Diffusion-weighted imaging (DWI)–based measures of brain connectivity do not add value to prediction of long term cognitive recovery after stroke over conventional predictors, a study shows.

While advanced MRI sequences performed in the early poststroke period give a good estimation of white matter integrity they are no better in predicting cognitive recovery than currently used clinical, neuropsychological, and conventional imaging variables, according to neurologists in the Netherlands.

In a prospective study, 217 patients with ischaemic stroke underwent DWI magnetic resonance imaging at five weeks after stroke to evaluate 4 measures of brain network connectivity: mean diffusivity, fractional anisotropy, weighted global network efficiency, and mean connectivity strength.

Patients also underwent neuropsychological assessment at five weeks and one  year for what the researchers termed poststroke cognitive disorder (PSCD).

Of the 135 patients with marked or modest PSCD at five weeks, 30% showed improvement to modest or no PSCD after one year.

The study investigators found that three out of four brain connectivity measures were significant predictors of cognitive recovery in univariable regression analysis, but there was no added value of these measures when added to a multivariable model that included conventional predictors such as level of education and MRI imaging findings of infarct size.

The researchers said the field of brain connectivity measures was evolving rapidly, and new methods such as fixel-based analysis and Single-Shell 3-Tissue Constrained Spherical Deconvolution could better differentiate between different tissue compartments, which may also provide additional information on white matter integrity.

“It may well be that these measures will evolve into useful predictors of cognitive outcome after stroke in the near future,” they concluded.

In an accompanying commentary, Professor Amy Brodtmann of the Florey Institute of Neuroscience and Mental Health, Melbourne, said the lack of added value of the diffusion  measures might have been because they were relatively crude measures of white matter integrity.

“Although they may be sensitive to gross white matter changes, they lack specificity in distinguishing from intracellular and extracellular water compartments or modelling crossing fibres, which are the rule rather than the exception in the brain’s white matter pathways,” she wrote.

Advances in DWI acquisitions  meant that clinical scanners were now capable of producing images of high enough resolution to detect changes in white matter integrity not ordinarily visible, she said.

However, for imaging markers to have the greatest usability in predictive models of stroke recovery they would need to be extractable from commonly collected clinical sequences, she added.