Abnormal transcranial doppler ultrasound (TCD) velocity is a risk factor for stroke in children with sickle cell anemia (SCA). The STOP trial demonstrated that transfusions reduce TCD velocity and stroke risk.¹ However, neither STOP¹ nor STOP II² results reported the unit decline in hemoglobin S (HbS) fraction per unit change in TCD velocity. In a pooled analysis of 10 studies demonstrating the relationship between hydroxyurea and TCD measurements, the average drop in TCD velocity was 21 cm/s (95% confidence interval [CI], 14.8-29.0)³ after hydroxyurea therapy. The pooled analysis did not demonstrate any change in the HbS fraction per unit change in TCD velocity. However, Hurlet-Jensen and colleagues demonstrated that a decline in HbS fraction has a more significant impact on cerebral blood flow (ie, rate of blood delivery to tissue; mL/100 g/min) than an increase in the total hemoglobin level.⁴ Here, we tested the hypothesis that HbS fraction has a more significant impact on the TCD velocity than total hemoglobin in individuals with SCA.

The SIT trial participants included children aged 5-15 years at the time of enrollment, with a confirmed diagnosis of hemoglobin SS or hemoglobin Sβ⁰ thalassemia; time-averaged maximum mean TCD velocity of the terminal portion of the internal carotid or the proximal portion of the middle cerebral artery <200 cm/s by non-imaging technique, or time-averaged maximum velocity <185 cm/s by imaging technique; and no prior history of overt strokes or seizures. Inclusion criteria for this secondary analysis: (a) SIT trial participant randomized to regular transfusions; (b) completed at least 10 transfusions; (c) at least two TCD exams completed, with the first TCD before the start of regular blood transfusion; (d) and the second TCD performed near to the time of transfusions; and (e) HbS at baseline and at the time of follow-up TCD. Exclusion criterion: any neurological event during the SIT trial. The trial was approved by the Institutional Review Boards (IRB) at all participating institutions and registered at https://clinicaltrials.gov/ (NCT00072761). Informed consents were performed per local IRB procedures. Participant characteristics were described by median and interquartile range (IQR) for continuous measures and by number and percent for categorical measures. Multivariable mixed regression models of longitudinal change in TCD were constructed, with age and time as covariates, and separate models for hemoglobin S or total hemoglobin because of sample size. Based on the evidence that decreasing HbS fraction and increasing hemoglobin levels are associated with a decrease in TCD velocities,⁵ a two-sided P value ≤.05 was considered statistically significant. Data analysis was performed using SPSS, version 26.0 (IBM, Armonk, New York).

From the randomized SIT cohort of 99 children who received approximately monthly blood transfusions, 50 completed at least 10 transfusions and had at least two TCD evaluations. Of these 50, 24 had paired TCD measurements, with one excluded for a neurological event and one excluded for missing HbS value, thus 22 had the first TCD measurement before transfusion (median 2.2 months, IQR 0.8-3.7), and the second TCD measurement completed within a median −0.1 month (IQR −0.6 to 0.1 months) of the last transfusion. The median HbS fraction of 20.5% at the time of the second TCD measurement provides evidence that the participants had reached a steady-state of regular blood transfusion therapy when the goal was to keep the maximum HbS percentage less than 30%. The median interval between the first and the last TCD was 2.5 years.

After initiating transfusion therapy, TCD assessment was not performed as part of the trial protocol because the TCD measurement after initiating transfusion has not been associated with a future incidence rate of ischemic strokes.⁶ All 22 children had HbS fractions available at the time of TCD examinations, and later when receiving regular blood transfusion therapy based on institutional practice. No statistically significant differences were observed between participants who completed at least 10 transfusions (n = 22) and participants with at least two TCD examinations but were not eligible for this analyses due to at least one exclusion criteria (n = 23), Table S1. As expected, with transfusion, the median TCD velocity decreased from a median of 142.5 cm/sec to 112.0 cm/sec (P < .001) while the median HbS fraction decreased from 89.0 to 20.5 (P < .001) and the median hemoglobin levels increased from 7.5 g/dL to 9.4 g/dL (P < .001), Table 1. One participant that was excluded from the analyses had a paired TCD measurement before and after transfusions and developed an overt stroke. The participant with a stroke had TAMMV TCD measurement before regular blood transfusion that was initially 141 cm/s and only decreased 7 cm/s, to 134 cm/s; the hemoglobin increased from 7.5 g/dL to 9.3 g/dL and the HbS fraction went from 91.6% to 25.0% over 1.63 years.

A mixed multivariable linear regression model, with random intercepts, controlling for age at TCD and times between TCD measurements, Table S2, found a 4.0 cm/s reduction (95% CI 1.3-6.6) in TCD velocity for every 10% reduction in HbS% (P = .004). Thus, clinically when the HbS fraction is reduced from 90% to 20%, the minimum expected decline in TCD velocity is 9 cm/s (−70%/10 × 1.3 cm/s) and the median expected decline of 28 cm/s (−70%/10 × 4 cm/s). A separate mixed model found a 4.6 cm/s reduction (95% CI −9.78 to 0.54) in TCD velocity for every 1 g/dL increase in hemoglobin (P = .078). Due to the small sample size, we were unable to include both the fraction of HbS and total hemoglobin in the same model. The small sample size may have contributed to the absence of a statistical association between change in total hemoglobin level and TCD velocity. Given the strong evidence that increasing hemoglobin levels are expected to decrease TCD measurements, a 1-tailed instead of a 2-tailed test could have been applied. The 1-tailed t test would have been statistically significant, demonstrating a relationship between increasing hemoglobin levels and P value of .039.

Similar to Hurlet-Jensen et al.,⁴ the current study demonstrated a significant relationship between HbS fraction and TCD velocity. However, Hurlet-Jensen used the xenon inhalation method, an experimental measure to assess cerebral blood flow in three participants with serial evaluations of HbS fraction after cessation of regular blood transfusion therapy.⁴ In contrast, our study used TCD velocity, a standard clinical test, and included 22 participants with serial assessment of HbS fraction and total hemoglobin levels after starting regular blood transfusion therapy for a median of 2.5 years.

The optimal decline of TCD velocity after blood transfusion therapy, if any, that confers a low risk of future strokes after starting regular blood transfusion therapy is unknown. Typically, TCD velocity is not obtained after starting regular blood transfusion therapy because, as in the STOP II Trial, TCD velocity obtained after receiving regular blood transfusion therapy was not predictive of future ischemic strokes.⁷ Potentially, comparing the expected to the observed decline of TCD velocities after starting regular blood transfusion or hydroxyurea therapy may help determine if there is an ongoing risk for ischemic strokes. However, in children with SCA and abnormal TCD measurements, ischemic strokes are rare after starting regular blood transfusion therapy (0.24 per 100 patient-years [95% CI 0.18-0.31]).⁷ Despite the low incidence rate of ischemic strokes with abnormal TCD measurements after blood transfusion therapy, in a large population of children with SCA, ischemic strokes will occur.⁷ A limitation in our study is the small sample size of 22 paired TCD measurements before and after regular blood transfusion therapy for at least a year. Despite this limitation, our study is one of the first to compare cerebral hemodynamic measurements prior to regular blood transfusion therapy and after the goal of transfusion therapy is reached, a hemoglobin S fraction less than 30%. In children with SCA receiving regular blood transfusion, there is a high biological association between HbS and total hemoglobin levels. Thus, both HbS and total hemoglobin levels were not included in the same multivariable regression model. The inability to include both independent variables in the same multivariable regression model means that we cannot measure the effect of HbS while controlling for total hemoglobin levels.

As a result of the collaborative efforts of our team with large cohorts of children receiving primary stroke prevention therapies (DISPLACE trial NCT04173026, and SPRING trial, NCT02560935), in future studies, we will be able to test the hypothesis that the magnitude of the drop in TCD velocities after starting blood transfusion or hydroxyurea therapy predicts future strokes.

CONFLICT OF INTEREST

M.R.D. and his institution are the sponsors of two externally funded research investigator-initiated projects. Global Blood Therapeutics (GBT) is providing funding for the cost of the clinical studies but will not be a cosponsor of either study. M.R.D. is not receiving any compensation for the conduct of these two investigator-initiated observational studies. He is a member of the GBT advisory board for a proposed randomized controlled trial for which he receives compensation, and he is the chairperson of a steering committee for an industry supported phase two trial (SPARTAN) for prevention of priapism in SCD.

M.D. receives research support from the American Heart Association, National Institutes of Health (NINDS, NIA, and NINR), and Lipedema Foundation. He also receives research-related support from Philips Healthcare, and consulting or advisory board payments from Global Blood Therapeutics, bluebird bio, and Pfizer.

AUTHOR CONTRIBUTIONS

Lori C. Jordan, Mark Rodeghier, Manus J. Donahue, and Michael R. DeBaun wrote the paper; Mark Rodeghier performed the statistical analyses, Michael R. DeBaun conceived the project and approach.