Immunological and short-term brain volume changes in relapsing forms of multiple sclerosis treated with interferon beta-1a subcutaneously three times weekly: an open-label two-arm trial

The trial was conducted in accordance with the International Conference on Harmonisation guidelines for Good Clinical Practice and applicable local regulations, as well as the Declaration of Helsinki. Written informed consent was given by patients before participation and the protocol was approved by the Institutional Review Board at the University at Buffalo Health Sciences.

Percent of whole brain volume change was measured from baseline (0 months) to 3 months, from 3 to 6 months, and from baseline to 6 months using SIENA analysis algorithms [30]. Serial scans were co-registered and relative edge motions (measuring expansion or contraction of the brain) were detected between scans and used to extrapolate the percent change in brain volume. SIENA measurements were improved by application of an in-house developed inpainting technique, nonuniformity correction [33], and intensity standardization [34]. Changes in GM and WM volumes were measured during the same study periods described above using the SX-MTP analysis algorithms on the inpainted, corrected, and standardized images. For GM/WM segmentation, 4D accuracy was improved through MTP modification via longitudinal regularization of a hidden Markov random field model.

For the immunological marker analysis, CD4⁺ and CD8⁺ T-cell expression of an array of immunological markers with known immunomodulatory properties was quantified at baseline and at 6 months. Blood samples were collected from patients. Peripheral blood mononuclear cells (PBMCs) were placed in serum-free medium and stimulated with phorbol myristate acetate (50 ng/mL) and Ionomycin (500 ng/mL; Sigma, St. Louis, MO) for 2 h and brefeldin A (1:1000 dilution; eBioscience, San Diego, CA) for an additional 3 h for the intracellular cytokine staining. PBMCs were fixed, permeabilized, and stained with fluorescein-conjugated antibodies against IFN γ, IL-4, IL-17A, IL-21, IL-22 (eBioscience, San Diego, CA), IL-17 F, brain-derived neurotrophic factor (BDNF; R&D Systems, Minneapolis, MN), IL-10, CD4, and CD8 (BD Biosciences, San Jose, CA). For nerve growth factor (NGF) staining, the cells were stained with primary antibody against NGF (Epitomics, Cambridge, MA), followed by staining with fluorescein isothiocyanate-conjugated secondary antibody (BD Bioscience). The percent of CD4⁺ and CD8⁺ T cells expressing each marker of interest was determined using a BD FACSCalibur™ flow cytometer and CellQuest software (BD Biosciences) and gated by size and granularity.

All statistical comparisons were planned in advance of data collection. The Wilcoxon signed-rank test was used to test for within-group differences in percent of brain volume change (whole brain, GM, and WM) at each of the study periods (from baseline to 3 months, from 3 months to 6 months, and from baseline to 6 months). The Wilcoxon rank-sum test was used to test for differences in percent of whole brain volume change, GM, and WM between patients and HCs during each study period. Correction for multiple comparisons by the Holm-Bonferroni method was applied. Spearman’s rank correlation was used to test the correlations between immunological parameters, Gd-enhancing lesions at baseline, and percent change in brain volume and GM and WM over 3 and 6 months of treatment. No corrections for multiple comparisons were made to the correlation results.

Whole brain volume analysis is shown in Table 1 and Fig. 1. In patients, the mean (SD) change from baseline to 3 months in brain volume was –0.95 % (1.71; P = 0.030 for difference from baseline; P = 0.090 after adjusting for multiple comparisons), –0.08 % (1.36; P = 0.960) from 3 to 6 months, and –0.91 % (1.88; P = 0.051) from baseline to 6 months. For comparison, in HCs the mean change from baseline to 3 months in brain volume was 0.24 % (1.07; P = 0.359), –0.32 % (1.50; P = 0.626) from 3 to 6 months, and 0.01 % (1.89; P = 0.952) from baseline to 6 months.

GM volume analysis is shown in Table 1 and Fig. 2. The pattern for GM was similar to that of whole brain. In patients, the mean (SD) change from baseline to 3 months in GM volume was –1.52 % (2.41; P = 0.004 and after adjusting for multiple comparisons, P = 0.029), –0.46 % (3.11; P = 0.176) from 3 to 6 months, and –1.66 % (1.84; P < 0.001 and after adjusting for multiple comparisons, P = 0.002) from baseline to 6 months. There were no significant differences observed for the HCs: mean change from baseline to 3 months was 0.01 % (SD 2.42; P = 0.934), –0.60 % (2.19; P = 0.426) from 3 to 6 months, and –0.51 % (2.47; P = 0.268) from baseline to 6 months.

WM volume analysis is shown in Table 1 and Fig. 3. Changes in WM tissue volume of patients were not significant at any timepoint. In patients, the mean (SD) change from baseline to 3 months in WM volume was –0.41 % (2.16; P = 0.390), 0.30 % (2.99; P = 0.626) from 3 to 6 months, and –0.21 % (3.20; P = 0.933) from baseline to 6 months. In HCs, the mean (SD) change from baseline to 3 months in WM volume was 0.51 % (1.94; P = 0.107), –0.03 % (1.96; P = 0.952) from 3 to 6 months, and 0.58 % (2.46; P = 0.463) from baseline to 6 months.

Only 8 of the patients with RRMS had any Gd-enhancing lesions at baseline, and there were no significant correlations between the number or volume of Gd-enhancing lesions at baseline and the changes in whole brain, GM, or WM volumes.

There were a number of significant correlations between whole brain and brain tissue-specific volumes and immunological markers following 6 months of treatment with IFN β-1a SC. Decreased percentage of IL-17 F-expressing CD4⁺ T cells from baseline to 6 months correlated significantly with decreasing whole brain volume over the same period (r = 0.51; P = 0.022; Fig. 4).

Decreased percentage of IL-17 F-expressing CD8⁺ T cells from baseline to 6 months correlated with decreasing GM volume over the same period (r = 0.47; P = 0.037; Fig. 5). In addition, decreased percentage of IL-17 F-expressing CD4⁺ T cells from baseline to 6 months correlated with a decreasing WM volume (r = 0.46; P = 0.043; Fig. 6), suggestive of pseudoatrophy.

A higher percentage of IL-22–expressing CD4⁺ T cells at baseline correlated with smaller decreases in GM volume from baseline to 6 months (r = 0.59; P = 0.005; Fig. 7).

The aim of this study was to use advanced imaging techniques to measure the extent of global and tissue-specific brain volume changes, and to examine the involvement of IFN β-1a 44 μg SC tiw, in the pseudoatrophy phenomenon in patients with RRMS. A better understanding of the pathophysiological mechanisms in RRMS, including a differentiation of the dynamics of inflammation and demyelination leading to brain volume loss observed in MS, could aid in devising means for halting disease progression and for treatment monitoring and optimization. In particular, the early study period measurement from baseline to 3 months represents an opportunity to investigate acute treatment-associated pseudoatrophy following IFN β-1a 44 μg SC tiw therapy initiation. The investigation of brain volume loss in whole brain tissue, as well as in GM and WM, allowed for a greater insight into the specificity of tissue volume changes suggestive of pseudoatrophy. In whole brain and GM, the effect of IFN β-1a SC treatment was associated with a reduction in brain tissue volume; in both cases the majority of the reduction occurred shortly (within 3 months) following treatment initiation, in comparison with the 3–6-month period, a result that is highly consistent with an acute pseudoatrophy effect.

Treatment-associated pseudoatrophy has been thought previously to be more evident in WM than in GM [9, 35]. This was not observed in the present study, where decline in whole brain tissue volume loss appeared to be driven by GM tissue volume loss. GM loss was more pronounced in the first 3 months than the second 3 months, and only loss in the first 3 months was statistically significant. While pseudoatrophy owing to WM volume has been noted during treatment with natalizumab [9], the longer term brain tissue changes in RRMS and clinical measures (such as physical and cognitive disability) have been associated more with GM volume changes rather than loss of WM volume [36, 37].

This conflicting finding of a GM rather than a WM effect may be due to a number of factors. Given the small sample size, issues of statistical power cannot be ruled out – in fact, the observed but not statistically significant WM volume changes were in the expected direction and were lower from 0–3 months than from 3–6 months. Methodological effects should also be carefully considered. Although the SIENAX-MTP method employed has been shown to more accurately measure GM volume change than was previously possible, it was not compared head-to-head with SIENAX or SPM for this study. It is noteworthy, though, that cortical GM measurement is much more susceptible to partial volume errors owing to its thin and convoluted nature, and that previous studies using less sensitive analysis techniques failed to detect significant brain atrophy progression over a 3-month period in RRMS [38] or in an open-label study of 30 patients with RRMS treated with IFN β-1b [5].

If correct, though, these exploratory results merit speculation about the potential pathophysiological basis for the observed GM changes. Over the past decade, it has become clear that GM plays an enormous role in both the pathology and the clinical picture of multiple sclerosis [39]. What remains unclear is whether the involvement of GM is secondary to WM damage or a semi-independent process [40]. Of particular relevance to the current work is the inflammatory nature of GM pathology. Histopathological studies have found much lower inflammation in the GM than in the WM [41, 42]. However, biopsies have shown substantial inflammation [43, 44]. Notably, cortical inflammation appeared in patients with early RRMS, much like the cohort studies described here. It is also possible that different therapeutic mechanisms of DMDs may affect any pseudoatrophy of specific tissue compartments differently.

Investigating correlations between brain volume and pro- and anti-inflammatory markers has provided some new insight into the proposed mechanisms of action of IFN β-1a SC tiw. The significant correlations between the decreased percentage of pro-inflammatory IL-17 F–expressing CD4⁺ or CD8⁺ T cells and reductions in whole brain, GM, and WM tissue volumes are supportive of an early anti-inflammatory therapeutic effect for IFN β-1a SC tiw.

The correlation between higher baseline percentage of IL-22–expressing CD4⁺ T cells and less GM tissue atrophy from baseline to 6 months in patients is also of potential interest. Although interpretation of cross-sectional markers should be undertaken with care, it is possible that higher baseline IL-22 levels may be indicative of a different pseudoatrophy response – either via less resolvable baseline inflammatory-related volumetric changes or via differences in treatment response.

The comprehensive battery of immunological markers, including BDNF, was used to screen for correlations with brain tissue volume changes. Beyond IL-17 F and IL-22, no further significant correlations were observed, despite previous reports that BDNF secretion was associated with inflammatory activity in MS lesions in the brain and with higher WM tissue volume [45].

We did not detect a relationship between brain volume and the resolution of Gd-enhancing lesions, likely due to the high proportion of patients with no Gd-enhancing lesions at baseline.

Our findings imply the need for caution when planning the timing of using MRI techniques to measure disease progression in terms of brain volume. They suggest that the baseline for MRI measures of brain volume in patients with RRMS with active disease embarking on DMD therapy should start only after several months of therapy, once the pseudoatrophy effect has run its course, rather than at the beginning of therapy, indicating a need for additional MRI measurements. Based on these results, the pseudoatrophy effect may account for a 1 % or more change in volume. Although seemingly small in an absolute sense, the yearly percent brain volume change in MS may be as low as 0.5 % (and 0.1 % in HCs). Thus, early pseudoatrophy over 3 months may produce volume changes commensurate with 2 years of MS disease effect, and completely mask treatment effects in clinical trials of that length. This is highly relevant, as such trials are becoming more common.

It is even conceivable that in the future more accurate MRI and analysis techniques might lead to individual patient care decisions based on rates of brain atrophy in addition to the conventionally used lesion burden. Any such initiatives will need to carefully account for the pseudoatrophy effect in order to be useful.

With the exception of GM tissue volume change from baseline to 3 months, the statistical significance of other associations observed was lost following adjustment for multiple comparisons. This may be as a result of this being a pilot study with a relatively small sample size, as well as the large number of exploratory comparisons that underwent correction. It is also possible that the improved SIENAX-MTP algorithm may be more powerful to detect real GM change than WM change [29]. Therefore, the lack of WM changes detected should be interpreted with caution.

The additional measurement of immunology marker changes from baseline to 3 months were also not feasible for this study, but would have enabled the same timepoint comparison with the brain volume analysis, contributing to the understanding of short-term changes in immunology alongside pseudoatrophy.

As GM and WM atrophy occurs with aging in normal controls [46‐49], the use of the HC group in this study as a comparison with the treated patients provided a control for the normal aging processes but not for time course of brain volume changes in patients with RRMS. In future studies, an active comparator group may help to establish the extent to which treatment-related pseudoatrophy is affected by treatment with IFN β-1a SC in comparison with other DMDs. In addition, more frequent measurements over the first 6 months and extension of measurement periods would allow the chance to better delineate the time course over which pseudoatrophy occurs in order to be able to differentiate it from true atrophy.