Children with sickle cell anemia (SCA) experience increased metabolic stress in the brain, measured by oxygen extraction fraction (OEF); regions with the highest OEF co-localize with regions at greatest risk for stroke. Yet, SCA affects cognition independent of stroke. We aim to investigate the relationship between metabolic stress and the cognitive effects of SCA. Neurocognitive abilities emerge from functional networks ascertained using unctional connectivity MRI (rs-fcMRI), potentially allowing functional connectivity (fc) to be used as a biomarker for cognitive dysfunction in SCA. We prospectively obtained cognitive testing and brain MRIs in children with SCA (unaffected by overt stroke or vasculopathy, not chronically transfused and without history of transplant) and controls in a single center to test our hypothesis that children with SCA experiencing the greatest metabolic stress will have the greatest disruption in fc. Brain MRI measured OEF (asymmetric spin echo) and fc (resting state BOLD), and the NIH Toolbox Cognition Battery (NIHTB) and Wechsler Abbreviated Scale of Intelligence Second Edition (WASI-II) assessed cognition. OEF processing is described in Fields et al. Blood. 2019; 133(22):2436-2444. Temporal correlation of the BOLD signal between 264 gray matter (GM) regions was calculated and assembled into region x region matrices sorted by 13 pre-defined functional networks. Homotopy, a metric of global connectivity, was measured by correlating GM voxels in the right hemisphere with corresponding left voxels, and averaging across GM. Group comparisons were made with Mann-Whitney U or chi-squared tests. Object oriented data analysis (OODA), post-hoc block permutation testing (p-values corrected for false discovery rate (FDR)), and group comparison of average within network correlations compared fc between cohorts. Bivariate correlations were described with Pearson's r. Significance was specified as p-value < .05, and the Benjamini-Hochberg procedure controlled for a final FDR of 0.05 for group comparison excluding permutation testing. Table 1 describes the 55 participants. There was no difference in cognition between cohorts (Table 1). Homotopy was lower in SCA (0.336 [0.310-0.379]) compared to controls (0.381 [0.341-0.394], p = 0.023). Figure 1 illustrates cohort fc matrices. The intra-network regions demonstrate strong positive correlations and inter-network regions demonstrate lower correlations for each cohort, indicating functional network architecture is similar between cohorts. However, the magnitude of fc is reduced in SCA. Using OODA, there was a significant difference between matrices (p=.001), with SCD selectively affecting the magnitude of fc within specific networks: Salience (Sal), Fronto-Parietal (FP), Cingulo-Opercular (CO), Sensory-Motor (SM), Sensory-Motor Lateral (SM-Lat), and Auditory (Aud) networks when comparing within network Pearson's r between cohorts (Table 1), and the Sal (FDR-corrected p < .001) and SM-Lat (FDR-corrected p = .039) networks using block permutation testing. Whole brain (Wb) OEF correlated with homotopy (r = -.420, p = .002), and the average Pearson r within specific networks: Default Mode (r = -.481, p < .001), CO (r = -.537, p < .001), Sal (r = -.549, p < .001), FP (r = -.461, p < .001), SM (r = -.298, p = .029), SM-Lat (r = -.381, p = .004), Aud (r = -.437, p = .001), and Visual (r = -.344, p = .011) networks (Figure 2). Wb OEF did not correlate with the average Pearson r within the Cerebellar (p = .713), Subcortical (p = .974), Memory (p = .104), Dorsal Attention (p = .228), or Ventral Attention (p = .162) networks. We conclude that there are differences in rs-fcMRI in this cohort of children with SCA unaffected by overt stroke or vasculopathy when compared to controls even though cohort differences were not found with cognitive testing. Differences were found in higher level cortical association systems (Sal, FP and CO) that are associated with executive function, which is known to be affected by SCA. Knowing that there is regional variation in metabolic stress, as measured by OEF, within the brains of children with SCA, we found an association between OEF and fc in select networks. These data suggest that those experiencing the greatest metabolic stress have diminished connectivity within select fc networks, and that these imaging metrics may provide neuroimaging biomarkers for cognitive decline in SCA. Disclosures Fields: Proclara Biosciences: Equity Ownership. Mirro:Nous Imaging Inc: Employment. King:Incyte: Consultancy; WUGEN: Equity Ownership; Tioma Therapeutics (formerly Vasculox, Inc.):: Consultancy; RiverVest: Consultancy; Novimmune: Research Funding; Amphivena Therapeutics: Research Funding; Bioline: Consultancy; Celgene: Consultancy; Cell Works: Consultancy; Magenta Therapeutics: Membership on an entity's Board of Directors or advisory committees.
Read full abstract