Access the full text.
Sign up today, get DeepDyve free for 14 days.
ORIGINAL RESEARCH ADULT BRAIN Effects of Arterial Stiffness on Cerebral WM Integrity in Older Adults: A Neurite Orientation Dispersion and Density Imaging and Magnetization Transfer Saturation Imaging Study J. Kikuta, K. Kamagata, M. Abe, C. Andica, Y. Saito, K. Takabayashi, W. Uchida, H. Naito, H. Tabata, A. Wada, Y. Tamura, R. Kawamori, H. Watada, and S. Aoki ABSTRACT BACKGROUND AND PURPOSE: Arterial stiffness is reported to be able to cause axonal demyelination or degeneration. The present study aimed to use advanced MR imaging techniques to examine the effect of arterial stiffness on the WM microstructure among older adults. MATERIALS AND METHODS: Arterial stiffness was measured using the cardio-ankle vascular elasticity index (CAVI). The high-CAVI (mean CAVI $ 9 points) and the low-CAVI groups (mean CAVI, 9 points) were created. The neuronal ﬁber integrity of the WM was evaluated by neurite orientation dispersion and density imaging and magnetization transfer saturation imaging. Tract-Based Spatial Statistics and the tracts-of-interest analysis were performed. Speciﬁc WM regions (corpus callosum, internal capsule, anterior thalamic radiation, corona radiata, superior longitudinal fasciculus, forceps minor, and inferior fronto-occipital fasciculus) were selected in the tracts-of-interest analysis. RESULTS: In Tract-Based Spatial Statistics, the high-CAVI group showed a signiﬁcantly lower myelin volume fraction value in the broad WM and signiﬁcantly higher radial diffusivity and isotropic volume fraction values in the corpus callosum, forceps minor, inferior fronto-occipital fasciculus, internal capsule, corona radiata, and anterior thalamic radiation than the low-CAVI group. In tracts-of-inter- est analysis using multivariate linear regression, signiﬁcant associations were found between the mean CAVI and radial diffusivity in the anterior thalamic radiation and the corona radiata; isotropic volume fraction in the anterior thalamic radiation and the corona radiata; and myelin volume fraction in the superior longitudinal fasciculus (P, .05). Additionally, partial correlation coefﬁcients were observed for the signiﬁcant associations of executive function with radial diffusivity and myelin volume fraction (P, .05). CONCLUSIONS: Arterial stiffness could be associated with demyelination rather than axonal degeneration. ABBREVIATIONS: ATR ¼ anterior thalamic radiation; CAVI ¼ cardio-ankle vascular elasticity index; CC ¼ corpus callosum; CR ¼ corona radiata; FA ¼ frac- tional anisotropy; FMi ¼ forceps minor; IC ¼ internal capsule; IFOF ¼ inferior fronto-occipital fasciculus; ISOVF ¼ isotropic volume fraction; NODDI ¼ neurite orientation dispersion and density imaging; MT ¼ magnetization transfer; MVF ¼ myelin volume fraction; RD ¼ radial diffusivity; SLF ¼ superior longitudinal fasciculus; TBSS ¼ Tract-Based Spatial Statistics; TMT ¼ Trail-Making Test; TOI ¼ tracts of interest rterial stiffness can be measured by several different methods. as an indirect index of arterial elasticity, but it is blood pressure– AThe pulse wave velocity has been measured by various methods dependent. Conversely, the cardio-ankle vascular elasticity index (CAVI) directly reflects vascular elasticity and is blood pressure–inde- pendent. Reports regarding the association between CAVI-measured Received July 12, 2022; accepted after revision October 15. From the Departments of Radiology (J.K., K.K., M.A., C.A., Y.S., K.T., W.U., A.W., S.A.), arterial stiffness and WM microstructure have still not been pub- Metabolism and Endocrinology (H.N., Y.T., R.K., H.W.), and Sportology Center (H.T., lished, though some studies reported the association between pulse Y.T., R.K., H.W.), Juntendo University Graduate School of Medicine, Tokyo, Japan; and Faculty of Health Data Science (C.A.), Juntendo University, Chiba, Japan. wave velocity–measured arterial stiffness and brain WM integrity. J. Kikuta and K. Kamagata contributed equally to this work. Neurite orientation dispersion and density imaging (NODDI) is This work is supported by the Strategic Research Foundation at Private Universities a new, advanced DWI technique that improves WM characterization (S1411006) and KAKENHI (18H03184, 18H02772, and 20K16737) from the Ministry of Education, Culture, Sports, Science and Technology of Japan. using a multicompartment model to describe different WM func- Please address correspondence to Junko Kikuta, MD, PhD, Department of tions. By enabling the estimation of neurite structure, NODDI can Radiology, Juntendo University Graduate School of Medicine, 2-1-1 Hongo, Bunkyo- ku, Tokyo 113-8421, Japan; e-mail: firstname.lastname@example.org provide more specific insight into the underlying WM microstruc- Indicates open access to non-subscribers at www.ajnr.org tural changes. Thus, NODDI could be useful in assessing arterial http://dx.doi.org/10.3174/ajnr.A7709 stiffness–associated WM microstructure, though no study has tested 1706 Kikuta Dec 2022 www.ajnr.org a Table 1: Demographic characteristics of the participants All Participants High-CAVI Low-CAVI High- vs Low-CAVI (n = 110) (n = 60) (n = 50) (P Value) Sex (men/female) 66:44 37:23 29:21 .70 Age (yr) 72.65 (SD, 4.90) 72.72 (SD, 5.00) 72.58 (SD, 4.82) .96 Mean CAVI 9.12 (SD, 0.84) 9.71 (SD, 0.57) 8.41 (SD, 0.46) ,.001 Antihypertensive therapy 71 43 28 .09 Education (yr) 14.31 (SD, 2.11) 14.30 (SD, 2.09) 14.32 (SD, 2.15) .96 Body mass index 22.54 (SD, 2.84) 22.79 (SD, 2.54) 22.24 (SD, 3.16) .31 Systolic blood pressure 137.47 (SD, 15.67) 139.60 (SD, 15.21) 134.92 (SD, 16.00) .12 Diastolic blood pressure 86.05 (SD, 9.35) 86.52 (SD, 8.49) 85.48 (SD, 10.34) .56 Heart rate 65.70 (SD, 23.73) 66.16 (SD, 33.88) 65.31 (SD, 9.85) .85 Montreal Cognitive Assessment (Japanese version) 25.19 (SD, 2.87) 25.31 (SD, 2.95) 25.04 (SD, 2.66) .61 Mini-Mental State Examination 27.93 (SD, 1.68) 28.07 (SD, 1.59) 27.76 (SD, 1.78) .34 TMT A 42.51 (SD, 13.90) 43.32 (SD, 14.24) 41.54 (SD, 13.56) .51 TMT B 115.82 (SD, 50.38) 116.78 (SD, 40.22) 114.66 (SD, 60.80) .83 TMT B minus A 73.31 (SD, 44.11) 73.47 (SD, 36.01) 73.12 (SD, 52.60) .97 Periventricular hyperintensity 1.14 (SD, 0.39) 1.17 (SD, 0.42) 1.10 (SD, 0.36) .372 Deep and subcortical WM hyperintensity 1.25 (SD, 0.53) 1.31 (SD, 0.60) 1.18 (SD, 0.48) .135 Data are means. this hypothesis. Moreover, myelin-sensitive imaging using MR imag- low-CAVI group (those with a mean CAVI of , 9points; 29 men and 21 women; mean age, 72.58 [SD, 4.82] years). Table 1 shows ing can evaluate the WM microstructure from a different viewpoint the demographic characteristics. Deep and subcortical WM hyper- from DWI. intensity and periventricular hyperintensity were evaluated using Arterial stiffness is indicated to cause axonal demyelination or 4,5 6 the Fazekas scale, according to axial FLAIR imaging. degeneration. Notably, Badji et al showed that carotid-femoral pulse wave velocity is significantly associated with both fractional anisotropy (FA) and radial diffusivity (RD) but not with the mye- Image Acquisition lin volume fraction (MVF). The results suggested that arterial stiff- MR imaging data were acquired using a 3T MR imaging scanner (Magnetom Prisma; Siemens) with a 64-channel head coil. We ness is associated with axonal degeneration rather than with acquired multishell DWI data using a spin-echo echo-planar imag- demyelination. However, reports supporting such results are still limited. We hypothesized that the impact of arterial stiffness on ing sequence, which included 2 b-values of 1000 and 2000 s/mm the brain WM microstructure could be better understood by using along 64 isotropic diffusion gradients uniformly distributed on a the above-mentioned multimodal WM-sensitive MR imaging sphere, with a simultaneous multisection echo-planar imaging techniques. Hence, this study aimed to explore the associations sequence in the anterior-posterior phase-encoding direction with between CAVI-measured arterial stiffness and WM-sensitive MR the following parameters: TR ¼ 3300 ms; TE ¼ 70 ms; FOV ¼ 229 229 mm; matrix size ¼ 130 130; section thickness ¼ 1.8 mm; imaging measures of the brain in older adults. resolution ¼ 1.8 1.8 mm; acquisition time ¼ 7minutes and 29 seconds. DWI acquisition was completed with a b ¼ 0image. MATERIALS AND METHODS Standard and antiphase-encoded blipping images were acquired The institutional review board of Juntendo University Hospital in without diffusion weighting to compensate for the distortion caused Japan approved this study in compliance with the World Medical by the magnetic susceptibility associated with the echo-planar imag- Association’s Code of Ethics (Declaration of Helsinki) for experi- ing acquisition. The predominant T1-weighted, proton density– ments involving humans. weighted, and magnetization transfer (MT)–weighted images were obtained using a 3D multiecho high-speed low-angle shot sequence Study Participants for calculating the MT saturation index. Thesettings for theMT The Bunkyo Health Study is a prospective cohort study of 1629 saturation sequences were as follows: for MT and MT scanning, older individuals. Of these, 160 participants underwent both off on TE ¼ 2.53 ms, TR ¼ 24 ms, flip angle ¼ 5°;for T1WI,TE ¼ FLAIR imaging and DWI. Exclusion criteria included major psy- 2.53 ms, TR ¼ 10 ms, flip angle ¼ 13°, with parallel imaging using chiatric or neurologic disorders, heart failure, stroke, and/or a his- generalized autocalibrating partially parallel acquisition with a fac- tory of alcohol or drug abuse. Ultimately, 110 older participants tor of 2 in the phase-encoding direction, 7/8 partial Fourier were included for the analysis. Arteriosclerosis was estimated by acquisition in the partition direction, bandwidth ¼ 260 Hz/pixel, CAVI determined by using an automatic waveform analyzer matrix ¼ 128 128, acquisition time ¼ 6 minutes 25 seconds, sec- (Vascular Screening System VaSera VS1500; Fukuda Denshi). tion thickness ¼ 1.8 mm, FOV ¼ 224 224 mm. High CAVI ($9.0) implies progression of carotid and coronary ar- teriosclerosis; thus, CAVI 9 was set as the cutoff value. The eligible participants were divided into the high-CAVI group (those with a Diffusion MR Imaging Processing mean [average of left and right values] CAVI of $ 9points; 37 For eliminating artifacts, the eddy (https://fsl.fmrib.ox.ac.uk/fsl/ men and 23 women; mean age, 72.72 [SD, 5.00] years) and the fslwiki/eddy/UsersGuide)and topup (https://fsl.fmrib.ox.ac.uk/ AJNR Am J Neuroradiol 43:1706–12 Dec 2022 www.ajnr.org 1707 Table 2: Summary of WM metrics Diffusion MR Imaging Parameter Explanation DTI FA Overall direction of water diffusion in brain tissue Mean diffusivity The magnitude of isotropic diffusion in brain tissue Axial diffusivity The coefﬁcient of diffusion across the long axis of the ellipsoid RD The coefﬁcient of diffusion perpendicular to the long axis NODDI Intracellular volume fraction Neurite density based on intracellular diffusion Orientation dispersion index Dispersion of neurites in the intracellular compartment ISOVF The measure of extracellular water diffusion fsl/fslwiki/topup) toolboxes, which are part of the FSL (www. 2a proportional to sina and sin 2a, respectively. The ratio of the 2 fmrib.ox.ac.uk/fsl), were used. The resulting images were fitted acquisitions was calculated using the following formula: to the NODDI model using the NODDI Matlab Toolbox 5 sina 1 (http://www.nitrc.org/projects/noddi_toolbox). Table 2 summa- ¼ : sin2a 2cosa rizes the parameters of DTI and NODDI. The maps of the orienta- tion dispersion index, isotropic volume fraction (ISOVF), and From there, the local flip angle a was calculated. intracellular volume fraction were generated using the Accelerated Microstructure Imaging via Convex Optimization. Furthermore, Tract-Based Spatial Statistics Analysis the DTIFit tool (https://open.win.ox.ac.uk/pages/fsl/fslpy/fsl.data. Voxelwise statistical analysis was performed using Tract-Based dtifit.html) was used to generate tensor-derived maps according to Spatial Statistics (TBSS; http://fsl.fmrib.ox.ac.uk/fsl/fslwiki/TBSS) the ordinary least-squares method using DWI data with b-values implemented in FSL. The TBSS procedure was as follows: First, of 0 and 1000 s/mm . using FMRIB’s Nonlinear Registration Tool (FNIRT; http://fsl. fmrib.ox.ac.uk/fsl/fslwiki/FNIRT), we aligned the FA maps of all Myelin-Sensitive Imaging Processing participants into the Montreal Neurological Institute 152 standard MT saturation (MT )data were analyzed using a Matlab script sat space with 1 1 1mm voxel size. Second, we created and (https://www.mathworks.com/help/matlab/ref/run.html). First, thinned a population-based mean FA image to establish the mean the apparent longitudinal relaxation rate (R ) was calculated as 1app FA skeleton, which shows the centers of all tracts common to the the following equation: group. The threshold of the mean FA skeleton was 0.2 to exclude the peripheral tracts and GM. Third, a binary mask of the FA ske- 1 S a =TR S a =TR T1 T1 T1 PD PD PD R ¼ : 1app letonizedimage was usedas the mask image to make a voxel-by- 2 S =a S =a PD PD T1 T1 voxel statistical comparison between the high- and low-CAVI Here, S and S indicate the signal strength of T1- and pro- groups of the 4D skeleton image file. This comparison was per- T1 PD ton-density (PD)-weighted imaging, respectively. TR and TR formed using FSL’s Randomise tool (http://fsl.fmrib.ox.ac.uk/fsl/ T1 PD reflect the TRs of T1- and PD-weighted imaging, respectively. fslwiki/Randomise/UserGuide), and the number of permutations a and a show the excitation flip angles of T1- and PD- T1 PD was set to 10,000. The data for other WM metrics was analyzed weighted imaging. Second, the apparent signal amplitude (A ) app using the tbss_non_FA script (https://fsl.fmrib.ox.ac.uk/fsl/fslwiki/ was calculated as the following: TBSS/UserGuide#Using_non-FA_Images_in_TBSS) to generate a 4D skeleton imagefilefor each metric. TR a =a TR a =a PD T1 PD T1 PD T1 A ¼ S S : app PD T1 S TRa S TR a T1 T1 PD T1 PD Tract-of-Interest Analysis Tracts of interest (TOI) were analyzed using the ICBM DTI-81 Third, the apparent d was calculated as the following app Atlas (http://www.bmap.ucla.edu/portfolio/atlases/ICBM_DTI-81_ equation: Atlas/). According to the TBSS results (Figure) and previous stud- d ¼ðA a =S 1ÞR TR a 2=2: app app MT MT 1app MT MT ies showing arterial stiffness–associated WM regions, 6 major WM tracts, namely, the corpus callosum (CC), internal capsule (IC), co- TR ,S ,and a show the TR, signal intensity, and excita- MT MT MT rona radiata (CR), inferior fronto-occipital fasciculus (IFOF), for- tion flip angle of the MT-weighted imaging respectively. The fol- ceps minor (FMi), and anterior thalamic radiation (ATR), were lowing formula was applied to fix the small residual high-order identified as TOI of RD and ISOVF. Additionally, 7 major WM dependency of MT on the local radiofrequency (RF) transmit sat tracts, namely, the CC, IC, CR, IFOF, FMi, ATR, and superior lon- field: gitudinal fasciculus (SLF), were selected as TOI of the MVF. These WM regions are reportedly vulnerable to increased arterial stiff- d ð1 0:4Þ app 4,5,20,21 MT ¼ : sat ness. Then, the mean value (the average of left and right val- 1 0:4 RF local ues) within each ROI was computed for each WM metric. RF was calculated using the dual-angle method. In addition, local we added 2 B maps with flip angles of 10° and 20°, respectively, Statistical Analysis All statistical data were analyzed using SPSS Statistics, Version 27 obtained by echo-planar imaging in about 10 seconds. The first and second images were obtained after excitation with flip angles a and (IBM). Demographic and clinical data were analyzed using the 1708 Kikuta Dec 2022 www.ajnr.org FIGURE. Comparison between the high- and low-CAVI groups. For TBSS, the low- and high-CAVI groups were compared (family-wise error– corrected P, .05, adjusting for age, sex, antihypertensive therapy use, intracranial volume, and systolic blood pressure). Red–yellow voxels demonstrate signiﬁcantly higher RD and ISOVF values in the high-CAVI group than in the low-CAVI group. Blue–light blue voxels illustrate a signiﬁcantly lower MVF value in the high-CAVI group than in the low-CAVI group. x or Mann-Whitney U test. A P value (2-tailed) , .05 was con- CAVI group. However, age, sex, education level, body mass index, sidered statistically significant. heart rate, systolic blood pressure, diastolic blood pressure, and anti- For TBSS, the high- and low-CAVI groups were compared hypertensive therapy history; the Mini-Mental State Examination; using the Randomize tool (family-wise error–corrected P, .05, the Montreal Cognitive Assessment (Japanese version); Trail adjusting for age, sex, antihypertensive therapy use, systolic blood Making Test (TMT) A, TMT B, TMT B minus A; deep and subcort- pressure, and intracranial volume). ical WM hyperintensity; and periventricular hyperintensity were not In all participants, we applied univariate linear regression analy- significantly different between the 2 groups. ses for each WM metric as a dependent variable and the mean Whole-Brain Analysis CAVI as an independent variable. Subsequently, multivariate linear TBSS results identified a significantly lower MVF in the high- regression analyses were conducted using backward linear regression CAVI group than in the low-CAVI group in the broad WM area to identify independent factors associated with the mean CAVI. The (family-wise error–corrected P, .05; Figure). RD was signifi- variable P,.2 in theunivariatemodel was included in the back- cantly higher in the high-CAVI group in specific WM areas such removal procedure with P-removal ¼ .1.Age,sex,antihypertensive as the CC, FMi, bilateral IFOF, bilateral ATR, bilateral CR, and therapy, systolic blood pressure, and intracranial volume were con- left IC compared with the low-CAVI group. The ISOVF was sig- sidered confounding covariates to separate from the strength of the nificantly higher in the high-CAVI group than in the low-CAVI relation between the mean CAVI and WM integrity. group in the CC, FMi, right IC, right CR, right ATR, and right Additionally, the partial correlation analyses between each WM IFOF. Most interesting, RD and ISOVF changes were relatively metric and the cognitive performance scores were examined sepa- overlapped, mainly observed in the anterior area. Whereas FA, rately for all participants, the high-CAVI group, and the low-CAVI mean diffusivity, axial diffusivity, the orientation dispersion group, adjusting for age, sex, and education level. Multiple com- index, and intracellular volume fraction did not significantly dif- parisons were corrected using the false discovery rate procedure fer between 2 groups. for each WM metric and region. The false discovery rate–cor- rected P, .05 was considered significant. Tract-Specific Analysis The univariate linear regression analysis revealed the significant associations of the mean CAVI with RD in the FMi; ISOVF in RESULTS the CC, CR, and FMi; and MVF in the CR, IC, FMi, IFOF, ATR, Participant Characteristics Table 1 shows all participant characteristics. The high-CAVI and SLF (false discovery rate–corrected P, .05; Table 3). In the group showed a significantly higher mean CAVI than the low- multivariate linear regression analyses adjusted for age, sex, AJNR Am J Neuroradiol 43:1706–12 Dec 2022 www.ajnr.org 1709 Table 3: Univariate and multivariate linear regression analyses adjusted for age, sex, anti- In all participants, the partial correla- hypertensive therapy, systolic blood pressure, and intracranial volume for the associa- tion analyses demonstrated significant tion of mean CAVI with WM metrics in specific regions associations of TMT B with RD in the Univariate Linear Regression Multivariate Linear Regression CR,FMi,and IC,or MVF in alltested P b P b WM regions (false discovery rate–cor- RD rected P, .05; Table 4). In the low- ATR .121 0.149 .015 1.549 CAVI group, TMT B was significantly CC .075 0.171 associated with RD in the FMi and MVF CR .079 0.168 .004 0.737 in theCR (falsediscovery rate–corrected FMi .01 0.245 .061 0.296 P , .05; Table 4). Furthermore, the par- IC .279 0.104 IFOF .163 0.134 tial correlation coefficients for the signifi- ISOVF cant associations of TMT B minus A ATR .16 0.135 .014 1.531 with RD in the CR, IC, and FMi and CC .036 0.2 with MVF in the ATR, CR, and FMi CR .031 0.206 .005 0.596 were noted in all participants (false dis- FMi .037 0.2 IC .271 0.106 covery rate–corrected P , .05; Table 5). IFOF .123 0.148 Meanwhile, the low-CAVI group had MVF partial correlation coefficients for the ATR .005 0.265 significant associations of TMT B minus CC .174 0.13 A with MVF in the CC, CR, FMi, IC, CR .015 0.231 FMi .028 0.21 IFOF, and SLF (false discovery rate–cor- IC .009 0.249 rected P , .05; Table 5). However, the IFOF .019 0.224 partial correlation coefficients showed no SLF .015 0.231 .036 0.218 significant associations among the Mini- Mental State Examination, the Montreal Cognitive Assessment (Japanese version), Table 4: Partial correlation coefficients between WM metrics and TMT B, adjusted for a and TMT A and WM metrics. age, sex, and education All Participants High-CAVI Group Low-CAVI Group DISCUSSION Corrected Corrected Corrected The present study evaluated the WM P Value r P Value r P Value r microstructural changes in older adults RD ATR .818 0.023 .898 0.017 .545 0.090 with arterial stiffness. The major find- CC .121 0.160 .658 0.078 .291 0.174 ings are as follows: First, whole-brain CR .046 0.220 .658 0.080 .107 0.307 voxelwise results identified a signifi- FMi .016 0.288 .650 0.135 .013 0.436 cantly lower MVF in the high-CAVI IC .046 0.221 .650 0.152 .276 0.197 group than in the low-CAVI group in IFOF .050 0.206 .650 0.121 .246 0.228 ISOVF the broad WM regions. TBSS results ATR .866 0.016 .983 0.008 .865 0.025 also showed significantly higher RD and CC .669 0.094 .983 0.003 .437 0.183 ISOVF in the high-CAVI group than in CR .464 0.139 .983 0.055 .272 0.250 the low-CAVI group in the CC, FMi, FMi .464 0.147 .983 0.044 .256 0.297 IFOF, IC, CR, and ATR. Second, the IC .822 0.059 .983 0.058 .865 0.047 IFOF .861 0.035 .983 0.108 .788 0.095 multivariate linear analysis noted signifi- MVF cant associations of the mean CAVI ATR .042 0.200 .549 0.116 .054 0.283 with RD and ISOVF in the ATR and CC .042 0.200 .659 0.060 .051 0.324 CR; and with MVF in the SLF. Finally, CR .022 0.275 .549 0.145 .027 0.414 FMi .026 0.235 .614 0.086 .051 0.306 we found partial correlation coefficients IC .033 0.219 .549 0.134 .054 0.283 for the significant associations between ISOF .022 0.250 .549 0.121 .051 0.331 the executive function scores and RD SLF .022 0.250 .549 0.205 .051 0.312 and MVF in specific WM areas. False discovery rate–corrected P value,.05. Whole-brain voxelwise results revealed that the high-CAVI group had a significantly lower MVF value than antihypertensive therapy, systolic blood pressure, and intracranial the low-CAVI group in the broad WM area. Low MVF values volume, the mean CAVI was significantly associated with RD indicate the loss of the myelin sheath insulating the nerves, and ISOVF in the ATR and CR, and the MVF in the SLF (false implying WM demyelination. In this study, the high-CAVI discovery rate–corrected P, .05; Table 3). group also had higher RD and ISOVF values than the low-CAVI 1710 Kikuta Dec 2022 www.ajnr.org Table 5: Partial correlation coefficients between WM metrics and TMT B minus A, adjusted participants was lower than that in the for age,sex,and education study of Badji et al (26.1 ([SD, 4.23] kg/ All Participants High-CAVI Group Low-CAVI Group m ). However, the average value of sys- Corrected Corrected Corrected tolic blood pressure (137.47 [SD, P Value r P Value r P Value r 15.67] mm Hg) in our study was higher RD than that in the study of Badji et al ATR .286 0.104 .280 0.172 .846 0.029 (125.66 [SD, 11.65] mm Hg). Suzuki et CC .060 0.199 .270 0.191 .508 0.127 al showed that pathologic processes CR .019 0.262 .270 0.200 .162 0.283 FMi .006 0.315 .270 0.265 .143 0.329 related to hypertension are associated IC .046 0.219 .270 0.203 .508 0.120 with image differences, suggesting IFOF .136 0.154 .419 0.109 .508 0.164 changes inWM axons. Inaddition, WM ISOVF integrity is particularly vulnerable to ATR .506 0.065 .333 0.187 .774 0.053 CC .233 0.153 .344 0.156 .774 0.099 obesity. A higher body mass index is CR .201 0.178 .333 0.178 .774 0.156 associated with lower FA in the FMi and FMi .120 0.225 .333 0.230 .774 0.176 29 CC. Therefore, except for the degree IC .395 0.095 .333 0.176 .774 0.043 of arterial stiffness, the difference in par- IFOF .386 0.110 .526 0.102 .774 0.051 ticipants’ physical characteristics may MVF ATR .038 0.232 .817 0.207 .082 0.256 influence WM integrity. However, exact CC .088 0.166 .855 0.025 .017 0.390 matching of these clinical findings is dif- CR .038 0.265 .817 0.114 .008 0.460 ficult. Additionally, our participants FMi .038 0.232 .817 0.074 .019 0.363 were all Japanese, whereas the partici- IC .057 0.192 .817 0.083 .037 0.314 IFOF .051 0.211 .847 0.048 .017 0.388 pants of Badji et al were all Canadian; SLF .057 0.191 0.817 0.088 .019 0.358 thus, racial differences may affect WM False discovery rate–corrected P value,.05. integrity. The present study also showed the group. High ISOVF values associated with high CAVI suggest existence of the partial correlation coefficients for the significant increased extracellular water diffusion and have been shown to associations between the executive function and WM metrics. In be related to increased inflammatory activation or blood-brain bar- all participants and the low-CAVI group, the MVF in several rier permeability. Moreover, elevated RD is also associated with WM areas was negatively associated with the executive function. demyelination. From these results, the WM in the high-CAVI These findings suggest that executive dysfunction may be associ- ated with demyelination. However, in the high-CAVI group of group could involve demyelination more than in the low-CAVI the study, there was no significant connection between WM met- group. Furthermore, there were no significant differences in the degree of WM hyperintensities between the high- and low-CAVI rics and TMT scores, which are measures of executive function. Furthermore, although there was no difference in the TMT scores groups in this study. Considering that TBSS showed significant dif- between the high- and low-CAVI groups, we found significant ferences in RD, ISOVF, and MVF between the two groups, our differences in WM metrics between two groups. The findings results suggest that WM microstructural changes precede WM imply that WM microstructural changes may have already hyperintensities and brain atrophy. In support of our findings, prior occurred before the impairment of executive function. studies have indicated that DTI metrics capture ultrastructural This study has some limitations. First, it is based on data changes in WM before the onset of WM hyperintensities and brain 26,27 obtained from Japanese individuals living in the city; hence, bias in atrophy. Therefore, it is thought that there was no correlation genetic factors and environmental factors, such as eating habits, between the degree of WM hyperintensities and WM metrics. may occur. Second, it was conducted in a single facility. Thus, mul- In the multivariate linear analyses, the mean CAVI was signifi- ticenter and epidemiologic studies are needed to examine further cantly associated with RD in the ATR, CR, and FMi; ISOVF in the these investigations. ATR and CR; and MVF in the SLF. These findings could reflect ar- teriosclerosis relating to demyelination. In previous studies, the WM microstructure has been reported to be vulnerable to circula- CONCLUSIONS 4-6,20 tory alterations and correlates with arterial stiffness. Notably, 6 Arterial stiffness could be strongly associated with demyelination Badji et al showed that carotid-femoral pulse wave velocity was sig- 17 rather than axonal degeneration. nificantly associated with both FA and RD but not with MVF. The present study indicated the significant associations between the Disclosure forms provided by the authors are available with the full text and mean CAVI and not only DTI and NODDI metrics but also MVF, PDF of this article at www.ajnr.org. possibly reflecting the progression of arteriosclerosis exacerbating demyelination. The conflicting results regarding MVF between this REFERENCES study and that of Badji et al might be caused by differences in the 1. Lim J, Pearman M, Park W, et al. Interrelationships among various characteristics of the target cohort. For instance, in our study, the measures of central artery stiffness. Am J Hypertens 2016;29:1024– average value of the body mass index (22.54 [SD, 2.84] kg/m )in all 28 CrossRef Medline AJNR Am J Neuroradiol 43:1706–12 Dec 2022 www.ajnr.org 1711 2. Matsushita K, Ding N, Kim ED, et al. Cardio-ankle vascular index and T1 relaxation obtained from 3D FLASH MRI. Magn Reson Med and cardiovascular disease: systematic review and meta-analysis 2008;60:1396–407 CrossRef Medline of prospective and cross-sectional studies. J Clin Hypertens 16. Weiskopf N, Suckling J, Williams G, et al. Quantitative multi-pa- (Greenwich) 2019;21:16–24 CrossRef Medline rameter mapping of R1, PD , MT, and R2 at 3T: a multi-center validation. Front Neurosci 2013;7:95 CrossRef Medline 3. Zhang H, Schneider T, Wheeler-Kingshott CA, et al. NODDI: practi- 17. Morrell GR, Schabel MC. An analysis of the accuracy of magnetic reso- cal in vivo neurite orientation dispersion and density imaging of nance flip angle measurement methods. Phys Med Biol 2010;55:6157– the human brain. NeuroImage 2012;61:1000–16 CrossRef Medline 74 CrossRef Medline 4. Maillard P, Mitchell GF, Himali JJ, et al. Aortic stiffness, increased 18. Smith SM, Jenkinson M, Johansen-Berg H, et al. Tract-Based Spatial white matter free water, and altered microstructural integrity: a Statistics: voxelwise analysis of multi-subject diffusion data. continuum of injury. Stroke 2017;48:1567–73 CrossRef Medline Neuroimage 2006;31:1487–505 CrossRef Medline 5. Tarumi T, de Jong DL, Zhu DC, et al. Central artery stiffness, baror- 19. Mori S, Oishi K, Jiang H, et al. Stereotaxic white matter atlas based eflex sensitivity, and brain white matter neuronal fiber integrity in on diffusion tensor imaging in an ICBM template. Neuroimage older adults. Neuroimage 2015;110:162–70 CrossRef Medline 2008;40:570–82 CrossRef Medline 6. Badji A, Noriega de la Colina A, Karakuzu A, et al. Arterial stiffness 20. Maillard P, Mitchell GF, Himali JJ, et al. Effects of arterial stiffness and white matter integrity in the elderly: a diffusion tensor and on brain integrity in young adults from the Framingham heart magnetization transfer imaging study. Neuroimage 2019;186:577– study. Stroke 2016;47:1030–36 CrossRef Medline 85 CrossRef Medline 21. Tamura Y, Shimoji K, Ishikawa J, et al. Subclinical atherosclerosis, 7. Someya Y, Tamura Y, Kaga H, et al. Skeletal muscle function and need vascular risk factors, and white matter alterations in diffusion ten- for long-term care of urban elderly people in Japan (the Bunkyo sor imaging findings of older adults with cardiometabolic diseases. Health Study): a prospective cohort study. BMJ Open 2019;9:e031584 Front Aging Neurosci 2021;13:712385 CrossRef Medline CrossRef Medline 22. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a 8. Shirai K, Utino J, Otsuka K, et al. A novel blood pressure-independ- practical and powerful approach to multiple testing. Journal of the ent arterial wall stiffness parameter: cardio-ankle vascular index Royal Statistical Society: Series B (Methodological) 1995;57:289–300 (CAVI). J Atheroscler Thromb 2006;13:101–07 CrossRef Medline 23. Duval T, Stikov N, Cohen-Adad J. Modeling white matter micro- 9. Izuhara M, Shioji K, Kadota S, et al. Relationship of cardio-ankle structure. Funct Neurol 2016;31:217–28 CrossRef Medline vascular index (CAVI) to carotid and coronary arteriosclerosis. 24. Suzuki H, Gao H, Bai W, et al. Abnormal brain white matter micro- Circ J 2008;72:1762–67 CrossRef Medline structure is associated with both pre-hypertension and hyperten- 10. Fazekas F, Chawluk JB, Alavi A, et al. MR signal abnormalities at 1.5 sion. PLoS One 2017;12:e0187600 CrossRef Medline Tin Alzheimer’s dementia and normal aging. AJR Am J Roentgenol 25. Song SK, Yoshino J, Le TQ, et al. Demyelination increases radial diffu- 1987;149:351–56 CrossRef Medline sivity in corpus callosum of mouse brain. Neuroimage 2005;26:132–40 11. Kamagata K, Zalesky A, Yokoyama K, et al. MR g-ratio-weighted CrossRef Medline connectome analysis in patients with multiple sclerosis. Sci Rep 26. Andica C, Kamagata K, Hatano T, et al. Free-water imaging in white 2019;9:13522 CrossRef Medline and gray matter in Parkinson’s disease. Cells 2019;8:839 CrossRef 12. Yamada H, Abe O, Shizukuishi T, et al. Efficacy of distortion correction Medline on diffusion imaging: comparison of FSL Eddy and Eddy_Correct 27. Maillard P, Carmichael O, Harvey D, et al. FLAIR and diffusion using 30 and 60 directions diffusion encoding. PLoS One 2014;9: MRI signals are independent predictors of white matter hyperin- e112411 CrossRef Medline tensities. AJNR Am J Neuroradiol 2013;34:54–61 CrossRef Medline 13. Daducci A, Canales-Rodríguez EJ, Zhang H, et al. Accelerated 28. Bettcher BM, Walsh CM, Watson C, et al. Body mass and white mat- Microstructure Imaging via Convex Optimization (AMICO) from ter integrity: the influence of vascular and inflammatory markers. diffusion MRI data. Neuroimage 2015;105:32–44 CrossRef Medline PLoS One 2013;8:e77741 CrossRef Medline 14. Basser PJ, Mattiello J, LeBihan D. Estimation of the effective self-diffu- 29. Kullmann S, Schweizer F, Veit R, et al. Compromised white matter sion tensor from the NMR spin echo. JMagn Reson B 1994;103:247– integrity in obesity. Obes Rev 2015;16:273–81 CrossRef Medline 54 CrossRef Medline 30. Liu G, Allen B, Lopez O, et al. Racial differences in gray matter in- 15. Helms G, Dathe H, Kallenberg K, et al. High-resolution maps of mag- tegrity by diffusion tensor in black and white octogenarians. Curr netization transfer with inherent correction for RF inhomogeneity Alzheimer Res 2015;12:648–54 CrossRef Medline 1712 Kikuta Dec 2022 www.ajnr.org
American Journal of Neuroradiology – American Journal of Neuroradiology
Published: Dec 1, 2022
Access the full text.
Sign up today, get DeepDyve free for 14 days.