Red cell exchange transfusions lower cerebral blood flow and oxygen extraction fraction in pediatric sickle cell anemia

Abstract

Blood transfusions are the mainstay of stroke prevention in pediatric sickle cell anemia (SCA), but the physiology conferring this benefit is unclear. Cerebral blood flow (CBF) and oxygen extraction fraction (OEF) are elevated in SCA, likely compensating for reduced arterial oxygen content (CaO₂). We hypothesized that exchange transfusions would decrease CBF and OEF by increasing CaO₂, thereby relieving cerebral oxygen metabolic stress. Twenty-one children with SCA receiving chronic transfusion therapy (CTT) underwent magnetic resonance imaging before and after exchange transfusions. Arterial spin labeling and asymmetric spin echo sequences measured CBF and OEF, respectively, which were compared pre- and posttransfusion. Volumes of tissue with OEF above successive thresholds (36%, 38%, and 40%), as a metric of regional metabolic stress, were compared pre- and posttransfusion. Transfusions increased hemoglobin (Hb; from 9.1 to 10.3 g/dL; P < .001) and decreased Hb S (from 39.7% to 24.3%; P < .001). Transfusions reduced CBF (from 88 to 82.4 mL/100 g per minute; P = .004) and OEF (from 34.4% to 31.2%; P < .001). At all thresholds, transfusions reduced the volume of peak OEF found in the deep white matter, a location at high infarct risk in SCA (P < .001). Reduction of elevated CBF and OEF, both globally and regionally, suggests that CTT mitigates infarct risk in pediatric SCA by relieving cerebral metabolic stress at patient- and tissue-specific levels.

Graphical abstract

Figure 1.

Exchange transfusions reduce CBF and OEF without changing CMRO₂. Wilcoxon signed-rank test compared CBF, OEF, and CMRO₂ before and after transfusion. CBF and OEF were significantly reduced after transfusion (P < .01); CMRO₂ was not altered (P = .10). **P < .01. ns, not significant.

Figure 2.

CBF and OEF maps from a child with SCA. This 7-year-old boy underwent an MRI scan before CTT initiation and again before and after an exchange transfusion (only CTT values included in cohort-level analyses). The whole-brain CBF was highest at his first scan (138 mL/100 g per minute). After 17 months of CTT, his pretransfusion CBF was lower than his initial scan (118 mL/100 g per minute; 14% drop) and further decreased after transfusion to 100 mL/100 g per minute (15% drop). The whole-brain OEF was highest at the first scan (46.1%), with dramatic reduction in OEF measured pretransfusion (29.4%; 38% drop) and only modest reduction posttransfusion (28.6%; 3% drop). His OEF maps were also notable for regionally elevated OEF in the deep white matter (blue arrows), which was most prominent before CTT initiation. This peak OEF was still detectable on the pretransfusion scan, although less prominent. After transfusion, the peak OEF region was absent, with restoration of homogeneous OEF across the brain.

Figure 3.

Hb is an independent predictor of OEF and CBF. Univariate correlations with Spearman’s rank test (ρ) of Hb (A) and Hb S percentage (B) with OEF and of Hb (C) and Hb S percentage (D) with CBF. Light lines connect individual patient values before and after transfusion. Blue lines indicate group-level regression. (E) Multivariate modeling of OEF and CBF as functions of Hb, Hb S percentage, age, and sex retain Hb (OEF and CBF) and age (CBF) as independent predictors. SE, standard error.

Figure 4.

Regional peak OEF diminishes with transfusion. Averaged pretransfusion (A) and posttransfusion (B) OEF maps. Left side demonstrates averaged map. Right side demonstrates regions above successive thresholds as applied to averaged maps. (C) Median percentage of hemispheric brain volume above OEF thresholds decreased after transfusion.

Figure 5.

CBF and OEF in a CTT SCA cohort lie between non-CTT SCA and healthy control cohorts. Mann-Whitney U tests directly compared the pre- and posttransfusion values with the control and also with the SCA nontransfused (no CTT) values. Wilcoxon signed-rank tests compared pre- and posttransfusion values. P values were corrected with Benjamani-Hochberg procedure for multiple comparisons. *P < .05 after correction, **P < .01 after correction.