Multiple Sclerosis Journal - October 2017 - 1485

P Caruana, K Lemmert
evaluating therapeutic efficacy by associating NK cell
profiles with clinical outcome measures such as MRI
(appearance of new or enhancing lesions) and NEDA
criteria in our MS cohort.
The fact that MRI measurements provide greater sensitivity as clinical outcome measures has already been
postulated in 1992.27 The absence of new/enlarging
T2 lesions was the only predictor of cortical, subcortical grey matter and thalamic atrophy rates,28 an irreversible outcome measure. More importantly, two
meta-analyses of different randomized MS trials indicated that the effect of a treatment on relapse rate can
be accurately predicted by the effect of that therapy
on MRI lesions.29,30 This could, in fact, explain why
in our relatively small sample size we only found a
correlation with MRI activity, but not clinical
outcome.
We applied an observation period for up to 2.4 years.
During this period, our total group had a NEDA of
46%. We were unable to demonstrate a correlation of
NK parameters with NEDA, Multiple Sclerosis
Severity Scale (MSSS),18 EDSS or EDSS progression, but could demonstrate a strong association with
MRI parameters (table 3). A longer follow-up might
detect an association of CD56 subsets with clinical
disability.
Decreased total NK numbers, but low proportion of
CD56 dim and, in turn, high proportion of CD56
bright were associated with a lack of MRI activity in
our study. A positive correlation of CD56 bright with
clinical response to beta-IFN has been reported, but
the study included limited MRI data (n = 15 with MRI
after 12 months) necessary to judge treatment response
according to Rio criteria.16
Lower proportion of CD56 dim means in general a
higher proportion of CD56 bright. Our results parallel
NK changes observed in pregnancy, with the most
pronounced changes occurring in the last trimester.
These changes correlate significantly with a lack of
clinical activity.10,11 It has also been demonstrated that
the extent of CD56 bright extension in the periphery
in response to daclizumab therapy predicted fewer
gadolinium enhancing lesions.31 Han et al.32 analysed
the cerebrospinal fluid (CSF) of 211 patients and
found that while there was a larger proportion of
CD56 bright in the CSF, there was no correlation
between the proportion of CD56 bright or dim in the
CSF and peripheral blood. This is contrasted by the
findings of Bielekova et al.,33 who clearly demonstrated that the increase in CD56 bright in the

journals.sagepub.com/home/msj

periphery was paralleled by an even larger increase in
CD56 bright in the CSF under treatment with daclizumab, which was associated with clinical benefit. In
addition, immune regulatory effects of CD56 bright
might well occur in the periphery and reduce the
available number of activated lymphocytes to be
recruited to the CSF.
It is of note that our study showed more clinical
events, namely, 19 patients with EDSS progression
and 47 patients with relapse when compared to MRI
events, which consisted of 32 patients with MRI progression. As not all relapses were associated with new
or gadolinium enhancing lesions, one can question
whether some of these relapses might have been
pseudo-relapses. On the other hand, new or gadolinium-enhancing lesions might have been missed
because of the timing or poor quality of the scan, as
some of the scans had only 5-mm-thick slices, some
(two patients) missed a cervical scan and most (94%)
did not undergo imaging of the full spine.
For routine screening of lymphocyte subsets, division
of CD56 into their respective dim and bright subsets
is not usually performed. We retrospectively gated the
populations of interest on flow cytometry. It would
have been of benefit to apply a number of additional
markers of NK cell activation, but this was impossible
in a retrospectively analysed sample of routine
testing.
Our study has several limitations, predominantly as
a result of the retrospective nature of the study. We
have included clinic patients with a wide range of
EDSS (0-7.5) and disease duration (0.6-38 years).
Lymphocyte subsets have been taken as per clinical
indication. As the length of treatment might influence
lymphocyte subsets, samples should be taken at set
time intervals in future prospective studies. In order to
take this into account, we have adjusted for the time
on treatment in our analysis. Being a retrospective
study, we had to review routine MRI scans of variable
time intervals and imaging quality; however, only
4.5% of patients had no imaging performed in the
observation period. Our patients had to be off treatment or on immune modulatory treatment for their
blood sample to be considered suitable for our study.
The minimum time interval on or off treatment of
28 days might have been too short to adequately
reflect treatment effect. However, the median time on
treatment was 887 days or 2.4 years.
Future studies should include blood collections at set
time intervals of a cohesive patient group including

1485

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

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