first outpatient HU prescription in subjects with a minimum of two prescriptions. Subjects with SCD-SS or Sb⁰ who initiated hematology care before age 1 between 2006 and 2022 were included. Subjects with known CVD or non-SCD neurologic disease were excluded. Subjects were followed from age 9 months until the earliest of incident CVD, loss to follow up, curative therapy, initiation of chronic red cell transfusions, or 10 years. All outcomes were manually reviewed.

Primary analysis examined the relationship between timely HU (initiation by age 2 years), and CVD, as well as any HU and CVD. HU was modeled as a time-varying exposure to account for immortal time bias. Analyses were stratified by birth year before versus in or after 2013, given significant changes in MRI screening frequency and HU prescribing practices recommended by SCD guidelines published in 2014.

Marginal structural models were implemented using 1) time-varying inverse probability of treatment weights to account for confounding by HU indication and 2) time-varying inverse probability of censoring weights to account for differential censoring. Baseline covariates included year of birth, sex, SCD genotype, and insurance category. Time-varying covariates included hospitalization for acute chest syndrome or splenic sequestration, degree of anemia, and diagnosis of obstructive sleep apnea.

Results: A total of 324 patients met inclusion criteria, of which 148 were born before 2013 (“early cohort”) and 176 were born in 2013 or later (“late cohort”). Most (96%) had SCD SS, and 50% were male. Median duration of follow-up was 5.9 years (IQR: 2.3-10).

Compared to subjects in the early cohort, those in the late cohort had higher rates of any HU (120/176, 68%, vs. 84/148, 57%) and timely HU (83/176, 47%, vs. 5/148, 3%). Median age at HU initiation was lower in the late cohort (1.2y, IQR [0.9, 2.5] vs 5.7y, IQR [3.8, 7.7]). Most patients with HU exposure (180/204, 88%) demonstrated laboratory evidence of HU at least once with a 10% rise in mean corpuscular volume and/or fetal hemoglobin.

Most patients in the early cohort (124/148, 84%) had at least one brain MRI/MRA, with similar rates in those prescribed HU vs. not. In contrast only 21% in the late cohort had MRI/MRA screening (37/176), with more HU-exposed patients screened (27/120, 23%) compared to those not exposed (11/176, 6%).

Over 1,770 person-years of follow-up, 44 patients developed CVD (2.5/100py). The most common outcomes were SCI (34/44) and abnormal TCD (6/44). In adjusted analysis, any HU initiation was significantly associated with decreased CVD development in the early cohort (HR 0.23, 95% CI: 0.06, 0.87, p = 0.03). We did not examine the impact of timely HU by age 2 in the early cohort due to the rarity of the exposure. HU initiation, regardless of timeliness, was not associated with a significant change in CVD in the late cohort. Sensitivity analyses requiring laboratory evidence of HU effect showed similar results.

Discussion: HU initiation was associated with a decrease in incident CVD, including SCI, in patients with SCD born 2005-2012. This relationship was independent of clinical factors influencing the likelihood of HU initiation. These data strengthen the recommendations for early and universal HU. Our novel findings, though preliminary, provide a therapeutic strategy for primary SCI prophylaxis. We were unable to confirm these results in a more recent cohort given decreased MR screening frequency. Prospective, multi-center studies are needed to determine optimal MR screening frequency and further explore approaches to primary CVD prevention.