Cerebral hemodynamics and metabolism are similar in sickle cell disease patients with hemoglobin SS and Sβ0 thalassemia phenotypes.

Abstract

Sickle cell disease (SCD) represents a group of genetically well-characterized hemoglobinopathies and leads to systemic morbidity frequently manifesting as pain crises, organ failure, and cerebrovascular disease (CVD).1 Sickle cell disease comprises a range of sickle phenotypes (HbSC, HbSS, HbSβ0 HbSβ+, HbSE and HbSD), with the most common, sickle cell anemia (SCA), characterized by the presence of hemoglobin (Hb)-S (HbSS). A less common but still frequently observed SCD phenotype is sickle beta-thalassemia null (HbSβ0). Both HbSS and HbSβ0 are considered similar in their potential for severe clinical manifestations, resulting in randomized trials often including both phenotypes as entry criteria.2 However, it is unclear whether HbSS and HbSβ0 confer similar risk of neurological injury. In the Cooperative Study of Sickle Cell Disease,3 the cerebrovascular event incidence was higher in individuals with HbSS compared to HbSβ0. In a more recent study among children with HbSβ0 (n = 22) or HbSS (n = 786), total hemoglobin was statistically higher for HbSβ0 (9.2 g/dL) vs HbSS (8.1 g/dL) individuals and transcranial Doppler (TCD) velocities statistically lower for HbSβ0 (112.6 cm/s) vs HbSS (135.6 cm/s) individuals. Yet the prevalence of cerebral infarcts was not statistically different for HbSβ0 (27.8%) vs HbSS (30.8%) individuals.4 These data collectively suggest that some hematological differences in phenotype may not manifest as tissue injury. To gain additional information, it is necessary to consider how hematological effects may influence tissue-level cerebral hemo-metabolic physiology. In individuals with SCD, autoregulatory increases in cerebral blood flow (CBF; rate of blood delivery to tissue) develop to maintain adequate oxygen delivery despite anemia and reduced oxygen carrying capacity. When insufficient, or in the presence of cerebral vasculopathy, oxygen extraction fraction (OEF) may increase, or, the cerebral metabolic rate of oxygen consumption may reduce. However, these studies have not considered different phenotypes. Here, we measure CBF and OEF in patients with HbSS or HbSβ0 and evaluate their association with indicators of disease severity. All volunteers provided informed, written consent and were comprised of adults and children with SCD (phenotype HbSS or HbSβ0) ages 6-40 years recruited sequentially from a comprehensive SCD clinic. Patients received oral hydroxyurea and/or transfusion; transfusion patients were scanned at least 3 weeks since their prior transfusion when hematocrit was near nadir. Hemoglobinopathy diagnosis was determined by high performance liquid chromatography. As is most common in clinical practice, genotyping was not performed and as such we refer consistently to SCD phenotype rather than SCD genotype. Magnetic resonance imaging (MRI) and angiography (MRA) were performed at 3.0 T (Philips Healthcare, Best, The Netherlands) using an anatomical head MRI/MRA protocol for prior infarct and vasculopathy determination. Also pseudo-continuous arterial spin labeling (pCASL)-MRI was used for CBF determination and T2-relaxation-under-spin-tagging (TRUST)-MRI for OEF determination. Physiological monitoring (In Vivo Research, Inc., Orlando, FL, USA) included arterial oxygenation saturation (Ya) via pulse oximetry, heart rate, and blood pressure. Major cervical and intracranial vessels for each participant were assessed for vasculopathy from MRA, and infarct determination from FLAIR (two-planes) and T1-weighted MRI, independently by two board-certified radiologists. CBF was quantified from the pCASL data utilizing a two-compartment model that accounts for differences between blood and tissue relaxation times, with participant-specific arterial blood longitudinal relaxation times (T1) calculated from measured hematocrit, and a SCD pCASL labeling efficiency of 0.72.5 For OEF quantification, T2 values from TRUST-MRI were converted to venous oxygen saturation (Yv) and subsequently OEF according to OEF = (Ya-Yv)/Ya. Due to participants having different hemoglobin phenotypes and different hemoglobin calibration curves being available, for completeness OEF is reported for Yv calculated from an HbSS model, HbAA model, fetal hemoglobin (HbFF) and bovine blood (Hb-bovine) model (see Bush et al.6 and references therein). Descriptive statistics including mean, SD, and median for continuous variables were calculated, along with percent for categorical variables. Group differences using bivariate analysis were evaluated using either an unpaired t test for continuous variables or a χ2 test for categorical variables. To understand whether hematocrit differed between groups after accounting for sex, we performed a linear regression using hematocrit as the dependent variable and (i) biological sex and (ii) hemoglobin phenotype as the explanatory variables. As treatment regimen will influence hematocrit, we utilized only the subgroup of patients that were medically managed on hydroxyurea (eg, excluding all participants on chronic blood transfusion) for this analysis, which yielded a remaining sample of 89 participants. Finally, two separate linear regressions were performed on data from 120 participants, which utilized either CBF or OEF as the dependent variable and (i) age, (ii) biological sex, (iii) hematocrit, and (iv) hemoglobin phenotype as the explanatory variables. Additional supplementary analyses that utilize subgroups of data, and also include additional exploratory co-variates, are included in Supplementary Material. In all analyses, significance criterion was two-sided P < .05. Additional study procedure details are provided in the Supplementary Methods. Of 120 participants, 13 (10.8%) had HbSβ0 variant and 107 (89.2%) HbSS hemoglobin variant. For HbSβ0 participants, 12 (92.3%) received hydroxyurea and one (7.7%) received chronic transfusion. For HbSS participants, 77 (72.0%) received hydroxyurea and 30 (28.0%) received chronic transfusions. No parameters were significantly different between groups on bivariate analysis (Table 1). When controlling for treatment, in the subgroup (n = 89) of participants on hydroxyurea only (n = 77 HbSS; n = 12 HbSβ0), no significant dependence of hematocrit on SCD phenotype (P = .23) after controlling for sex (P = .21) was observed using linear regression. When considering the possible dependence of CBF on explanatory variables, CBF was inversely related to hematocrit (P < .001) with no significant relation to age (P = .46), sex (P = .92), or SCD phenotype (P = .93). When considering the possible dependence of OEF on explanatory variables, results were similar when using the HbAA, HbFF, and Hb-bovine calibration models, which found that OEF was inversely related to hematocrit (P ≤ .012 in these models) and directly related to age (P < .001 in these models). Neither sex (P-value range across these models = 0.13-0.15) nor hemoglobin phenotype (P-value range across these models = 0.14-0.15) were significantly related to OEF. By contrast, when the HbSS model was used, OEF was found to be directly related to hematocrit (P = .049), and, similar to other models directly related to age (P < .001) and unrelated to SCD phenotype (P = .15) and sex (P = .48). Full and additional regression analyses are summarized in Supplementary Materials. As such, we evaluated 120 participants with SCD phenotypes HbSS or HbSβ0 followed in our SCD clinics between 2015 and 2019 with anatomical, angiographic, and hemo-metabolic neuroimaging to understand whether either of these hemoglobin phenotypes had any significant impact on CBF or OEF. No significant dependence of either CBF or OEF on hemoglobin phenotype was found. These findings are largely consistent with a prior study of HbSβ0 (n = 22) and HbSS (n = 786) pediatric patients, which included genotyping whereby a lower intracranial TCD velocity was observed in HbSβ0 vs HbSS participants, but with no significant impact on infarct prevalence.4 Certain individuals with intracranial stenosis would be expected to have elevated TCD velocities and as such this prior work is consistent with no patients here with HbSβ0 having vasculopathy. Additionally, in this prior study there was an observation of slightly elevated hematocrit in the HbSβ0 cohort compared to the HbSS cohort, which was observed here as well although our finding did not meet criteria for significance. These findings may be collectively interpreted as while subtle or negligible differences in hematocrit or vasculopathy may exist between these cohorts, these differences have limited discriminatory impact on cerebral hemo-metabolic health or infarct risk. We did observe significant relationships between both CBF and OEF and hematocrit. As oxygen carrying capacity reduces due to anemia, CBF will increase and cerebrovascular reserve will reduce in an attempt to maintain a constant CMRO2 and OEF. No dependence of CBF or OEF on hemoglobin phenotype was observed however, and importantly, all OEF calibration models yielded no significant dependence of OEF on SCD phenotype (however mean OEF values did differ, as expected). Limitations of this work include that participants were not genotyped. We used available chromatography information to classify patients. While some misclassification may exist, this procedure is representative of clinical practice. Second, this study included 120 SCD participants, with 13 HbSβ0 participants. While this fraction approximately reflects the fraction that would be expected in typical North American SCD clinics, larger sample sizes may elucidate more subtle differences. A post-hoc power calculation revealed that using the measured differences found here (Table 1), 650 and 140 patients per group would be required for the observed CBF and OEF differences, respectively, to be significant at two-sided P = .05 with 80% power. Finally, we did not measure dyshemoglobins in this study, which may also account for a small fraction (generally <5%) of total hemoglobin. In conclusion, we performed quantitative measures of CBF and OEF in 120 SCD participants with differing hemoglobin phenotype. Results suggest that both CBF and OEF are not significantly different between these SCD phenotypes, and also on bivariate analysis that conventional risk factors such as hematocrit, infarct, and vasculopathy were not different between cohorts. These results provide added support that randomized trials should consider including both phenotypes as entry criteria. Funding for this work has been provided by the National Institutes of Health (NIH)/National Institute of Neurological Disorders and Stroke (NINDS) and American Heart Association: NIH/NINDS 5R01NS096127, NIH/NINDS 5R01NS078828, American Heart Association 14CSA20380466, and NIH/NINDS 5R01NS097763. Manus J. Donahue is the CEO of biosight consulting; provides paid consulting and/or advisory board services to Global Blood Therapeutics, bluebird bio, and Pfizer Inc; and receives research-related support from Philips North America. Appendix S1: Supplementary Methods Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.