Multiple Sclerosis Journal - October 2017 - 1457

PM Matthews
functional recovery after axonal loss with inflammatory demyelination.7 Applications of DTI have complemented other experimental measures of
demyelination and axonal loss, particularly as they
provide measures related to the progressive loss of
brain connectivity with disease progression.8 They
have been powerful tools for generating hypotheses
regarding clinico-pathological correlations.
Nonetheless, they have substantial limitations for
testing hypotheses, particularly in the context of MS.
The phenomena that they are sensitive to are only
related indirectly to the concepts of real interest. This
can lead to fundamental confounds to interpretation.
For example, fMRI depends on the comparability of
neurovascular coupling mechanisms in health and
disease, in different stages of disease or with treatment. Astrocyte activation state and the microvasculature can be altered in MS.9 Differences therefore
may reflect the influences of MS pathology (or possible drug effects) on the neurovascular coupling
mechanism rather than on neuronal activation.
For DTI, the issues are different, but the general conclusion is similar. Interpretation of DTI measures in
cerebral white matter depends on an understanding
of the tissue microstructure that we cannot have in
the context of the evolving neuropathology of MS.
The maximum practical linear resolution of a singleimage voxel (>1 mm3) is so much larger than the
cross-sectional area of individual axons (<5 µm3) that
diffusion measures are always averaged across many
fibres. In contrast to the spinal cord, where axons are
highly linearly oriented, cerebral white matter has a
complex microanatomy of crossing or angulated
fibres through most of its volume. Diffusion along
orthogonal crossing fibres averaged within the same
voxel may not display significant anisotropy, despite
the highly oriented diffusion in each. Loss or dystrophy of axons in either component of such orthogonal
crossing fibres will give rise to a 'paradoxical'
increase in diffusion fractional anisotropy.10 At the
whole voxel level, this would be similar to what might
be anticipated with remyelination!
The measures also have fundamentally limited measurement properties. Both fMRI and DTI intrinsically
generate much lower signals than does conventional
MRI, as only small fractions of the total water in the
tissue are being used to generate the image signal.
Unless the effect sizes for the phenonmenon of
interest scale similarly, the measures may not improve
the efficiency of a Phase II trial. As they are most
commonly implemented, they are also expressed as
relative, rather than absolute measures. Hence, it is
journals.sagepub.com/home/msj

difficult to interpret changes in the measures in terms
of absolute changes in the brain tissue. This makes
comparisons of results across disease stages (or
different forms of pathology) problematic, as the
contexts for normalisation change.
None of this is to suggest that fMRI and DTI are not
powerful tools for characterisation of in vivo neuropathology or for exploratory pharmacology (e.g. following demonstration of drug-associated changes in
the same measures in an animal model). However, as
relative measures of MRI biophysical phenomena
are only indirectly related to the functional neuropathology of interest, with sensitivities to relevant drug
responses that are uncharacterised or based on only
relatively small clinical studies, these measures are
far from 'clinical trial ready'. Contextualising
BOLD fMRI signals using simultaneous measures
of baseline blood flow and neurovascular reactivity
and advanced data acquisition and analytics for DTI
could address some of the concerns that I have
raised.9 However, with the exception of classical
evoked potential measures, these tools still are at
very early stages of development.
Declaration of Conflicting Interests
The author(s) declared the following potential conflicts of interest with respect to the research, authorship and/or publication of this article: P.M.M. is in
receipt of generous personal and research support
from the Edmond J Safra Foundation and Lily Safra,
the Medical Research Council, the UK Dementia
Research Institute and the UK Engineering and
Physics Science Research Council. P.M.M. acknowledges consultancy fees (paid to his institution) from
Roche, Adelphi Communications and OrbiMed. He
has received honoraria or speakers' fees (also paid to
his institution) from Novartis and Biogen, and has
received research or educational funds from Biogen,
Novartis and GlaxoSmithKline.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship and/or publication of this article: This study was supported by the
Imperial College Healthcare Trust Biomedical
Research Centre.

References
1. De Groot CJ, Bergers E, Kamphorst W, et al. Postmortem MRI-guided sampling of multiple sclerosis
brain lesions: Increased yield of active demyelinating
and (p)reactive lesions. Brain 2001; 124(Pt. 8):
1635-1645.

1457

https://journals.sagepub.com/home/msj

Table of Contents for the Digital Edition of Multiple Sclerosis Journal - October 2017