DETECTION OF SMALL INFARCTS IN THE CAUDATE NUCLEUS ON 7 TESLA MRI: THE SMART-MR STUDY

DETECTION OF SMALL INFARCTS IN THE CAUDATE NUCLEUS ON 7 TESLA MRI: THE SMART-MR STUDY

Poster Presentations: Sunday, July 16, 2017 P441 Table 1 Demographic and clinical characteristics of study participants Subjects for classifier trai...

895KB Sizes 19 Downloads 62 Views

Poster Presentations: Sunday, July 16, 2017

P441

Table 1 Demographic and clinical characteristics of study participants Subjects for classifier training

N Age, y Age of onset, year Women Education, y APOE ε4 present* MMSE Vascular risk factors DM Hypertension Hyperlipidemia History of IHD History of stroke

aMCI patients

AD patients

CN individuals AD patients p-value Total

Nonconverters Converters p-value Total

Slow decliners

Fast decliners

p-value

869 65.4 (9.0) 599 (68.9) 11.7 (4.9) 135 (22.8) 28.5 (2.0)

473 73.0 (9.4) 68.5 (10.7) 307 (64.9) 9.4 (5.3) 180 (55.6) 18.2 (5.5)

< 0.001 0.133 < 0.001 < 0.001 < 0.001

79 69.7 (8.8) 67.1 (8.6) 46 (58.2) 12.1 (4.6) 29 (39.2) 26.4 (2.4)

53 (67.1) 69.1 (9.0) 66.3(8.6) 29 (54.7) 12.1 (4.4) 15 (29.4) 27.0 (1.9)

26 (32.9) 70.9 (8.6) 68.5 (8.4) 17 (65.4) 12.1 (5.0) 14 (60.9) 25.4 (3.0)

0.410 0.308 0.366 0.971 0.010 0.015

27 70.4 (7.6) 67.2 (7.7) 18 (66.7) 10.3 (5.1) 4 (66.7) 21.4 (3.1)

14 (51.9) 72.0 (6.6) 68.0 (6.9) 10 (71.4) 9.5 (6.2) 2 (50.0) 21.1 (3.0)

13 (48.1) 68.8 (8.4) 66.4 (8.7) 8 (61.5) 11.2 (3.6) 2 (100.0) 21.7 (3.3)

0.276 0.595 0.695 0.392 0.467 0.616

178 (20.5) 260 (29.9) 238 (27.4) 110 (12.7) 32 (3.7)

112 (23.7) 206 (43.6) 90 (19.0) 43 (9.1) 26 (5.5)

0.174 < 0.001 0.001 0.049 0.118

35 (44.3) 31 (39.2) 25 (31.6) 17 (21.5) 2 (2.5)

26 (49.1) 21 (39.6) 19 (35.8) 9 (17.0) 2 (3.8)

9 (34.6) 10 (38.5) 6 (23.1) 8 (30.8) 0 (0.0)

0.225 0.921 0.251 0.161 0.316

6 (22.2) 14 (51.9) 6 (22.2) 2 (7.4) 1 (3.7)

5 (35.7) 8 (57.1) 3 (21.4) 1 (7.1) 1 (7.1)

1 (7.7) 6 (46.2) 3 (23.1) 1 (7.7) 0 (0.0)

0.165 0.568 1.000 1.000 -

Values are mean (SD) or N (%). Statistical analyses were performed with Chi-square. Fisher’s exact or Student’s t-tests. * APOE genotyping was performed in 916 (68.3%) of the 1,342 subjects for classifier training: 74 (93.7%) of the 79 patients with aMCI; and 6 (22.2%) of the 27 patients with AD. respectively. Abbreviations: N ¼ number; SD ¼ standard deviation; CN ¼ cognitively normal; AD ¼ Alzheimer’s disease; aMCI ¼ amnestic mild cognitive impairment; APOE ¼ apolipoprotein E; MMSE ¼ mini-mental state examination; DM ¼ diabetes mellitus; IHD ¼ ischemic heart disease. Table 2 Mixed effects model of the relationship between the classified groups by AD risk score and the rate of decline in neuropsychological results over time in patients with aMCI and AD aMCI patients Converters vs. Non-converters

AD patients Fast decliners vs. Slow decliners

Time by group

Estimate

SE

p-value

Estimate

SE

p-value

Attention Language Visuospatial Memory Frontal/executive SNSB-D total MMSE CDR CDR-SB

-1.2 -2.4 -2.7 1.9 -5.7 -18.4 -3.2 0.2 1.7

0.6 0.8 2.2 5.5 3.1 5.9 0.7 0.1 0.3

0.033 0.004 0.211 0.726 0.066 0.003 < 0.001 0.003 < 0.001

-2.7 -7.5 -8.8 -9.1 -15.4 -43.6 -7.0 1.2 5.9

0.9 1.8 4.1 4.7 3.7 10.1 1.7 0.2 0.7

0.003 < 0.001 0.036 0.058 < 0.001 < 0.001 < 0.001 < 0.001 < 0.001

Linear mixed effects model were performed using age, education, group, time, and the interaction term between group and time (group by time) as fixed effect and patient effect as random effect. Abbreviations: aMCI ¼ amnestic mild cognitive impairment; AD ¼ Alzheimer’s disease; SE ¼ standard error; SNSB-D ¼ Seoul Neuropsychological Screening Battery-Dementia version; MMSE ¼ mini-mental state examination; CDR ¼ Clinical Dementia Ratina; CDR-SB ¼ Clinical Dementia Rat in 2 sum of boxes.

the first year (p ¼ 0.028), and third year of follow-up (p ¼ 0.027). Linear mixed models showed that the results of neuropsychological tests were significantly worsened in converters of aMCI patients and fast decliners of AD patients during follow-up periods. Conclusions: The AD risk score is a novel approach for the prediction of dementia risk, as well as the trajectories of AD on an individual subject level. It will facilitate risk stratification for not only prevention trials but personalized therapy. P1-424

DETECTION OF SMALL INFARCTS IN THE CAUDATE NUCLEUS ON 7 TESLA MRI: THE SMART-MR STUDY

Rashid Ghaznawi1, Mirjam I. Geerlings2, Jeroen Hendrikse2, Jeroen de Bresser1, Theo Witkamp1, Maarten Zwartbol2, Yolanda van der Graaf2, 1UMC Utrecht, Utrecht, Netherlands; 2 University Medical Center Utrecht, Utrecht, Netherlands. Contact e-mail: [email protected]

Background: Small infarcts are among the key imaging features of cerebral small vessel disease (CSVD), but remain largely undetected on conventional 1.5t MRI scans. The use of 7t MRI scans enables detection of these small infarcts. We aimed to establish imaging criteria for the detection of small infarcts in the caudate nucleus on 7t MRI; to determine intra-and inter-rater agreement; and to estimate the frequency in patients with symptomatic atherosclerotic disease. Methods: Cross-sectional analysis withing a prospective cohort study among patients with a history of arterial disease (SMART-MR study). Results: Based on our assessment in 90 patients (6868 years), we defined imaging criteria for cavitated and non-cavitated small infarcts in the caudate nucleus. Intra- and inter-rater agreement, measured in a separate set of patients with atherosclerotic disease (n¼23), was very good/excellent for presence (Cohen’s kappa: 1.00/0.86), number (intraclass correlation coefficient: 0.99/0.98) and individual locations (Dice similarity coefficient: 0.96/0.88) of small infarcts. In the 90 patients, more

P442

Poster Presentations: Sunday, July 16, 2017

as well as a sensitivity of 82.6% and a specificity of 76.7% for distinguishing AD patients from controls. For aMCI and AD, the area under the curve for the ROC was separately 0.825 and 0.864. Conclusions: The results provide insight into the features of dynamic FC alterations and abnormal brain state properties in AD progression

Figure 1. Two cavitated small infarcts in the body of the left caudate nucleus in a 67-year old female shown on sagittal FLAIR (A). T1-weighted (B) and T2-weighted images (C). These lesions are hypointense with a hypointense rim on the FLAIR image, hypointense on the T1-weighted image and hyperintense on the T2-weighted image.

infarcts were detected (12 patients (13%); 20 cavitated, 1 non-cavitated; mean size 5.2mm), compared to 48 primary care patients not selected on disease status (1 patient (2%); 1 cavitated; p¼0.031; Figure 1). Conclusions: We established reliable imaging criteria for the detection of small infarcts in the caudate nucleus on 7t MRI that can be used in future studies to provide new insights into the pathophysiology of CSVD. P1-425

ABNORMAL BRAIN CONNECTIVITY DYNAMICS AND BRAIN ACTIVITY STATES IN ALZHEIMER’S DISEASE

Yu Sun1, Zhaojun Zhu2, Xuanyu Li1, Jiachen Li1, Xiaoni Wang1, Guanqun Chen1, Liu Yang1, Haijing Niu2, Ying Han3,4,5, 1XuanWu Hospital of Capital Medical University, Beijing, China; 2Beijing Normal University, Beijing, China; 3XuanWu Hospital, Capital Medical University, Beijing, China; 4Beijing Institute for Brain Disorders, Beijing, China; 5National Clinical Research Center for Geriatric Disorders, Beijing, China. Contact e-mail: wuyou91@ 126.com Background: Recent advances in analysis approaches of functional connectivity (FC) in brain at rest have provided a new avenue of research towards the nature of intrinsic brain activity, which exhibits spontaneous spatiotemporal dynamics with disparate intermittent fluctuations in the connectivity patterns. Here, we used resting-state functional near-infrared spectroscopy (fNIRS) to perform FC dynamics and FC states pattern in amnesic mild cognitive impairment (aMCI) and Alzheimer’s disease (AD). Methods: We employed fNIRS and sliding-window correlation analysis as well as k-means clustering of windowed correlation matrix to systematically investigate the whole-brain FC dynamics, brain FC states and the association between brain FC fluctuation and clinical performance in controls (n¼31), aMCI (n¼27) and AD (n¼24). Functional connectivity fluctuation strength (FCFS) was used to characterize spontaneous FC fluctuation, and a network-based statistic (NBS) approach was introduced to compare group difference in FCFS. Analysis of multiple linear regressions was conducted to test the association between clinical variables and FCFS or FC state metric. Results: We revealed increased FCFS in aMCI and AD compared to controls (p < 0.005). The significant variable connections involved in long-distance connections which mainly belong to the default mode network, frontal-parietal network and internetwork connections. Among mainly FC states, we reported significantly reduced emergency frequency for state 1 in contrast to increased frequency for state 2 in AD relative to controls (all p < 0.05). Significant opposite correlations were found between metrics of state 1 and state 2 with cognitive measures. The mean FCFS in the NBS-based connections (p < 0.005,) exhibited a sensitivity of 84% and a specificity of 70% for distinguishing aMCI patients,