White matter integrity and processing speed in sickle cell anemia

Abstract

Objective: The purpose of this retrospective cross-sectional study was to investigate whether changes in white matter integrity are related to slower processing speed in sickle cell anemia.

Methods: Thirty-seven patients with silent cerebral infarction, 46 patients with normal MRI, and 32 sibling controls (age range 8-37 years) underwent cognitive assessment using the Wechsler scales and 3-tesla MRI. Tract-based spatial statistics analyses of diffusion tensor imaging (DTI) and neurite orientation dispersion and density imaging (NODDI) parameters were performed.

Results: Processing speed index (PSI) was lower in patients than controls by 9.34 points (95% confidence interval: 4.635-14.855, p = 0.0003). Full Scale IQ was lower by 4.14 scaled points (95% confidence interval: -1.066 to 9.551, p = 0.1), but this difference was abolished when PSI was included as a covariate (p = 0.18). There were no differences in cognition between patients with and without silent cerebral infarction, and both groups had lower PSI than controls (both p < 0.001). In patients, arterial oxygen content, socioeconomic status, age, and male sex were identified as predictors of PSI, and correlations were found between PSI and DTI scalars (fractional anisotropy r = 0.614, p < 0.00001; r = -0.457, p < 0.00001; mean diffusivity r = -0.341, p = 0.0016; radial diffusivity r = -0.457, p < 0.00001) and NODDI parameters (intracellular volume fraction r = 0.364, p = 0.0007) in widespread regions.

Conclusion: Our results extend previous reports of impairment that is independent of presence of infarction and may worsen with age. We identify processing speed as a vulnerable domain, with deficits potentially mediating difficulties across other domains, and provide evidence that reduced processing speed is related to the integrity of normal-appearing white matter using microstructure parameters from DTI and NODDI.

Figure 1. Neurocognitive and hematologic variables

(A) Differences in hemoglobin (Hgb; left), arterial oxygen content (Cao₂; middle), and oxygen saturation (Spo₂), between patients with (SCI+) and without (SCI−) silent cerebral infarction. (B) Differences in processing speed index (PSI) between healthy controls (HCs) and patients (SCI−, SCI+). *p < 0.05; **p < 0.01 (after Bonferroni correction for multiple comparisons). Horizontal line represents mean PSI in the normative population. NS = not significant.

Figure 2. Correlations between predictors of processing speed

Correlogram visualizing relationships between variables included in the exploratory regression analysis. Values are zero-order Pearson correlation coefficients. Shaded areas represent significant relationships. Blue colors represent positive relationships, whereas red colors represent negative relationships. Intensity signifies the strength of relationships. PSI = processing speed index; SCI = silent cerebral infarction; SES = socioeconomic status.

Figure 3. Correlations between diffusion metrics and processing speed

(A) Blue voxels indicate areas in which fractional anisotropy (FA) correlated with processing speed index (PSI) (34,392 voxels, p < 0.05). Red voxels indicate areas in which intracellular volume fraction (ICVF) correlated with PSI (70,659 voxels, p < 0.05). (B) Yellow voxels indicate areas in which radial diffusivity (RD) correlated with PSI (67,296 voxels, p < 0.05). Purple voxels indicate areas in which mean diffusivity (MD) correlated with PSI (82,663 voxels, p < 0.05). Results were age, sex, education decile (SES), and threshold-free cluster enhancement corrected and overlaid on the group white matter skeleton (green) and the study-specific mean FA template. Adj. = adjusted; SCI = silent cerebral infarction; SES = socioeconomic status.