Article Text

Download PDFPDF
Towards multicentre diffusion MRI studies in cerebral small vessel disease
  1. Alberto de Luca1,2,
  2. Geert Jan Biessels1
  1. 1 Department of Neurology, UMC Utrecht Brain Center, University Medical Center Utrecht, Utrecht, The Netherlands
  2. 2 Department of Radiology, University Medical Center Utrecht, Utrecht, The Netherlands
  1. Correspondence to Dr Geert Jan Biessels, Department of Neurology, UMC Utrecht, Utrecht 3508GA, The Netherlands; g.j.biessels{at}

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Diffusion MRI metrics help to explain variance in cognitive performance in patients with cerebral small vessel disease

Diffusion MRI can provide unique insights into the structure of the human brain. By exploiting motion of water molecules as a contrast mechanism, diffusion MRI informs about tissue properties at a scale well beyond conventional structural MRI. Diffusion MRI metrics derived with diffusion tensor imaging and higher order models are emerging as powerful biomarkers of cerebral white matter injury in the context of small vessel disease. For example, the diffusion tensor imaging-derived metric peak width of skeletonised mean diffusivity (PSMD) was shown to better explain interindividual variation in cognitive function than conventional small vessel disease MRI lesion markers.1 While many diffusion MRI metrics have shown promise in dedicated monocentre studies, their applicability in multicentre settings or clinical practice is hindered by limited evidence of repeatability and reproducibility, scanner dependence and lack of normative data.2 Egle et al 3 present a study on the added value of diffusion tensor imaging metrics over conventional lesion makers to predict cognitive function across six different cohorts with varying small vessel disease burden. Despite substantial heterogeneity in study populations, MRI acquisition parameters and assessment of cognitive outcomes, in all six cohorts the diffusion tensor imaging-derived metric mean diffusivity (MD) median consistently explained as much variance in cognitive performance than all other lesion markers combined. Furthermore, baseline MD median predicted conversion to dementia in three cohorts with longitudinal data.

Egle’s study thus provides further evidence on the value of diffusion MRI as indicator of white matter injury underlying cognitive dysfunction in small vessel disease. Yet, it also highlights issues that warrant further evaluation. For example, in the included cohort of patients with cerebral autosomal-dominant arteriopathy with subcortical infarcts and leucoencephalopathy (CADASIL), MD median explained remarkably less variance in cognition (~16%)3 than PSMD in a very similar population in a previous study (46%).1 By contrast, that previous study reported that in patients with sporadic small vessel disease (the ‘RUN DMC’ cohort) PSMD explained 8.8% variance in cognition, slightly less than the variance explained by MD median (~11%) in that same cohort in Egle’s study. Lesion burden, which was much higher in the patients with CADASIL than in the RUN DMC cohort, may explain these opposing findings for different diffusion MRI metrics. Hence, better insight in the value of different diffusion MRI metrics in small vessel disease is essential, also considering performance according to lesion burden and including further instrumental validation and head-to-head comparison of multiple metrics.2

Although Egle’s study involved six cohorts, diffusion MRI acquisition and analysis were essentially conducted as six independent monocentre studies. To move towards true multicentre studies with pooled analyses of aggregated data, additional harmonisation steps are needed. Clearly, standardisation of MRI acquisition protocols is essential for prospective studies. Of note, emerging post processing techniques may also allow harmonisation of already acquired diffusion MRI data4 to remove cross-site differences. Such techniques might also help to amend the disruptive effect of scanner updates in long-term monocentre longitudinal studies. Finally, standardisation of processing is essential, as subtle changes in processing protocols may introduce variability. This can be resolved with centralised processing or by deploying ‘containerised’ analysis tools which are indifferent to local configurations and user settings.

Ethics statements

Patient consent for publication



  • Contributors GJB and AdL have authored this commentary.

  • Funding The research of AdL and GJB is supported by Vici Grant 918.16.616 from ZonMw, The Netherlands, Organisation for Health Research and Development (PI, GJ Biessels).

  • Competing interests None declared.

  • Provenance and peer review Commissioned; internally peer reviewed.

Linked Articles

  • Cerebrovascular disease
    Marco Egle Saima Hilal A M Tuladhar Lukas Pirpamer Edith Hofer Marco Duering James Wason Robin G Morris Martin Dichgans Reinhold Schmidt Daniel Tozer Christopher Chen Frank-Erik de Leeuw Hugh S Markus