Early brain pseudoatrophy while on natalizumab therapy is due to white matter volume changes

Introduction

Brain volume loss is a common and early feature in patients with multiple sclerosis (MS) and it occurs at an accelerated rate when compared with healthy controls.^1,2 Because of the high clinical relevance of brain atrophy and the ability of magnetic resonance imaging (MRI) to reproducibly and reliably measure brain volume changes, atrophy outcomes have been incorporated into recent clinical trials;^3,4 however, although most disease modifying drugs (DMD) have been shown to improve the curve of brain atrophy accrual in MS patients through time, these drugs may initially produce a greater loss of brain volume in patients receiving active treatment, as compared to those receiving placebo. ³ Along these lines, the investigation of atrophy data from the AFFIRM trial (natalizumab versus placebo in treatment of relapsing–remitting MS) has demonstrated that there is an accelerated rate of volume loss during the first year of natalizumab therapy, which was interpreted as a pseudoatrophy effect. ⁵ It is not known whether pseudoatrophy is a global phenomenon, or whether it mostly affects grey or white matter. Following the marketing of natalizumab, several observational studies regarding the effectiveness and safety of treatment with natalizumab were published;^6–12 however, only a few studies included brain MRI data^7,10,12 and to our knowledge, none of them analyzed the changes in brain volume that occur in patients receiving natalizumab for this indication. The main objective of the present work is to assess the evolution and underpinnings related to brain volume changes during the first 2 years of natalizumab therapy, specifically focusing on grey and white matter volume changes and their relation to both global brain volume changes and the degree of baseline inflammation when undergoing natalizumab therapy.

Methods

Results

We identified 45 patients who had been treated with natalizumab for over 2 years in our center, who also had brain MRI examinations that were amenable to volumetric analysis with our SIENA and SPM5 software. We calculated PBVC for the first and second year of treatment, as well as GMF and WMF at both baseline and 12-month MRIs, for all 45 patients. All segmentation outputs were checked and we excluded six patients from further analyses due to missegmentation. Baseline demographic data and clinical features of the study samples (n = 39) are detailed in Table 1. There were no significant differences found with regard to baseline demographic, clinical and MRI characteristics when compared with the whole cohort of the natalizumab-treated patients in our center (data not shown). Gd-enhancing lesions were seen in 22 out of 39 patients’ baseline MRI scans. All patients had received at least one disease-modifying treatment before natalizumab. The first-line treatments received were as follows: interferon beta (89.7% of patients), glatiramer acetate (5.7% of patients) and an investigational drug (5.7% of patients). The median time from the first MS treatment onset to the first infusion with natalizumab was 48.3 months (range, 9.35–162.41).

OPEN IN VIEWER

Table 1.

Baseline characteristics of study participants.

	n = 39
Age at natalizumab onset (mean; SD)	34.95 (8.75)
Gender (female) (n; %)	28 (71.8)
Disease duration (years) (median; range)	10.98 (1.88–32.68)
Previous use of DMD (n; %)	39 (100)
1 DMD	24 (61.5)
2 DMD	10 (25.6)
> 2 DMD	5 (12.9)
Previous immunosuppressant treatment (n; %)	6 (15.4)
Patients with Gd-enhancing lesions (n; %)	22 (56.4)
Number of Gd-enhancing lesions (median; range)	1 (0–40)
MS relapses during the previous year (median; range)	2 (0–6)
Baseline EDSS score (median; range)	4.0 (1.5–6.0)

DMD: disease modifying drugs; Gd: gadolinium; EDSS: Expanded Disability Status Scale; MS: multiple sclerosis.

Global brain volume data

For the entire cohort (n = 39) the mean PBVC was -1.10% (SD 1.43%) during the first year and -0.51% (SD 0.96%) during the second year (p = 0.037). PBVC during the first year of treatment, but not during the second, was correlated with the number of Gd-enhancing lesions present on baseline MRI (Rho = -0.411, p = 0.009 for the first year; Rho = 0.050, p = 0.763 for the second). When analyzing brain volume changes based on the presence of baseline Gd-enhancing lesions, we found that the patients with baseline Gd-enhancing lesions had a greater brain volume loss during the first year, as compared to those without baseline MRI inflammation (p = 0.026). No differences were found between PBVC during the second year in patients with or without baseline Gd-enhancing lesions (p = 0.480) (Figure 1). Accordingly, brain volume loss occurred at a different rate during the two 1-year periods in those with Gd-enhancing lesions at baseline (−1.55% SD 1.59%, in the first year, versus −0.41% SD 1.01%, in the second; p = 0.005); but it occurred at a similar rate during the two periods, for those without any baseline Gd-enhancing lesions (−0.53% SD 0.98%, in the first year, versus −0.64% SD 0.92% in the second; p = 0.781) (Figure 1). In a linear regression analysis (adjusting for age, corticosteroid treatment in the month previous to the baseline MRI scan and treatment between the baseline and 1-year MRI scans), we found that the presence of Gd-enhancing lesions at baseline were able to predict the changes in PBVC (partial correlation, pr = −0.38; p = 0.022) during the first year of treatment.

OPEN IN VIEWER

Figure 1.

Brain volume changes during the first and second year of natalizumab treatment based on baseline gadolinium- enhancing lesions.

White and grey matter volume data

During the first year of therapy, we observed that a +1.15% (SD 4.92%) increase occurred for GMF (p = 0.151) and a −0.72 (SD 4.3%) decrease occurred for WMF (p = 0.017). In comparing WMF and GMF changes according to the presence of Gd-enhancing lesions at baseline, we observed that there was a trend toward a greater WMF loss during the first year of treatment in those patients with Gd-enhancing lesions at baseline (−2.81% SD 4.01%), as compared to those without (0.31% SD 4.36%) (p = 0.071). Such differences were not observed for GMF (+1.56% SD 5.13%, versus +0.62% SD 4.73%; p = 0.563) (Figure 2). In a linear regression analysis adjusting for the same variables detailed above, we found that the presence of Gd-enhancing lesions at baseline could predict the changes in WMF (pr = −0.37; p = 0.026) during the first year of treatment, while no predictors were found for changes in GMF.

OPEN IN VIEWER

Figure 2.

Grey and white matter volume changes in the first year of natalizumab treatment based on gadolinium-enhancing lesions.

Discussion

To the best of our knowledge, this is the first published study that evaluates global and regional brain volume changes occurring during natalizumab treatment, and investigates its relationship with pre-existing inflammation, showing that the pseudoatrophy effect involves only the brain’s white matter and that it can be predicted by pre-existing inflammation. Even though brain atrophy is present in the early phases of MS and it is known to be related to disability,^1,2,16,17 the underlying mechanism of brain tissue loss in patients with MS is not fully understood. Some proposed mechanisms are directly related to the inflammation and lesion formation processes,^1,18,19 while others would be independent of inflammatory phenomena.^1,16,20–23 DMD treatments for MS have been shown to have a clear beneficial effect, reducing relapses and inflammatory lesions; however, the brain volume changes occurring in MS patients who are on disease-modifying therapy are sometimes difficult to interpret, as they may be influenced by inflammation present at therapy onset.^3,4,18

Regarding natalizumab treatment, the MRI volume measurements of the AFFIRM trial ⁵ showed that although the mean percentage of reduction in brain parenchymal fraction (BPF) was similar between both treatment arms (placebo versus natalizumab) during the 2 years of the study, patients receiving natalizumab had an accelerated rate of atrophy during the first year, as compared to placebo; this trend reverted during the second year, similar to what was previously observed in the intramuscular interferon beta 1a pivotal trial in RRMS patients.^4,24 The authors of the AFFIRM trial attributed these dual findings to an initial pseudoatrophy phenomenon and to a potential beneficial effect for natalizumab therapy in preventing axonal loss, by decreasing new lesion formation during the second year of treatment. ⁵ Pseudoatrophy is an accelerated brain volume reduction that is thought to be attributable to a loss of cerebral water content, due to the resolution of the disease-related abnormal permeability of the blood-brain barrier and not as a consequence of cell death and tissue injury. ⁴

For a better understanding of this phenomenon, we first evaluated global brain volume changes occurring during the first 2 years of natalizumab treatment, with a special focus on the relationship between pre-existing inflammation and the subsequent brain volume loss, as well as its relationship with disability scores. In the whole group, we found there was an accelerated loss of brain volume during the first year, as compared to the second, similar to what was already described in the AFFIRM trial. ⁵ We found that brain volume loss was more pronounced for those patients with baseline Gd-enhancing lesions. Furthermore, the number of baseline Gd-enhancing lesions correlated with the amount of brain volume loss during the first year, but not during the second one. Previous studies demonstrated that brain volume, and specifically white matter volume changes, may be influenced by the presence of active lesions and it has been suggested that grey matter volume changes would be insensitive to pseudoatrophy effects.^25–27 In order to study whether this accelerated brain volume loss during the first year of treatment was a global phenomenon or was mostly due to either grey or white matter changes, we calculated the grey and white matter brain fractions during this period of treatment; we found that brain volume loss occurring during the first year of natalizumab treatment was mostly due to white matter volume changes and that the presence of Gd-enhancing lesions at baseline predicted those subsequent changes in white matter volume. Taken together, our findings support the idea that volume loss occurring in the first months of natalizumab therapy is due to the blood-brain barrier stabilization and the abrupt resolution of ongoing white matter inflammation. Global brain volume changes during the second year were similar in groups with or without baseline Gd-enhancing lesions, meaning that after the resolution of inflammation, other disease mechanisms are to be held responsible for brain volume changes. It might still be surprising that such a relevant pseudoatrophy effect could occur, even though all patients had received at least one DMD before commencing natalizumab; however, all these patients switched to natalizumab because of first-line treatment inefficacy, and up to 56.4% of them presented baseline Gd-enhancing lesions.

To date, no studies assessing grey and white matter changes on natalizumab-treated patients have been conducted and there are very few studies published about DMD-treatments.^28,29First-line DMD have shown to prevent the appearance of new cortical lesions ²⁸ and to slow the progression of grey matter atrophy.^28,29 The increase in grey matter volume during the first year of natalizumab treatment that we observed in our sample could be due to either a neuroprotective effect of natalizumab, ⁵ a reduction in the formation of cortical lesions (that were not detectable with the sequences used), or both. Interestingly, we observed a greater GMF increase in patients with baseline Gd-enhancing lesions, as compared to those without, although this was not statistically significant. This might suggest that a potential beneficial effect of natalizumab on grey matter damage could be especially relevant to patients with more severe blood-brain barrier disruption; however, the lack of a control arm does not allow us to ascribe all grey matter changes to the therapy.

We have demonstrated a significant association between the disability changes observed during the first year of treatment and the number of Gd-enhancing lesions at baseline, but not with brain volume changes occurring in the same period. This lack of association between the EDSS change and PBVC during the first year may be partly explained by the confounding pseudoatrophy effect, so that patients with higher baseline inflammatory activity will be those with a greater improvement on the EDSS score, due to a reversion of inflammation-related transient disability; but they are also those at a higher risk of brain volume loss. Also, EDSS changes were not associated with grey nor white matter fraction changes.

Concerning the absolute figures for volume changes, we found almost 2-fold values of brain volume loss in both 1-year periods, as compared to the AFFIRM trial (in our cohort, PBVC of −1.02% for the first year and −0.59% for the second; in the AFFIRM trial, BPF changes of −0.56% for the first year and −0.24% for the second). ⁵ The fact that our cohort had more aggressive disease, with a higher mean number of baseline Gd-enhancing lesions, than patients in the natalizumab arm of the AFFIRM trial (clinical trial’s mean number of Gd-enhancing lesions was 2.2 ± 4.7 versus 4.9 ± 7.9 in our cohort) could partly explain these findings. However, despite having patients with a more aggressive disease, we observed a 95.4% reduction in patients with Gd-enhancing lesions and a 90% reduction in patients with new T2-lesions at 24 months, similar to what was reported in the AFFIRM trial. ⁵

Some limitations of our study should be taken into account. Firstly, and most importantly, because there was not a control group, the brain volume changes cannot be fully attributed to natalizumab therapy; however, this does not affect the main conclusions of the present study, that indicate that global brain atrophy is not an appropriate surrogate marker of disability in such settings, as it is importantly affected by changes in the patients’ white matter volume due to a resolution of their ongoing inflammation. Secondly, as some evidence seems to indicate that this pseudoatrophy effect may be occurring in the very first months after treatment initiation, ²⁴ it is possible that MRI scans performed earlier than the 12-month point might have detected such an effect.

In summary, this study leads us to conclude that the important and early brain volume loss occurring during the first year of natalizumab therapy, but not during the second, was related to inflammatory activity at the onset of therapy and that it was mainly due to white matter volume changes, which supports the idea that brain volume changes during the first year of treatment are mostly due to a pseudoatrophy effect.