Association Between Gray Matter Volume Reduction And Cognitive Impairment In Neuromyelitis Optica

BACKGROUND AND PURPOSE:Whether gray matter impairment occurs in neuromyelitis optica(NMO)is a matter of ongoing debate,and the association of gray matter impairment with cognitive deficits remains largely unknown.By recruiting a large sample of NMO patients(n=50),we performed a voxel-based morphometry(VBM)analysis and a battery of cognitive assessment to investigate gray matter volume(GMV)reductions and their association with cognitive decline and clinical variables,in order to elucidate patterns of gray matter abnormalities in NMO patients and their clinical associations.MATERIALS AND METHODS:A total of 50 right-handed NMO patients who fulfilled the revised Wingerchuk diagnostic criteria for NMO and 50 sex-,age-,handedness-and education-matched healthy controls(HCs)were included in this study.Sagittal 3D T1-weighted images and axial fast spin-echo T2-weighted images were acquired using a GE 3.0-Tesla MR system.All of the participations completed a battery of cognitive assessments.For each patient,we carefully recorded information of the onset age,the relapsing frequency,the number of attack,the disease duration,the expanded disability status scale(EDSS)scores and the seropositive for anti-aquaporin 4(AQP4)antibody.The brain lesion volumes(BLV)were measured by two radiologists using mricron software.1.All preprocessing steps for 3D-T1 high-resolution structural data were carried out using the statistical parametric mapping(SPM8,http://www.fil.ion.ucl.ac.uk/spm)package based on Matlab platform.The preprocessing of the 3D-T1 high resolution structural data including segmentation,spatial normalization and smoothing with a Gaussian kernel of 6 × 6 × 6 mm3 full-width at half maximum.2.GMV differences were compared between NMO group and HC group.The voxel-based comparisons were performed to identify the brain regions with significant group differences in GMV using a two-sample two-tailed t-test,and the multiple comparisons were corrected using the false discovery rate(FDR)method(P < 0.05),cluster size > 200 voxels.3.The voxel-based partial correlation analysis was performed to test the relationships of the GMV with clinical parameters within brain regions that exhibited significant group differences in GMV in the NMO patients,and the correlation between GMV and BLV was investigated in 37 NMO patients with abnormal brain MRIs.Age,gender and years of education were entered as covariates of no interest.Multiple comparisons were corrected by the Alphasim method(single voxel P = 0.01,5000 simulations,full-width at half maximum = 6 mm,cluster connection radius = 2.5 mm,within a significant GM mask,which resulted in a corrected threshold of P < 0.05 and a cluster size threshold of > 33 voxels).4.A general linear model was implemented to compare differences in the cognitive scores between the NMO and the HC groups using SPSS 18.0(P < 0.05).The voxel-based partial correlation analysis was performed to test the relationships of the GMV with cognitive scores within brain regions that exhibited significant group differences in GMV in the NMO patients.Age,gender and years of education were entered as covariates of no interest.BLV was also added as an additional nuisance covariate to explore whether it would affect the correlation between GMV and cognitive scores.Multiple comparisons were corrected by the Alphasim method(single voxel P = 0.01,5000 simulations,full-width at half maximum = 6 mm,cluster connection radius = 2.5 mm,within a significant GM mask,which resulted in a corrected threshold of P < 0.05 and a cluster size threshold of > 33 voxels).RESULTS:1.Compared to healthy controls(635.9 ml ± 51.18 ml),NMO patients(602.8 ml ±51.03 ml)had a 5.21% decrease in the mean GMV of the whole brain(P = 0.002).The significant GMV reduction in NMO affected the frontal and temporal cortices,as well as the right thalamus(FDR correction,P < 0.05).No significantly increased GMV was found in the NMO patients.2.In the NMO patients,the EDSS scores had a negative correlation(Alphasim correction,P < 0.05)with the GMV in the left inferior temporal gyrus(ITG)(pr =-0.500,P < 0.001)and the right medial prefrontal cortex(MPFC)(pr =-0.465,P =0.001).Voxel-based analysis revealed that the NMO patients exhibited significant negative correlations(Alphasim correction,P < 0.05)between BLV and GMV in the right thalamus(pr =-0.376,P = 0.022)and left ITG(pr =-0.510,P = 0.001).When we excluded the two patients with visible lesions in these two regions,the correlations between BLV and GMV in the right thalamus(pr =-0.380,P = 0.024)and left ITG(pr =-0.499,P = 0.002)remained significant.No significant correlations were observed between the relapsing frequency,disease duration,onset age and GMV.3.The cognitive impairments in NMO were deficits in short-and long-term memory,speed of information processing and verbal fluency on semantic stimuli.Voxel-based analysis revealed that the NMO patients had a significantly positive correlation(Alphasim correction,P < 0.05)between the Immediate Recall of trial 2 and the GMV of the medial prefrontal cortex(pr = 0.563,P < 0.001).The Long-Delay(pr =0.420,P = 0.005)and Short-Delay(pr = 0.401,P = 0.007)Free Recalls also had positive correlations with the GMV of the right thalamus.These cognitive scores did not show any negative correlations with the regional GMV in NMO patients.No significant correlations were observed between other cognitive scores and the GMV.Even after further controlling for the effect of BLV,these correlations remained significant.CONCLUSIONS:1.GMV reduction is present in NMO patients.We found GMV reductions in the frontal and temporal lobes and the thalamus,the range of which largely surpassed the brain regions connected to the spinal cord and optic nerves.Therefore,the mechanism of axonal degeneration secondary to the damage of the spinal cord and optic nerves cannot fully explain these extensive reductions in the GMV.2.GMV reduction is an important pathological process in NMO patients,which may be related to disease severity.Our findings also suggest a potential role of GMV reductions in predicting the severity of the disorder.Besides that,we found negative correlations between the GMVs of the right thalamus and left ITG and the BLV,even after excluding patients with visible lesions in these regions.This finding suggests that the GMV reductions in these regions are secondary to brain lesions,which may be mediated by a mechanism of axonal degeneration.However,the GMV reduction in the MPFC was not correlated with the BLV,indicating that the axonal degeneration mechanism cannot explain the medial prefrontal cortex reduction observed in NMO.We suggest that both neurodegeneration and axonal degeneration mechanisms may contribute to GMV reductions in NMO patients.3.The cognitive impairments in NMO were deficits in short-and long-term memory,speed of information processing and verbal fluency on semantic stimuli.The frontal cortex and thalamus are closely associated with cognitive function.The GMV reductions in these regions may affect cognitive performance.