Advanced MRI Technology May Help Predict Dementia After Stroke

By Marilynn Larkin

September 27, 2019

NEW YORK (Reuters Health) - An advanced MRI technique called diffusion tensor image segmentation (DSEG) detects microstructural brain damage in patients with small vessel disease (SVD) and can identify those at increased risk of dementia, researchers say.

"Following a group of patients with SVD - a disease often associated with strokes - over five years, we predicted who would develop difficulties with planning, organizing information and processing speed," Dr. Rebecca Charlton of Goldsmiths University of London, UK, told Reuters Health. "We also predicted who would go on to develop vascular dementia."

"Importantly, this technique relies on just one (DSEG) scan, which is readily available on most MRI systems," she said by email. "Previous research predicting cognitive decline and dementia generally use several scans and/or several different clinical tests. By using just one scan, the patient can spend less time in the MRI scanner, and the results are equally accurate."

Dr. Charlton and colleagues studied data from 99 patients (average age, 68; about one-third women; most, white) enrolled in the St George's Cognition and Neuroimaging in Stroke study in London from 2007 to 2015. Participants underwent annual MRI scanning for three years and cognitive assessment for five years. DSEG at the angle theta (DSEG-theta) was used as a whole-cerebrum measure of SVD severity.

Cardiovascular risk factors were also recorded, including hypertension (systolic blood pressure >140 mm Hg or diastolic >90 mm Hg or treatment with antihypertensive drugs); hypercholesterolemia (serum total cholesterol >5.2 mmol/L or treatment with a statin); type 2 diabetes; and smoking status.

As reported online September 12 in Stroke, DSEG-theta was significantly related to decline in executive function and global cognition. Eighteen (18.2%) patients converted to dementia, with a mean time to conversion of 3.31 years.

No significant differences between those who did and did not convert to dementia were seen in demographic or vascular risk factors, including age, sex, and premorbid intelligence.

However, there were significant differences in baseline DSEG-theta, executive function, information processing speed, working memory, episodic memory and global cognition, as well as the Mini-Mental State Exam, with patients who developed dementia showing a higher level of overall brain damage and poorer cognitive functions.

Baseline DSEG-theta predicted dementia with a balanced classification rate=75.95% and area under the receiver operating characteristic curve=0.839. The best classification model included baseline DSEG-theta; change in DSEG-theta; age; sex; and premorbid intelligence quotient (balanced classification rate of 79.65%; area under the receiver operating characteristic curve=0.903).

Dr. Charlton said, "The DSEG technique requires automatic computer analysis after the scan. This analysis compares each individual to a reference image. In this study, we compared patients to a healthy middle-aged adult brain."

"Over time, we will generate and develop new comparison images, which will allow for patient-specific comparisons - e.g., age, sex, matched, etc.," she added. "Future studies will examine those at risk for different types of dementia. We will also investigate whether the technique can distinguish development of different types of dementia."

Dr. Malaz Boustani, a geriatrician and implementation neuroscientist at the Sandra Eskenazi Center for Brain Care Innovation and the Regenstrief Institute in Indianapolis, Indiana commented in an email to Reuters Health, "I have concerns about the external validity of the study. The study population is not representative and included a very small and high-risk population to start with - those who were relatively young hospitalized patients with clinically significant SVD-based stroke."

Further, he added, the study does not include a comparison group with a similar imaging-based SVD burden, but no clinically diagnosed stroke.

"The value of repeating MRI to run the DSEG is very small and very expensive," he said. "One can reach a similar predictive value with much more scalable methods such as a simple global cognitive test."


Stroke 2019.