Quantitative MRI of the brain in children with sickle cell disease reveals abnormalities unseen by conventional MRI

Abstract

Conventional MRI (cMRI) has shown that brain abnormalities without clinical stroke can manifest in patients with sickle cell disease (SCD). We used quantitative MRI (qMRI) and psychometric testing to determine whether brain abnormalities can also be present in patients with SCD who appear normal on cMRI. Patients 4 years of age and older with no clinical evidence of stroke were stratified by cMRI as normal (n = 17) or abnormal (n = 13). Spin-lattice relaxation time (T1) of gray and white matter structures was measured by the precise and accurate inversion recovery (PAIR) qMRI method. Patient cognitive ability was assessed with a standard psychometric instrument (WISC-III or WISC-R). In all 30 patients with SCD, qMRI T1 was lower than in 24 age- and race-matched controls, in cortical gray matter (P < .0006) and caudate (P < .0009), as well as in the ratio of gray-to-white matter T1 (P < .008). In the 17 patients who were shown to be normal by cMRI, qMRI T1 was still lower than in controls, in both cortical gray matter (P < .02) and caudate (P < .004). Histograms of voxel T1 show that the proportion of voxels with T1 values intermediate between gray and white matter (ie, consistent with encephalomalacia) was 9% higher than controls in patients shown to be normal by cMRI (P < .05) and 15% higher than controls in patients shown to be abnormal by cMRI (P < .0005). The full scale intelligence quotient (FSIQ) of all patients with SCD was 75, compared to the FSIQ of 88 in a historical control group of patient siblings (P < .001). The FSIQ of patients shown to be normal by cMRI was 79, significantly lower than the FSIQ of patient siblings (P < .04). The FSIQ of 71 in patients shown to be abnormal by cMRI was significantly lower than both the patient siblings (P < .005) and the patients shown to be normal by cMRI (P < .04). Patients shown to be abnormal by cMRI scored lower than patients shown to be normal by cMRI, specifically on the subtests of vocabulary (P = .003) and information (P = .03). Cognitive impairment is thus significant, even in patients with SCD who were shown to be normal by cMRI, suggesting that cMRI may be insensitive to subtle neurologic damage that can be detected by qMRI. Because cognitive impairment can occur in children normal by cMRI, our findings imply that prophylactic therapy may be needed earlier in the course of SCD to mitigate neurologic damage.