Cerebral hemodynamics and oxygen metabolism in patients with milder and severe forms of sickle cell disease and thalassemia

Abstract

Silent cerebral infarcts (SCIs) are present in patients with sickle cell disease (SCD) and thalassemia, but the pathophysiology of SCIs is not fully understood. Previous studies mainly focused on cerebral hemodynamics and oxygen metabolism in patients with severe SCD (HbSS/HbSβ°) but not in milder forms of SCD (HbSC/HbSβ⁺) and thalassemia despite the high prevalence of SCIs in these patients. In this work, we studied the cerebral hemodynamics and oxygen metabolism, and SCI lesion load in 75 severe and 26 mild adult SCD patients, 18 thalassemia patients (as anemic comparison group), and 30 healthy controls before and after a vasodilatory challenge with acetazolamide. Cerebral blood flow was significantly higher in patients with severe SCD and thalassemia compared to patients with mild SCD and controls (p < 0.05). Conversely, oxygen extraction fraction and cerebral metabolic rate of oxygen (CMRO₂) were significantly lower in patients with severe SCD and thalassemia compared to other groups (p < 0.01). In contrast, no difference in SCI volumes was found between mild and severe SCD and thalassemia patients. After acetazolamide administration, oxygen delivery increased less in severe SCD and thalassemia patients compared to other groups (p < 0.01) and CMRO₂ decreased only in severe SCD patients (p < 0.01). Given the reduced CMRO₂ values in severe SCD and thalassemia patients, we conclude that reduced cerebral oxygen consumption in these patient groups is mostly related to anemia. Our data suggest that the pathophysiology of SCIs in patients with milder forms of SCD might be more related to prior episodes of anemia or other sickle cell‐related factors.

INTRODUCTION

Sickle cell disease (SCD) is an inherited form of hemoglobinopathy resulting in hemolytic anemia, recurrent vaso‐occlusions, multi‐organ damage, and a reduced life expectancy. ¹ , ² The brain is particularly vulnerable, with a silent cerebral infarct (SCI) prevalence of up to 37% in children, which further increases with age. ³ , ⁴ , ⁵ Although SCIs do not cause overt neurological symptoms, they are associated with learning disabilities and lower quality of life. ⁶ To date, the pathophysiology of SCIs in SCD is not fully understood, but several mechanisms have been postulated. First, anemia has been suggested to cause SCIs as a result of local cerebral ischemia, especially in the deep watershed areas. ³ , ⁷ Second, local vaso‐occlusions in the small vasculature, caused by adhesive leukocytes and sickled red blood cells (RBCs), may contribute to the formation of SCIs. Finally, oxygen offloading may be limited by functional shunting, defined as the rapid transit of blood through the capillaries caused by increased cerebral blood flow (CBF) while leaving vulnerable areas of the brain with inadequate oxygen supply. ⁸ , ⁹

To get more insight into the pathophysiology of SCIs, magnetic resonance imaging (MRI) has been used to investigate cerebral hemodynamics and oxygen metabolism in patients with SCD. ¹⁰ , ¹¹ , ¹² Increased CBF has been reported in patients with severe forms of SCD (sickle cell anemia [HbSS] and HbSβ°‐thalassemia), to compensate for lower oxygen delivery caused by anemia. ¹² , ¹³ , ¹⁴ Efficient oxygen delivery and extraction of oxygen are important for sustaining a proper rate of cerebral metabolism. Any alterations in these factors can impact cerebral oxygenation, and consequently, may influence cerebral functionality. Although whole‐brain (WB) oxygen delivery is normal in SCD, multiple studies have reported altered oxygen metabolism in these patients, including oxygen extraction fraction (OEF) and cerebral metabolic rate of oxygen (CMRO₂). ¹¹ , ¹⁵ , ¹⁶ Additionally, studies have reported that CBF increases less after a vasoactive stimulus in patients with SCD compared to healthy controls. ¹⁰ , ¹²

Most of the studies investigating cerebral hemodynamics and oxygen metabolism in SCD patients have been performed in patients with severe genotypes as these patients have more vascular disease and SCIs. ¹⁷ , ¹⁸ However, SCIs are also present in SCD patients with milder genotypes, such as sickle‐hemoglobin C disease (HbSC) and HbSβ⁺‐thalassemia. ⁶ , ¹⁹ , ²⁰ We recently analyzed cerebral hemodynamics in these patients and found significantly lower CBF values in patients with milder SCD compared to severe forms, which were even comparable to CBF values found in healthy controls. ¹⁰ These results suggest that CBF alone cannot explain the high prevalence of SCIs in the relatively mild SCD genotypes. Interestingly, SCIs have also been detected in patients with thalassemia, a hereditary form of anemia that is not characterized by vaso‐occlusion. SCIs have been detected in up to 60% of the patients with thalassemia with hypercoagulability and anemia as the potential risk factors for SCIs. ²¹ , ²² , ²³ , ²⁴ To better understand the pathophysiology of SCIs in these different forms of hereditary anemia, it is important to examine the brain perfusion, along with cerebral oxygen supply and consumption in patients with mild and severe SCD and thalassemia patients.

In the present study, we assessed cerebral hemodynamics and oxygen metabolism in patients with mild and severe forms of SCD, thalassemia patients, and healthy controls. Given the milder clinical complications and the higher hemoglobin levels in patients with mild SCD, we hypothesized that in these patients, OEF and CMRO₂ would be less affected compared to those with severe SCD and thalassemia. Although thalassemia patients are also known to have anemia‐related organ damage, patient with thalassemia do not suffer from recurrent vaso‐occlusion of the microvasculature like severe SCD patients, but have poorer clinical outcomes compared to mild SCD patients. Therefore, we expected their OEF and CMRO₂ values to fall between those of the mild and severe SCD groups. To study SCI among the participant groups, we assessed the presence and volumes of SCIs in each subject and compared the results across groups. Our hypothesis was that smaller SCIs would be detected in patients with mild SCD and thalassemia compared to those in severe SCD, and that the healthy control group would have the least SCIs compared to the other groups. The effect of a vasodilatory stimulus on cerebral circulation was studied to test the impact of increased CBF on cerebral oxygen metabolism. We hypothesized that upon vasoactive stimulus, OEF and CMRO₂ would decrease in patients with mild SCD but to a lesser extent than in those with severe SCD, and that CMRO₂ in healthy controls would remain unaffected.

METHODS

Demographics

The local medical ethics committee approved the study at the Amsterdam University Medical Centers (UMC) in The Netherlands and Children's Hospital Los Angeles (CHLA) in the United States. The study was performed in accordance with the Declaration of Helsinki and all participants provided written informed consent. SCD patients with the genotypes HbSS, HbSβ⁰‐thal, HbSβ⁺‐thal, and HbSC were recruited, and they were divided into two groups based on the severity of their disease. The severe group consisted of patients with genotypes HbSS and HbSβ⁰‐thal, while the mild group included patients with genotypes HbSβ⁺‐thal and HbSC. In addition, thalassemia patients and healthy age and ethnicity‐matched controls with the genotype HbAA, HbAS and HbAC were included. All participants were 16 years or older and were recruited between 2018 and 2022. Exclusion criteria were MRI or acetazolamide (ACZ) contraindications, history of diagnosed stroke or cerebral pathology, vasculopathy or arterial stenosis, sickle cell crisis at inclusion, hospitalization up to one month before the study day, pregnancy, or breastfeeding.

Data acquisition

Before MRI, blood was drawn to determine hematological laboratory parameters including hemoglobin, hematocrit, reticulocytes, and fetal hemoglobin (HbF). MRI scans were acquired on a 3.0 T MRI scanner (Philips Medical Systems; Best) with a 32‐channel head‐coil and body‐coil transmission. The scan parameters of 3D fluid‐attenuated inversion recovery (FLAIR), T1‐weighted anatomical, time‐encoded pseudo‐continuous arterial spin labeling (te‐pCASL), M0, 2D phase‐contrast scans, and T₂‐relaxation‐under‐spin‐tagging (TRUST) are reported in Supporting Information S1: Section 1.1. The functional scans such as te‐pCASL, M0, 2D phase‐contrast, and TRUST were performed before and 10 min after ACZ administration (16 mg/kg i.v., max 1400 mg).

Data analyses

For the SCI volumes, two observers (SC&SS) manually marked the centroids of white matter (WM) hyperintensities in the FLAIR images and then counted them to obtain SCI count. The centroid of SCIs was used to identify the boundaries of the SCIs and to create masks using an automated tool previously published. ²⁵ SCI masks were manually checked by two other observers (KB&LAH) and manually corrected. The segmented SCI maps were then used for SCI volume calculations. The post‐processing pipeline used for WM, gray matter (GM) and WB CBF is explained in Supporting Information S1: Section 1.2. For OEF measurements, venous blood T₂ values were derived from the TRUST scan and converted to venous oxygenation (Yv) using the calibration models reported in Supporting Information S1: Section 1.3. This section also provides the equations for OEF, oxygen delivery, and CMRO₂ calculations.

To test the effect of ACZ as a vasoactive stimulus, we calculated the differences between the baseline WM and GM‐CBF and oxygen delivery, OEF, and CMRO₂ and the corresponding post ACZ values.

Statistical analyses

Statistical analyses were performed in SPSS v28 (IBM) and p < 0.05 was considered statistically significant. Analysis of variance (ANOVA) or a Kruskal–Wallis test (in case of non‐normality) was used to compare the clinical parameters between the groups. A one‐way analysis of covariance (ANCOVA) or Quade's test (in case of non‐normality) was conducted with post hoc Bonferroni correction to test for statistically significant differences between the participant groups on WM and GM CBF and oxygen delivery, OEF, CMRO₂ and SCI volume controlling for age and sex. To determine if CMRO₂ differed between groups independently of hemoglobin levels, we used the residuals from the regression between hemoglobin and CMRO₂ to analyze the differences. For the ANCOVA tests between the groups on ∆WM and ∆GM‐CBF and oxygen delivery, ∆OEF, and ∆CMRO₂, their respective baseline values were added as an additional covariate. Paired sample t‐tests or Wilcoxon tests were used to study the differences in imaging parameters before and after ACZ. In case the patients received treatment, independent t‐tests or Mann–Whitney tests were conducted to test the statistically significant differences in imaging parameters between SCD patients with and without treatment. The same tests were conducted to determine a statistically significant difference between patients with and without SCIs on imaging parameters. To get more insight into the effect of HbF, we studied the correlations between HbF and laboratory and imaging parameters in the severe SCD group using correlation analysis.

RESULTS

Among 154 participants that were included, 103 were recruited from Amsterdam UMC, Amsterdam, the Netherlands while 51 were recruited from CHLA. One patient with severe SCD, two patients with thalassemia, and one healthy control were excluded from the study due to incomplete data. One patient with severe SCD was excluded because of a complete occlusion of the right MCA detected on magnetic resonance angiography (MRA). The remaining sample consisted of 75 patients with severe SCD (HbSS and HbSβ⁰‐thal), 26 patients with mild SCD (HbSβ⁺‐thal and HbSC), 18 patients with thalassemia, and 30 healthy age and ethnicity‐matched controls. The demographical and clinical parameters of the study participants are shown in Table 1. In the SCD cohort, 40 patients (53%) from the severe group and 10 patients (39%) from the mild group were on stably‐dosed (stable for at least 90 days prior to participation and with no anticipated need for dose adjustment) hydroxyurea. The thalassemia group consisted of six α‐thalassemia and 12 β‐thalassemia patients. Nine patients (12%) from the severe SCD group and 11 (61%) patients from the thalassemia group were treated with chronic transfusion and were examined between 0‐6 days before their routine transfusion. Of the nine transfusion‐dependent SCD patients, five patients received exchange transfusion and four were on simple transfusion.

Table 1.

Mean and standard deviation or median and interquartile range of demographic and clinical parameters per group with test statistics and p‐values between the groups.

	Severe SCD	Mild SCD	Thalassemia	Controls	Test statistics (p‐values)
N	75	26	18	30	‐
Demographics					‐
Ethnicity
African (%)	85.3	84.6	‐	63.3
Hispanic American (%)	8	3.8	‐	33.3
Asian (%)	6.7	11.5	66.7	3.3
European (%)	‐	‐	16.7	‐
Middle East (%)	‐	‐	16.7	‐
Age	28.0 ± 14.0	29.5 ± 16.3	29.0 ± 9.4	31.5 ± 11.5	H: 2.0 (0.58)
Sex	33 F (44%)	8 (31%)	12 (67%)	16 (53%)	X 2: 6.3 (0.1)
Clinical parameters
Genotype (N)	64 HbSS 11 HbSβ0	16 HbSC 10 HbSβ+	7 NTDT 11 TDT	1 HbAC 5 HbAS 24 HbAA	‐
Hemoglobin (g/dL)	8.8 ± 1.3	11.8 ± 1.8	9.3 ± 1.1	13.1 ± 1.3	F: 80.0 (<0.01)
Hematocrit (%)	25.8 ± 4.3	34.2 ± 4.7	30.0 ± 2.8	39.8 ± 3.7	F: 91.7 (<0.01)
Reticulocytes (109/L)	258.5 ± 156.7*	120.5 ± 43.8*	160.0 ± 258.3*	63.9 ± 31.7*	H: 61.1 (<0.01)
HbF (%)	6.2 ± 11.0*	3.2 ± 5.7*	1.1 ± 5.6*	‐	H: 70.0 (<0.01)
Hydroxyurea N (%)	40 (53%)	10 (39%)	‐	‐	X 2: 37.7 (<0.01)
Chronic transfusion N (%)	4 simple (5%) 5 exchange (7%)	‐	11 (61%)	‐	X 2: 44.0 (<0.01)

Post hoc statistics (p‐values)
	S‐M	S‐T	S‐C	M‐T	M‐C	T‐C
Demographics
Age	p = 0.40	p = 0.21	p = 0.45	p = 0.91	p = 0.75	p = 0.65
Sex	p = 0.26	p = 0.12	p = 0.40	p  = 0.03	p = 0.11	p = 0.55
Clinical parameters
Hemoglobin (g/dL)	p  < 0.01	p > 0.99	p  < 0.01	p  < 0.01	p  < 0.01	p  < 0.01
Hematocrit (%)	p  < 0.01	p  < 0.01	p  < 0.01	p  < 0.01	p  < 0.01	p  < 0.01
Reticulocytes (109/L)	p  < 0.01	p  < 0.01	p  < 0.01	p = 0.88	p  < 0.01	p = 0.10
HbF (%)	p  < 0.01	p  < 0.01	‐	p = 0.98	‐	‐
Hydroxyurea N (%)	p = 0.26	‐	‐	‐	‐	‐
Chronic transfusion N (%)	‐	p  < 0.01	‐	‐	‐	‐

Note: Test statistics were calculated using ANOVA or Kruskal–Wallis in case of nonnormality or Chi‐square test for categorical variables. Asterisks denote the reported median and interquartile range. Bold values denote statistically significant p‐values.

Abbreviations: ANOVA, analysis of variance; C, healthy controls; HbF, fetal hemoglobin; M, mild SCD; S, severe SCD; T, thalassemia.

Baseline imaging parameters between the groups

Participant groups differed significantly in baseline WM and GM‐CBF, GM oxygen delivery, OEF, and CMRO₂, but not in WM oxygen delivery, while adjusted for age and sex (Table 2 and for covariate statistics Supporting Information S1: Section 1.4). Representative CBF maps of patients with severe and mild SCD, thalassemia, and healthy controls before and after ACZ administration are shown in Figure 1. WM and GM‐CBF values were significantly higher in patients with severe SCD and thalassemia compared to mild SCD and healthy controls and the GM‐CBF values in severe SCD patients were significantly higher compared to the thalassemia group (Table 2 and Figure 2A,B). GM oxygen delivery was significantly lower in thalassemia patients compared to healthy controls (Figure 2D). OEF was significantly lower in severe SCD and thalassemia patients compared to the mild SCD patients and controls, but significantly higher in the thalassemia group compared to severe SCD (Table 2 and Figure 2E). CMRO₂ was significantly lower in patients with severe SCD and thalassemia compared to the patients with mild SCD and healthy controls. No significant differences in CMRO₂ were found between severe SCD patients and thalassemia patients (Table 2 and Figure 2F). After adjusting CMRO₂ for hemoglobin levels, it remained significantly lower in severe SCD compared to both mild SCD (p < 0.01) and healthy controls (p = 0.02), and was also lower in thalassemia patients compared to mild SCD patients (p = 0.02).

Table 2.

Mean and standard deviation or median and interquartile range of hemodynamic parameters, parameters of oxygen metabolism, and SCI volumes per group with test statistics and p values.

	Severe SCD	Mild SCD	Thalassemia	Controls
WM CBF (mL/100 g/min)
Pre‐ACZ	40.6 ± 12.2*	30.6 ± 9.9*	39.2 ± 8.9	27.8 ± 3.2
Post‐ACZ	55.1 ± 18.6*	45.1 ± 12.1	52.1 ± 9.3	38.8 ± 5.4
Paired tests	Z = −7.3, p < 0.01	Z = −3.9, p < 0.01	t = −13.5, p < 0.01	t = −13.6, p < 0.01
GM CBF (mL/100 g/min)
Pre‐ACZ	85.2 ± 16.3	60.5 ± 9.8	73.4 ± 17.6*	57.5 ± 8.1
Post‐ACZ	108.5 ± 20.1	81.3 ± 20.6	90.7 ± 13.2*	81.0 ± 13.0
Paired tests	t = −14.5, p < 0.01	t = −8.5, p < 0.01	Z = −3.4, p < 0.01	t = −14.7, p < 0.01
WM oxygen delivery (mL/100 g/min)
Pre‐ACZ	4.7 ± 0.7	4.8 ± 0.8	4.6 ± 0.9	4.7 ± 0.5
Post‐ACZ	6.2 ± 1.3*	6.1 ± 2.1*	6.2 ± 0.9	6.6 ± 0.9
Paired tests	Z = −7.2, p < 0.01	Z = −3.7, p < 0.01	t = −13.2, p < 0.01	t = −13.5, p < 0.01
GM oxygen delivery (mL/100 g/min)
Pre‐ACZ	9.6 ± 1.4	9.2 ± 1.2	8.3 ± 1.5	9.2 ± 1.1*
Post‐ACZ	12.1 ± 2.6*	12.8 ± 2.0	10.6 ± 1.3	13.8 ± 2.9
Paired tests	Z = −7.2, p < 0.01	t = −9.1, p < 0.01	t = −11.5, p < 0.01	Z = −4.4, p < 0.01
OEF (%)
Pre‐ACZ	28.4 ± 5.5*	40.5 ± 4.3	33.1 ± 5.4	39.5 ± 5.1
Post‐ACZ	19.4 ± 7.6*	29.8 ± 5.1	26.6 ± 4.4	29.1 ± 6.5
Paired tests	Z = −6.9, p < 0.01	t = 10.1, p < 0.01	t = 6.3, p < 0.01	t = 11.7, p < 0.01
CMRO2 (mL O2/100 g/min)
Pre‐ACZ	2.1 ± 0.4	2.9 ± 0.4	2.2 ± 0.4	3.0 ± 0.4
Post‐ACZ	1.8 ± 0.8*	2.9 ± 0.4	2.3 ± 0.4	3.0 ± 0.6
Paired tests	Z = −4.1, p < 0.01	t = −0.67, p = 0.51	t = −0.44, p = 0.67	t = −0.12, p = 0.91
SCI presence N [%]	53 [70.7]	18 [69.2]	15 [83.3]	19 [63.3]
SCI volume (mL) in all subjects	0.11 ± 0.6*	0.07 ± 0.3*	0.10 ± 0.2*	0.02 ± 0.1*
SCI volume (mL) in all subjects with SCI	0.27 ± 0.7*	0.12 ± 1.0*	0.12 ± 0.1*	0.06 ± 0.2*

Test statistics between the groups
	df (between, within)	Test statistics (p‐value)	p‐value (S‐M)	p‐value (S‐T)	p‐value (S‐C)	p‐value (M‐T)	p‐value (M‐C)	p‐value (T‐C)
WM CBF (mL/100 g/min)	3137	F  = 39.4 (<0.01)	<0.01	0.58	<0.01	<0.01	0.10	<0.01
GM CBF (mL/100 g/min)	3138	F = 47.1 (<0.01)	<0.01	<0.01	<0.01	0.02	>0.99	<0.01
WM oxygen delivery (mL/100 g/min)	3134	F = 0.18 (0.91)	>0.99	>0.99	>0.99	>0.99	>0.99	>0.99
GM oxygen delivery (mL/100 g/min)	3135	F = 3.6 (0.02)	0.56	0.06	>0.99	>0.99	0.30	0.04
OEF (%)	3139	F = 46.6 (<0.01)	<0.01	0.04	<0.01	<0.01	>0.99	<0.01
CMRO2 (mL O2/100 g/min)	3131	F = 47.2 (<0.01)	<0.01	>0.99	<0.01	<0.01	>0.99	<0.01
∆WM CBF (mL/100 g/min)	3122	F = 0.08 (0.97)	>0.99	>0.99	>0.99	>0.99	>0.99	>0.99
∆GM CBF (mL/100 g/min)	3124	F = 1.4 (0.26)	>0.99	0.51	>0.99	0.84	>0.99	0.32
∆WM oxygen delivery (mL/100 g/min)	3116	F = 2.21 (0.09)	0.54	>0.99	0.35	0.62	>0.99	0.46
∆GM oxygen delivery (mL/100 g/min)	3118	F = 14.8 (<0.01)	<0.01	>0.99	<0.01	<0.01	0.88	<0.01
∆OEF (%)	3125	F = 4.9 (<0.01)	0.73	<0.01	0.56	0.97	>0.99	0.39
∆CMRO2 (mL O2/100 g/min)	3111	F = 8.2 (<0.01)	<0.01	0.03	<0.01	>0.99	>0.99	>0.99
SCI presence	3146	X 2 = 1.68 (0.64)	0.75	0.31	0.62	0.52	0.51	0.21
SCI volume (mL) in all subjects	3142	F = 3.1 (0.03)	>0.99	>0.99	0.02	>0.99	0.65	0.47
SCI volume (mL) in all subjects with SCI	3102	F = 5.2 (<0.01)	0.75	>0.99	<0.01	>0.99	0.41	0.45

Note: Paired sample t‐tests or Wilcoxon tests were used to study the differences before and after ACZ. ANCOVA or Quade's test with post hoc Bonferroni was performed to study the differences between the groups. Asterisks denote the reported median and interquartile range. Bold values denote statistically significant p‐values.

Abbreviations: ANCOVA, analysis of covariance; C, healthy controls; CBF, cerebral blood flow; CMRO₂, cerebral metabolic rate of oxygen; GM, gray matter; M, mild SCD; OEF, oxygen extraction fraction; S, severe SCD; SCD, sickle cell disease; SCI, silent cerebral infarcts; T, thalassemia, WM, white matter.

Figure 1.

Six axial slices of cerebral blood flow (CBF) before and after acetazolamide (ACZ) administration of representative patients with severe sickle cell disease (SCD), mild SCD, and thalassemia, and a healthy control.

Figure 2.

Dot plots of WM CBF (A), GM CBF (B), WM oxygen delivery (C), GM oxygen delivery (D), OEF (E), and CMRO₂ (F) for all groups. Error bars denote the mean and standard deviation. Asterisks denote statistically significant differences compared to the severe SCD group and diamonds denote statistically significant differences compared to the Thalassemia group. CBF, cerebral blood flow; CMRO₂, cerebral metabolic rate of oxygen; Con, healthy controls; GM, gray matter; Mild, mild SCD patients; OEF, oxygen extraction fraction; severe, severe SCD patients; Thal, thalassemia patients; WM, white matter.

To analyze the differences between patients with or without hydroxyurea treatment, mild and severe groups were taken together while patients receiving blood transfusion were excluded. WM and GM oxygen delivery were significantly higher in SCD patients receiving hydroxyurea compared to those who did not (WM: mean = 4.96 ± 0.7 mL/100 g/min vs. 4.43 ± 0.6 mL/100 g/min, Z = −4.0, p < 0.01, GM: mean = 9.89 ± 1.4 mL/100 g/min vs. 8.93 ± 1.4 mL/100 g/min, t = −3.3, p < 0.01). OEF was slightly lower in patients receiving hydroxyurea compared to those not receiving it (OEF: mean = 30.20 ± 7.3 vs. 33.25 ± 7.7, Z = −2.1, p = 0.03). CMRO₂ was not significantly different between the two groups. To analyze the effect of transfusion therapy on cerebral hemodynamics and oxygen metabolism patients receiving hydroxyurea were excluded from the analyses. GM and WM oxygen delivery were significantly higher in patients receiving chronic transfusion compared to those without (WM: mean = 5.10 ± 0.6 mL/100 g/min vs. 4.43 ± 0.7 mL/100 g/min, t = −2.5, p = 0.04, GM: mean = 10.46 ± 0.8 mL/100 g/min vs. 8.93 ± 1.4 mL/100 g/min, t = −2.7, p = 0.01). Given the small size of the subset without transfusion, statistical analyses were not performed in the thalassemia group.

Effect of acetazolamide on imaging parameters

After ACZ administration, WM and GM‐CBF, and oxygen delivery increased and OEF significantly decreased in all groups (Table 2). CMRO₂ significantly decreased after ACZ in the severe SCD group, but not in other groups (Table 2). ∆GM oxygen delivery, ∆OEF, and ∆CMRO₂, but not ∆GM and ∆WM CBF and ∆WM oxygen delivery were significantly different between the participant groups (Table 2 and Figure 3). ∆GM oxygen delivery was significantly lower in severe SCD and thalassemia patients compared to mild and controls (Table 2 and Figure 3D). In thalassemia patients, OEF decreased less after ACZ compared to severe SCD patients (Table 2 and Figure 3E). ∆CMRO₂ was larger in the severe SCD group compared to mild SCD patients, thalassemia patients, and controls (Figure 3F). An alternative analysis studying the differences in cerebrovascular reserve (CVR) between the groups yielded comparable findings, as detailed in the Supporting Information S1: Section 1.5. No significant differences were found between patients receiving hydroxyurea or a chronic transfusion and those not receiving these treatments.

Figure 3.

Dot plots of ∆WM CBF (A), ∆GM CBF (B), ∆WM oxygen delivery (C), ∆GM oxygen delivery (D), ∆OEF (E), and ∆CMRO₂ (F) for all groups. Error bars denote the mean and standard deviation. Stars denote statistically significant differences compared to the severe SCD group and diamonds denote statistically significant differences compared to the Thalassemia group. CBF, cerebral blood flow; CMRO₂, cerebral metabolic rate of oxygen; Con, healthy controls; GM, gray matter; Mild, mild SCD patients; OEF, oxygen extraction fraction; Severe, severe SCD patients; Thal, thalassemia patients; WM, white matter.

Silent cerebral infarcts

Significant larger SCI volume was found in the severe SCD group as compared to the healthy controls (Table 2 and Figure 4). No significant differences in SCI volume were found between the other groups. Subsequently, the other imaging parameters were compared between SCD patients with SCIs and those without SCIs. Baseline WM and GM oxygen delivery were significantly higher in SCD patients without SCIs compared to SCD patients with SCIs (Figure 5C,D). In SCD patients with SCI, ΔGM CBF (t = −2.9, p < 0.01), ΔWM CBF (Z = −2.2, p = 0.03), and ΔWM oxygen delivery (Z = −2.2, p = 0.03) were significantly higher compared to patients without SCIs. No significant differences in other imaging parameters were found between patients with and without SCIs.

Figure 4.

Violin plot with dot plots of SCI volumes for all groups. Con, healthy controls; Mild, mild SCD patients; SCI, silent cerebral infarct; Severe, severe SCD patients; Thal, thalassemia patients.

Figure 5.

Dot plots of WM CBF (A), GM CBF (B), WM oxygen delivery (C), GM oxygen delivery (D), OEF (E), and CMRO₂ (F) for SCD patients with and without SCIs. Error bars denote the mean and standard deviation. CBF, cerebral blood flow; CMRO₂, cerebral metabolic rate of oxygen; GM, gray matter; OEF, oxygen extraction fraction; SCI, silent cerebral infarct; WM, white matter.

HbF was positively associated with hemoglobin level (r = 0.40, p < 0.01) and negatively associated with reticulocyte count (r = −0.37, p < 0.01), and HbS% (r = −0.34, p < 0.01) in severe SCD patients. In addition, we found a negative association between HbF and OEF (r = −0.31, p = 0.01), but no associations were found with WM‐CBF (r = −0.16, p = 0.18), GM‐CBF (r = −0.16, p = 0.20), WM oxygen delivery (r = 0.21, p = 0.1), GM oxygen delivery (r = 0.22, p = 0.08) and CMRO₂ (r = −0.09, p = 0.46) in these patients.

DISCUSSION

In this study, we assessed cerebral hemodynamics, oxygen metabolism, and SCI volumes in different genotypes of SCD and thalassemia patients, and matched healthy controls to further explore the pathophysiology of SCIs. Our results showed that in contrast to patients with severe SCD and thalassemia, CBF, OEF and CMRO₂ were not affected in patients with milder SCD, suggesting a more balanced oxygen supply versus demand in these patients. After administering ACZ, there was an increase in oxygen delivery in patients with milder SCD and in healthy controls, compared to other groups. This indicates that, unlike in severe SCD and thalassemia patients, in these two groups, vessel dilation in response to a vasoactive stimulus can enhance oxygen supply. In patients with severe SCD, CMRO₂ was further impaired following ACZ administration, indicating a functional shunting effect in this group. SCI volumes gradually increased from healthy controls to thalassemia, mild SCD, and severe SCD patients, though the difference was only statistically significant between healthy controls and patients with severe SCD.

In patients with severe SCD, CBF is increased to compensate for the lower oxygen delivery caused by anemia. Nonetheless, elevated CBF can decrease the vasodilatory reserve capacity due to chronic vessel dilation at rest which may result in a higher susceptibility to ischemia when additional dilatation is required. ¹⁰ , ¹² Additionally, increased CBF may also limit the oxygen offloading due to functional shunting which may ultimately result in local cerebral ischemia. ⁸ , ⁹ In our study, we found impaired oxygen extraction and consumption in patients with severe SCD which may have significant implications for brain health. When both global OEF and CMRO₂ are diminished, there is an imbalance between oxygen supply and demand, potentially increasing the risk of silent infarcts that could eventually result in cognitive impairment and reduced quality of life. However, in line with previous studies investigating global CMRO₂, we did not find an association between CMRO₂ and SCIs. ¹¹ , ¹⁶ The lack of association could have several explanations. First, our cross‐sectional design limits the opportunity to demonstrate a possible causal relationship. Our study showed that SCD patients with SCIs have lower oxygen delivery, which can be further impacted during acute episodes of oxygen deprivation such as episodes of worsened anemia or hypoxia. This could potentially contribute to the development of SCIs that may not be present at the time of the MRI analysis and CMRO₂ assessment. Therefore, a longitudinal study might be needed to follow the association between CMRO₂ and SCIs over time in order to provide more insights in this regard. Secondly, we cannot rule out that WB CMRO₂ might not be the most suitable parameter for examining associations with SCIs localized in the watershed areas of the WM. Fields et al. investigated OEF and CMRO₂ more locally in deep white matter and demonstrated decreased CBF, increased OEF, and a relatively decreased CMRO₂ in these specific areas which co‐localize with regions of high infarct density. ²⁶ Additionally, we observed that differences in CMRO₂ persisted even after adjusting for hemoglobin levels, indicating that factors beyond anemia contribute to the reduced oxygen consumption in patients with severe SCD and thalassemia. Vascular damage, which is commonly found in these patients, might be one such contributing factor. Future research should focus on investigating vascular damage within the brain and its relationship with cerebral oxygen consumption.

By including thalassemia patients as a nonsickling form of hereditary anemia, we were able to study the contribution of HbS polymerization in the cerebral hemodynamics and oxygen metabolism of patients with SCD. Contrary to the hypothesis, comparable CMRO₂ values were found between thalassemia patients and those with severe SCD, suggesting that lower oxygen metabolism is primarily related to the level of anemia in both patient groups and, to a lesser extent to HbS polymerization. However, CMRO₂ declined after ACZ in severe SCD patients but not in thalassemia patients, suggesting diffusion‐limited oxygen availability in severe SCD. The mechanism of this phenomenon is unknown but likely reflects functional shunting. During functional shunting the increased CBF values found in severe SCD patients may limit offloading of oxygen resulting in decreased cerebral oxygen metabolism; an effect that may be less pronounced in mild SCD due to lower CBF. ⁸ , ⁹ Although venous outflow as a marker of functional shunting has been observed in two transfusion‐dependent thalassemia patients, ²⁷ it has not been studied in a larger group of thalassemia patients or patients with mild SCD.

GM oxygen delivery was elevated in mild SCD and healthy controls following ACZ administration compared to other groups, while changes in ΔGM CBF showed no significant differences across the groups. This stability in ΔGM CBF implies that the differences in oxygen delivery may not be primarily driven by alterations in blood flow or vascular dilation. Our results suggest that the variations in ΔGM oxygen delivery could be linked to the higher hemoglobin levels observed in both the mild SCD and control groups. Unexpectedly, there was a significant difference in ΔOEF between severe SCD and thalassemia patients, which remain not fully understood. A possible explanation is that ACZ might have a reduced effect on the hemodynamics and oxygen metabolism in thalassemia patients. While these differences were not statistically significant, our data indicate a pattern where both GM CBF and GM oxygen delivery increase less after acetazolamide administration, and the OEF decreases to a lesser extent in the thalassemia group compared to other groups. However, the mechanisms behind these findings could not be explained within the scope of our current study. Consequently, further research with a larger sample of thalassemia patients is required to explore the effects of vasodilation in this group more thoroughly.

An unexpected finding was that despite the significantly higher OEF and CMRO₂ found in patients with mild SCD in comparison with severe SCD, no statistical difference in SCI volume was found between these groups. One potential explanation is that the observed SCIs might have arisen from earlier anemic crises or other disease‐related complications, which we cannot determine with our cross‐sectional study design. Another possibility is that in patients with milder SCD forms, anemia might not be the sole contributing factor to SCI development. Other mechanisms may play a role in the pathogenesis of SCI such as RBC deformability, or rheology aspects of sickle RBCs which have also been suggested to play a role in the sickle cell‐related retinopathy. ²⁸ , ²⁹

Interestingly, we did not find an association between HbF and CMRO₂ in patients with severe SCD, which was contrary to our previous study showing a negative association between HbF and CMRO₂. ⁸ However, our current results showed that HbF was positively correlated with hemoglobin levels and negatively with OEF confirming that higher HbF levels decrease anemia but may also decrease OEF due to the higher oxygen affinity. The unchanged CMRO₂ might be attributed to a counteracting alteration of hemoglobin and OEF.

In our study, the OEF values observed in severe SCD patients are decreased which differs from observations in studies using asymmetric spin echo (ASE) MRI technique or TRUST studies using other calibration models. ³⁰ , ³¹ These discrepancies in OEF values in SCD patients may cause confusion. Therefore, it is crucial to explain these differences to ensure an accurate interpretation of the results. As recently clarified by Stout et al., OEF values measured by TRUST are flow‐averaged and determined by the comparison of the oxygenation of the blood in the venous sinus with the arterial saturation, while those obtained from the ASE method are spatially averaged, determining the venous oxygenation more locally in the brain. ³² Consequently, the OEF derived from the ASE method offers more localized oxygen extraction data, while the OEF from TRUST provides insights into the overall brain extraction of oxygen. Furthermore, OEF values are also strongly dependent on the calibration model used. The Bush‐Li model used for severe SCD patients in our study is based on a sickle cell‐specific calibration model with a larger range of hematocrit, as compared with the bovine calibration model or HbA‐model. ³¹ , ³³ In contrast to severe SCD patients, no calibration model is available for patients with mild SCD like HbSC and HbSβ⁺‐thalassemia, therefore, we chose the HbA model which was the most suitable model according to the CMRO₂‐hemoglobin association graph (see Supporting Information S1: Section 1.3). However, the features of RBCs in these SCD patients are not the same as healthy controls, and therefore disease group‐specific calibration models are required to calculate the venous oxygenation in patients with mild SCD and possibly also in thalassemia.

ACZ was used as a vasoactive stimulus, which is widely used in healthy controls and several patient groups including patients with carotid stenosis, carotid occlusions, moyamoya disease, dementia and Alzheimer's disease and SCD. ¹² , ³⁴ , ³⁵ , ³⁶ , ³⁷ , ³⁸ , ³⁹ A previous study has reported that daily use of ACZ could lead to sickle cell crisis in patients with SCD. ⁴⁰ However, in the study described here, patients received ACZ only once during their MRI scan. Of the 103 patients initially included, only three experienced symptoms of a sickle cell crisis, which resolved within a few days without any complications.

This study has some limitations. First, we have unequally sized groups which reduced power for some comparisons. Second, both in the severe SCD and thalassemia groups, patients receiving blood transfusions were included, which may have influenced the comparison we did. However, we did not find significant differences between patients receiving blood transfusions and those not receiving them. Third, assessing CBF in white matter using te‐pCASL remains a challenge due to its lower signal‐to‐noise ratio (SNR), which may impact the findings. Notwithstanding the lower SNR, we were still able to find differences in WM CBF between the participant groups. Finally, no specific calibration models were available for patients with mild SCD, as well as for transfusion‐dependent and independent thalassemia patients. Although the absence of group‐specific calibration models may have influenced the results, our analysis represents the initial attempt to explore oxygen metabolism in these groups. We encourage future studies to develop group‐specific calibration models for these groups, thereby enabling more accurate insights into the cerebral oxygenation of these patients.

In conclusion, we found that cerebral oxygen consumption in severe SCD and thalassemia patients is reduced, which appeared to be related to anemia. Patients with mild SCD appeared to have a more balanced oxygen supply versus demand. However, they had comparable SCI volumes, which could either be attributed to earlier acute SCD complications or SCD‐related factors not tied to cerebral oxygen dynamics. The reduction in CMRO₂ following the vasodilatory stimulus in severe SCD patients points to functional shunting in this group, possible due to their higher CBF compared to the other groups.

AUTHOR CONTRIBUTIONS

Liza Afzali‐Hashemi involved in study conception and design, data acquisition, data analysis, interpretation of the data, and manuscript drafting. Koen P. A. Baas involved in data acquisition, analysis, interpretation of the data, and manuscript revision. Anouk Schrantee involved in the interpretation of the data and manuscript revision. Erfan Nur involved in the research set up and manuscript revision. Chau Vu involved in data acquisition and analysis and manuscript revision. Soyoung Choi involved in data analysis. Silvie Suriany involved in data analysis. John C. Wood involved in data acquisition, interpretation, and manuscript revision. Aart J. Nederveen involved in study conception and design, data acquisition, analysis, interpretation of the data and manuscript revision. Bart J. Biemond involved in study conception and design, analysis, interpretation of the data and manuscript revision.

CONFLICT OF INTEREST STATEMENT

Erfan Nur has received a research grant from Novartis and participated in advisory boards and the speaker's office of Novartis. Bart J. Biemond has received research grants from Sanquin, Novartis, GBT, and BMS and participated in advisory board meetings of Novartis, Celgene, Novo Nordisk, CSL Behring, and GBT/Pfizer. Liza Afzali‐Hashemi, Koen P. A. Baas, Anouk Schrantee, Chau Vu, Soyoung Choi, Silvie Suriany, John C. Wood, and Aart J. Nederveen report no disclosures.

FUNDING

This investigation was supported by grant HL136484‐01A1 to JCW, AJN, and BJB from the National Institutes of Health R01, USA.

Supporting information

Supporting information.

DATA AVAILABILITY STATEMENT

For original data, please contact b.j.biemond@amsterdamumc.nl.