Previous studies have reported diffuse hypoperfusion in normal appearing white matter (NAWM) in patients with relapsing-remitting multiple sclerosis (MS), using dynamic susceptibility contrast (DSC) MRI^1–3 and arterial spin labeling (ASL) MRI⁴. The extent of this hypoperfusion is unclear and ranges from few percent up to a factor of almost two. The present study aims to investigate this issue with yet another technique, using dynamic contrast-enhanced (DCE) MRI.

The statistical power of a DCE-MRI acquisition to reveal hypoperfusion in MS was estimated using a Monte-Carlo simulation. An arterial input function (AIF) from a pilot study⁵ was used to generate synthetic tissue curves with a contrast-to-noise ratio of 7.5, for MS patients and control group; using previously reported perfusion values from a DSC-MRI study¹ and an ASL study⁴. A compartment-uptake model⁶ was fitted to these curves, yielding estimates of cerebral blood flow (CBF), cerebral blood volume (CBV) and permeability-surface area product (PS) as a measure of vascular permeability. This was repeated n=1000 times. Mean and standard deviation of the resulting distributions were used to calculate the statistical power of a DCE MRI study to detect perfusion differences with n=16 subjects in MS and control group.

In an IRB-approved study, patients with relapsing-remitting MS (n=24, mean age 36 years, 17 female, mean EDSS score 3.25) and patients without history or symptoms of neurological disorder (n=16, mean age 49 years, 9 female) underwent a DCE-MRI examination with a previously established MRI protocol⁵ (3D SPGR sequence, 2.1s temporal resolution, 44 slices, spatial resolution 1.7*1.7*3mm³). Regions were defined manually in the middle cerebral artery, in frontal, periventricular and occipital NAWM, in the pons, the splenium and in the thalamus (Fig. 1), and CBF, CBV and PS were quantified as described previously⁵.

Parameter differences between MS and control groups were evaluated using a mixed linear model with subjects as random effect and controlling for age and gender. CBF, CBV and PS were modeled independently. A p-value of less than 0.05 was considered to indicate statistical significance.

CBF distributions before and after DCE MRI analysis are shown in Fig. 2. With group sizes of n=16, DCE MRI can detect NAWM hypoperfusion as reported in ¹ with a statistical power of 96%. The statistical power reduces to 18% for differences as reported in ⁴. Fig. 3 displays representative tissue curves and model fits. Boxplots visualizing quantitative results in all regions are shown in Fig. 4.

Mean CBF in NAWM was 11.0 (15.1) and 10.4 (8.2) ml/100ml/min in MS and control group, respectively. Mean CBV in NAWM was 0.50 (0.45) ml/100ml in MS and 0.48 (0.28) ml/100ml in control group. No significant differences were observed for CBF, CBV and PS between patient groups (p=0.44, p=0.20, p=0.66). PS was not significantly different from zero in all regions, but slightly higher in periventricular matter than in other regions (Fig. 4). Age and gender had no significant influence on any of the three parameters.

We found no significant differences between MS and control patients in any of the investigated perfusion parameters. For effect sizes as reported in ¹, our simulation study indicates a high statistical power to detect such differences. For smaller effect sizes, as reported in ⁴, this power is markedly reduced, indicating that small differences may not be reliably detected with a DCE MRI study. Taken together, these findings suggest that, if a perfusion difference exists in our patient cohort, it must have a substantially smaller effect size than reported in previous DSC studies. It is also possible that effect sizes were overestimated in these studies, a problem which occurs in particular when sample sizes are small⁷.

Absolute values of CBF and CBV in our study were lower than literature values. This is a well-known effect in cerebral DCE-MRI and may be attributed to effects of limited water exchange^8,9 across the vessel wall. We do, however, not expect that this effect played a role in hiding potential perfusion differences, since PS was not significantly different between groups.

A potential limitation of our study is the age mismatch between MS and control groups. However, statistical analysis yielded no significant influence of age on any of the three parameters.

Despite high statistical power, we could not confirm previous reports of NAWM hypoperfusion in MS. This indicates that, at least in in our patient cohort, potential hypoperfusion is much less pronounced than reported in previous studies.