Multiple Sclerosis Journal 23(11) Table 3. Correlation of NK (subset) cells and proportions and clinical outcome parameters. Dependent variable CD56% CD56 cells CD56 bright % CD56 bright cells CD56 dim % CD56 dim cells Treatment response variable NEDA MRI activity EDSS progression Relapses p value p value p value p value 0.08 0.476 0.94 0.288 0.94 0.532 0.025 0.01 0.04 0.707 0.04 0.008 0.546 0.828 0.334 0.483 0.334 0.881 0.834 0.408 0.226 0.371 0.226 0.331 NK cells: natural killer cells; NEDA: No Evidence of Disease Activity; MRI: magnetic resonance imaging; EDSS: Expanded Disability Status Scale; CD56%: proportion of NK cells within the lymphocyte count; CD56 cells: total number of NK cells; CD56 bright%: proportion of bright NK cells within the NK cell count; CD56 bright cells: total number of bright NK cells; CD56 dim%: proportion of dim NK cells within the NK cell count; CD56 dim cells: total number of dim NK cells. P-values in bold indicate significant correlations. Table 4. Correlation of NK (subset) cells and NK subset proportions with MRI activity. Dependent variable CD56% CD56 cells CD56 bright% CD56 bright cells CD56 dim% CD56 dim cells MRI activity No activity Activity No activity Activity No activity Activity No activity Activity No activity Activity No activity Activity Mean 11.5 15.7 0.156 0.212 10.5 8.1 0.014 0.015 89.5 91.9 0.141 0.196 Std. error 1.0 1.5 0.0 0.0 0.6 0.9 0.0 0.0 0.6 0.9 0.0 0.0 95% Confidence interval p value Lower bound Upper bound 9.4 12.6 0.1 0.2 9.2 6.2 0.0 0.0 88.2 90.0 0.1 0.2 13.5 18.7 0.2 0.2 11.8 10.0 0.0 0.0 90.8 93.8 0.2 0.2 0.025 0.01 0.04 0.71 0.04 0.008 NK cells: natural killer cells; MRI: magnetic resonance imaging; CD56%: proportion of NK cells within the lymphocyte count; CD56 cells: total number of NK cells; CD56 bright%: proportion of bright NK cells within the NK cell count; CD56 bright cells: total number of bright NK cells; CD56 dim%: proportion of dim NK cells within the NK cell count; CD56 dim cells: total number of dim NK cells. P-values in bold indicate significant correlations. Given the difficulty in ascertaining treatment-naive patients, we chose to use patients that were off any form of treatment for at least 28 days. However, we do acknowledge that some residual effects on NK levels arising from previous treatment may exist in this control cohort. To assess this possibility, we ascertained a small group of n = 7 patients who were completely naive to any MS treatment and measured the NK subsets in these patients. The results of this analysis showed that the mean values for all NK subsets in the treatment naive sample were not significantly different to the 'off treatment' controls used in the analysis proper (p > 0.1). This finding justifies using our off 1484 treatment control group and, in turn, supports our conclusions about treatment effects. With the large number of disease treatments available to MS patients, the current dilemma faced by neurologists is not only to make the first treatment choice but also when to change treatments. There is a trend to move away from basing treatment decisions on clinical relapses alone. Newer paradigms such as NEDA criteria including MRI parameters are rapidly taken up, but there is still a desperate need for biomarkers predicting treatment response. We explored the possible application of NK cell expression as a tool for journals.sagepub.com/home/msjhttps://journals.sagepub.com/home/msj