Brain atrophy in multiple sclerosis: mechanisms, clinical relevance and treatment options

1Multiple sclerosis (MS) is an immune-mediated disease that affects the entire central nervous system (CNS). Magnetic resonance imaging (MRI) lesions are well-scattered at white matter (WM) and grey matter (GM), while normal-appearing brain tissue in MRI also seems to be affected in pathological studies. Brain atrophy, the gradual loss of brain volume, is quite extensive in MS, nearly 0.5–1.35% per year, far off the limits of normal aging. It arises early in the course of the disease, accelerates with disease progression but is attenuated by disease-modifying drugs.

2There has been increasing interest in measuring tissue loss in CNS, as it represents the net effect of all destructive pathogenic processes during the disease course. It is worth recalling that neurons occupy almost half (46%) of the tissue volume, myelin is 24%, and glial and other cells almost 30%. GM holds much less myelin than WM (about one tenth), while neurons comprise its most abundant component. Relative to glial cells, oligodendrocytes outweigh the number of astrocytes, microglia and oligodendrocyte progenitor cells, although the exact percentage is still unknown.

3Atrophy in MS is often considered to be the result of extensive axonal transection and demyelination. The contribution of neuroglia may be less clear; reactive gliosis has the potential to mask considerable tissue loss in WM lesions. Measurement of brain atrophy is also considerably influenced by the amount of tissue fluids, which is increased by active inflammation and vasogenic edema in WM plaques, and decreased during treatment with agents with strong anti-inflammatory properties (pseudoatrophy effect).

4Transient volume changes could also be attributed to idiosyncrasic and technical factors. Dehydration may affect functional integrity of neuroglial cells, while decreased protein levels mainly affect synaptic densities. Unlike demyelination, water volume fluctuations and transient biological factors, neuroaxonal damage is irreversible in CNS, and atrophy is primarily considered to reflect this neurodegenerative component in MS. Finally, the atrophy rates may also be influenced to some extent by the genetic makeup of a person; Human leukocyte antigen (HLA) genotypes considered as ‘high risk for MS’, namely DRB1 and DQB1, have been associated with significantly lower WM and GM volumes, alongside with higher mean annualized percentage of brain volume change (PBVC) compared with medium and low risk HLA genotypes independent from patients clinical features (age, gender, disease course) or the DMTs used.

5Focal tissue loss in WM plaques is undoubtedly a major contributor to brain atrophy. However, the correlation between demyelination foci and whole brain atrophy is still a matter of debate. Some studies have found a strong association, while others have not, suggesting that separate pathologic processes may also contribute to tissue destruction.

6Chard et al. in a longitudinal 14-year study found that atrophy is more related to early rather than late focal lesion volumes. Inflammation may be an important contributor to global tissue loss in early disease stages (i.e. in clinically isolated syndrome). As the disease progresses, additional mechanisms emerge that are, at least partly, independent from WM injury, such as microglia activation, meningeal inflammation, iron deposition, oxidative stress and diffuse axonal damage in normal appearing white matter (NAWM). The lack of a significant relationship between white matter fraction (WMF) and T2 lesion load further support this hypothesis. Biopsy studies also confirm that the atrophy may proceed even in the absence of inflammation.

7Regional atrophy studies may also be helpful. Indeed, the volume loss of deep GM structures may be present in the early stages of the disease and it is strongly correlated with the disease course. In MS, brain atrophy may develop in different CNS structures and varies depending on the clinical disease phenotypes; ventricular enlargement is more prominent in relapsing–remitting MS [RRMS], whereas cortical atrophy seems to be more important in the progressive forms of the disease.

8All things considered, it has been suggested that the pathogenic trajectory of brain atrophy changes with disease progression; from primarily inflammatory to less inflammatory and primarily neurodegenerative in the late stages of the disease.

9In the initial stages of MS, many different components of the adaptive and the innate immunity induce demyelination and neuronal loss. The activation of auto-reactive CD4+ T lymphocytes in the peripheral immune system is necessary for their migration across the blood–brain-barrier (BBB) and into the CNS. After myelin destruction, T cells are in situ reactivated by antigens within myelin debris and their clonal expansion results in multifocal demyelinating plaques. Peripheral B lymphocytes are involved in the antigen presentation and initial stimulation of CD4 T cells. Also, they are an essential source of pro- and anti-inflammatory cytokines (IL-6 among others) promoting every autoimmunity response (driven by Th1, Th2, Th 17 cells) driving MS. In addition, the presence of chemokines (CXCL13) and survival factors (BAFF and APRIL) in the CSF of patients with MS, promotes the formation of meningeal follicle like structures, in progressive phases but also in early RRMS. T cells and B cells may, therefore, play an equally important role in the immunopathology of MS.

10Axonal destruction is quite extensive (up to 60–80%) in all active WM lesions and the extend of axonal loss is related to the number of immune cells within the plaques. Activated immune cells (T and B cells) and microglia/macrophages release a number of pro-inflammatory cytokines (e.g. TNFa, INFγ), proteolyticenzymes (e.g. perforin, granzymes) and free radicals (e.g. nitric oxide, glutamate) that can directly damage axons. Additionally, axons may die secondarily, due to the loss of pre- and post-synaptic signals (i.e. dying-back and Wallerian axonal degeneration) in regions far from the lesion site.

11Active MS lesions are characterized by profound heterogeneity regarding their demyelination pattern, which is persistent over time. The most commonly observed patterns are pattern II, which is a complement- and antibody-mediated demyelination, and pattern III, in which the initial event in lesion formation is a brief yet exorbitant oligodendrocyte injury. In other patients with RRMS, new lesions are associated with T cells, and activated microglia only. Pathologic heterogeneity across individuals in demyelination may imply different stimuli in the initial inflammation or different vulnerability to tissue loss across individuals.

12In WM lesions, inflammation and brain edema, demyelination, axonal loss, gliosis, and remyelination, all happen simultaneously. Brain edema which increases brain volume might bias atrophy measurements, but it resolves in the first few weeks after lesion formation. Notably, CNS has the capacity to use a great number of compensatory mechanisms (i.e. remyelination, redistribution of sodium channels, expression of neurotrophic factors etc.) to re-establish lost functioning to demyelinated foci.

13To conclude, tissue loss due to inflammation and demyelination maybe partly reversible in RRMS, while tissue loss and axonal damage due to mechanisms other than inflammation is irreversible, and remains the major component of brain atrophy especially in the progressive disease stages.

14While the destruction of CNS myelin is associated with clinical relapses, acute or late axonal loss is considered to be the main cause of permanent clinical disability in MS. Axons are more vulnerable to acute injury by inflammatory mediators, due to their shape and structure, compared to cell bodies or dendrites, while thin axons (< 2.5 μm in diameter) are mainly affected. Neurofilament light chain (NfL) protein is only expressed in neurons. It is an essential component of the axonal cytoskeleton, and reflects the axonal integrity and the stability of neurons. Under conditions of acute axonal transection, NfL are released and can be found as a result, in the cerebrospinal fluid (CSF) and blood of patients with MS. Of note, ultra high versus low blood NfL levels have been associated with MRI related (increased number of gadolinium enhancing or T2 lesion load, whole brain atrophy) and clinical measures (number of relapses, disability worsening) of disease activity and evolution and may, therefore, have prognostic value for patients and clinicians.

15Transected axons and ovoids are abundant in MS lesions but, abnormalities have also been reported in chronic inactive plaques, in normal appearing white matter (NAWM), and cortical areas, in which inflammation is less prominent. Therefore, additional mechanisms of axonal loss coexist with disease progression. It should be noted that these mechanisms have been postulated for both acute and late axonal loss (i.e. “late” signifying the absence of apparent inflammation):

16Several ion channels show compensatory changes a few weeks after demyelination a process that eventually promotes energy deficiency, and neurodegeneration. Aberrant expression of sodium channels, acid sensing ion channels, increased expression of maladaptive isoforms (i.e. Nav1.6 channels), paranodal (Caspr) and juxtparanodal (Kv1.2) protein alterations have also been detected in WM lesions, in NAWM, and GM. Alternation in the expression of these ion channels lead to intra-axonal calcium accumulation, and subsequent axonal degeneration and atrophy, particularly in secondary progressive MS.

17There has been increasing interest in the role of mitochondrial injury in MS demyelination and axonal destruction. In acute inflammatory lesions mitochondrial nicotinamide adenine dinucleotide-hydrogen (NADH) oxidase and complex IV defects (COX I) have been described, in axons, oligodendrocytes, and astrocytes. In chronic inactive plaques, ionic imbalance and high energy demands result to swollen and dysfunctional mitochondria, a phenomenon in which is partially reversed in remyelinating axons. There are also additional mtDNA deletions in GM structures of patients with SPMS. Furthermore, the respiration deficient neurons were diffusely distributed in the subcortical WM resulting in axonal loss in the absence of demyelination or inflammation. In oligodendrocytes, mitochondrial damage results in cell death and demyelination. Progenitor cells are also impaired, regarding their capacity to differentiate and produce myelin. Plus, genetic defects in mitochondrial genes potentiate MS lesions. From what can be deducted, mitochondrial dysfunction, in neurons and glia, is recognized as an important cause of atrophy and degeneration in MS and in other primarily neurodegenerative deceases such as Alzheimer’s disease and Parkinson’s disease.

18Iron [Fe] loading accumulates with age and in patients with MS, it can further increase oxidative tissue loss. In the CNS, iron is mainly stored in oligodendrocytes, binding with ferritin. Under conditions of oxidative stress, such as MS lesions, when oligodendrocytes are destroyed, free iron [Fe2+] is released in the extracellular space and becomes an additional source of reactive oxygen species (Fenton reaction: Fe2+ + H2O2 = Fe3+ + OH. + OH−). Further, iron is released by activated glial cells, which become dystrophic and disintegrate, leading to a second wave of Fe2+release.

19Diffuse T2hypointenselesions, which represent increased iron deposition are commonly found in patients with MS in cortical and deep GM areas (i.e. thalamus, basal ganglia, dentate nucleus and WM plaques. Notably, T2 hypointensity has been associated with brain atrophy and early axonal loss. Furthermore, in progressive MS, there is a significant decrease in iron levels in NAWM. Iron is important for myelin synthesis and neurogenesis, and iron depletion in normal appearing tissue, may further promote diffuse axonal loss and CNS atrophy.

20Several lines of evidence suggest that glutamate could also mediate injury to myelin, oligodendrocytes and neurons in the autoimmune experimental encephalomyelitis (EAE) model and in MS. Glutamate levels are elevated in CSF, in the centre of active plaques, on the borders of chronic active lesions, and in NAWM.

21There are two factors intertwining for glutamate accumulation: increased glutamate production by activated microglial cells and lymphocytes, and impaired clearance by resident cells such as astrocytes. High levels of glutamate lead to the over-activation of N-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors (permeable for calcium and sodium ions) and subsequent calcium overload and oligodendocyte and neuron cell death.

22Clinical symptoms and signs do not usually correlate with changes seen on conventional MRI measures (the “clinical-MRI paradox”). Whole brain atrophy, on the other hand, has a significant imaging association with physical disability as measured by Expanded Disability Status Scale (EDSS) score. In a longitudinal study, whole brain (WB) and cortical atrophy as well as other MRI related metrics such as the enlargement of ventricular CSF spaces have been associated with disability progression over a 10 year follow up. Furthermore, brain volume changes during the first year after disease onset, estimated by PBVC, were the best predictor of future neurologic impairment regardless of the intermediate relapse rate. Increased brain volume loss (BVL) has been correlated disability progression, independent from the number of previous relapses or the T2 lesion load in RRMS.

23In a similar vein, when patients with clinically definite MS were compared to patients with clinically isolated syndrome (CIS), at baseline, all brain volume metrics, except for cortical GM, were significantly lower in the MS cohort. Over a mean follow-up period of about 3 years, the annual PBVC values were significantly lower in CIS patients when compared to the MS cohort. Neuropsychological impairment, affecting mental speed processing, episodic memory, executive functions and attention, may be present in up to 50% of patients with MS and has been found to occur early in the disease course. Changes of brain parenchymal fraction (BPF) have been shown to predict cognitive impairment over 2 years in patients with early MS. Cortical atrophy was the best predictor of poor cognitive functioning, even when mild impairment was detected. Poorcognitive functioning has been associated with significant cortical thinning, especially in the fronto-parietal cortical and subcortical regions. Pravatà et al. specifically reported that the thinning of the right precuneus and high T2 lesion load were the best predictors of cognitive impairment. Strong correlations have also been reported between cognitive impairment and thalamic atrophy. Not surprisingly, patients with brain atrophy and higher education or high “cognitive reserve” are relatively protected against cognitive decline.

24Other clinical aspects of CNS atrophy include mood and personality disorders (i.e. euphoria, disinhibition, aggression, major depressive disorder) autonomic dysfunction and sexual disorders. Fatigue has been reported to be associated with GM atrophy in frontal regions and depressed patients were found to present selective cortical thinning in the fronto-temporal regions, while the frontal thinning was found to be the best predictor for depression in MS patients.

25Taken together, this growing body of evidence suggests that brain atrophy is a valid and sensitive measure of disease burden and progression in MS patients and may effectively be used in routine clinical practice and treatment trials.

26The need of agents to control the inflammatory process in multiple sclerosis pathology is obvious, but the need for medications to halt brain atrophy progression and neurodegeneration is also evident. Currently approved treatments for MS differ in their effects on brain atrophy (Table 1 for the first line therapies, Table 2 for the second line therapies and Table 3 for the emerging therapies).

27In general, studies of traditional injectable treatments have not exerted robust beneficial effects in the rate of brain atrophy. Intramuscular IFN-β-1a produced lower rates of brain volume loss (BVL) when compared to placebo during the second year of treatment in relapsing–remitting MS patients (− 0.23% vs − 0.51%; p = 0.03). However, the subcutaneous (sc) IFN-β-1a produced inconsistent results in both CIS and RRMS patients. BV data for intramuscular INF-β-1a in CIS patients and for subcutaneous INF-β-1b in relapsing MS patients has not been made available to date. The addition of monthly oral methylprednisolone pulses to subcutaneous interferon beta-1a treatment provided no further gain in normalized BV change in two published trials against placebo. The approved long-acting pegylated interferon beta-1a has only shown limited and inconclusive evidence for a beneficial effect on BV change in RRMS. A possible delayed effect in reducing brain atrophy has been reported for Glatiramer acetate [GA]. In the PReCISe clinical trial, GA failed to show an immediate effect on brain volume outcomes versus placebo (− 0.38% vs 0.33%), but the subsequent open label phase of the trial showed a clear–cut benefit on PBCV for the early treatment group, when compared to patients with delayed treatment onset (40% reduction, p = 0.0209). In relapsing–remitting MS, data from the extension phase of the European/Canadian GA trial come back as negative.

28Available oral therapies (Fingolimod, Teriflunomide, Dimethyl fumarate) have shown various effects on BV decline. Fingolimod has been reported consistentin reducing median PBVC by approximately 30 to 45% versus placebo or IFNβ-1a, in its three phase III clinical trials and their extensions. Of note, this reduction was observed as early as 6 months after treatment onset. In the extension phase of the TRANFORMS trial, patients switching from intramuscular (i.m.) INF β1a to FTY720 slowed their median PBVC, and patients continuing on FTY720 sustained low atrophy rates, over the following 4.5 years of therapy. However, no similar effects were reproduced in patients with the primary progressive form of MS, a finding that that could have otherwise strengthened the evidence for a direct action of fingolimod on brain cellular components. Finally, further condoning the aforementioned observations, in a study by Yousuf et al., cortical GM, alongside T2 lesion volume, remained stable in the cohort treated with fingolimod, as compared to the untreated group, where it decreased and increased respectively, in the first 2 years of treatment.

29Regarding Teriflunomide, brain volume outcomes have been reported for clinically isolated syndrome and relapsing- remitting MS in the TOPIC and TEMSO clinical trials respectively. Both doses of 7 mg or 14 mg failed to show a clear effect on slowing BVL when compared to placebo. However, when tissue specific volume changes were examined a significant reduction in the rate of WM loss was detected for the 14 mg teriflunomide treatment arm versus placebo. Similar results have recently been reported in 4 retrospective analyses of TOWER and TEMSO trials when an alternative method of brain loss evaluation was implemented.

30Dimethyl fumarate (DMF/BG12) showed a 21% reduction in BVL compared to placebo in the DEFINE study (the 240 mg twice daily regimen only) and produced only marginal but beneficial effects in BVL reduction in the CONFIRM study. A recent pilot study of 20 patients with RRMS showed a protective effect of DMF treatment in whole brain atrophy (PBVC: − 0.37 ± 0.49% vs. − 1.04 ± 0.67%, p = 0.005) and putamen atrophy (− 0.06 ± 0.22 vs. − 0.32 ± 0.28 ml, p = 0.02), but no effect on other subcortical volumes or total GM atrophy.

31Natalizumab, a monoclonal antibody against the cell adhesion molecule a 4-integrin, in two pivotal clinical trials was found to increase the rate of BVL in the first year of treatment and then significantly reduced it when compared to the placebo in the second year. Post–marketing observational studies confirmed that most of the BVL occurring while on Natalizumabtherapy takes place during the first months of therapy, and that it primarily involves WM volume changes. One trial has shown superiority of Natalizumab over conventional MS therapies (IFN-β and GA) and placebo regarding cortical atrophy. Recently, treatment with Natalizumabdid not affect the loss of brain volume compared to placebo in secondary progressive MS patients (ASCENT). Τhe study by Arpín et al. also suggests a neuroprotective effect of Natalizumab, after the measurement and comparison of the corpus calosum index, and the absence of brain atrophy in several patients under treatment during the follow up.

32Alemtuzumab, a monoclonal antibody against cells that express the CD52, has demonstrated greater MRI and clinical improvement in comparison to IFNb-1a in its three pivotal studies in active relapsing MS patients. Additionally, most patients remained free of disability and MRI progression, for the following 6 years of the initial treatment. Brain atrophy measures showed that brain parenchymal fraction was smaller in Alemtuzumab compared to the INF β-1a treatment arm either in treatment naïve patients or in participants who had relapsed on prior therapy. Extension studies showed sustained low brain atrophy rates, in the absence of continuous treatment with Alemtuzumab or other DMTs during the follow up period. The CARE-MS II 5-year follow-up study (2017) provided class III evidence that Alemtuzumab slows brain atrophy; the annual BVL rate continued to drop during the third year and remained low through the fourth and fifth year as well.

33The immune-modulatory agent Daclizumab in a 3-year post hoc analysis of 899 RRMS patients was compared to IFN beta-1a on brain volume change. Median annualised PBVC was significantly reduced in the DAC treatment group during both the first and the second year of treatment (baseline—24th weeks: − 0.674 vs − 0.745; 24th–96th weeks: − 0.511 vs − 0.549; all p < 0.0001) in comparison to INF β treatment, a finding which was consistent with previous longitudinal data.

34Ocrelizumab is a humanized mAb designed to target CD20+ B cells. MRI outcomes hint towards a positive effect on BVL and clinical disability progression. Treatment with Ocrelizumab has significantly slowed brain atrophy rates in comparison to INF-β1a (baseline to 96 weeks: 23.5% p < 0.0001 in OPERA 1 and 23.8% p < 0.0001 in OPERA 2) along with clinical disability. Ocrelizumab reduced the rate of whole BVL in PPMS from week 24 to week 120 by 17.5%120 (p = 0.0206) compared with placebo (ORATORIO).

35Several new agents are currently undergoing clinical development, including immuno-modulatory, neuroprotective or remyelinating compounds.

36Laquinimod, a linomide derivative, has also shown promising results on PVC rates in RRMS, most probably as a result of reduced astrocytic activation within the CNS. In the ALLEGRO clinical trial, after adjusting for baseline active inflammation, laquinimod markedly reduced BVL as compared to the placebo. Positive effects on PBVC are replicated in one active comparator trial [BRAVO] versus im IFN-β-1a. At present, the agent is further investigated in RRMS [CONCERTO] and PPMS patients [ARPEGGIO].

37Cladribine, an antiproliferative agent that takes effect by interfering with DNA synthesis, has shown significant effects in terms of relapse rate and disability progression. Data from CLARITY study suggested that at 18 months, patients treated with cladribine had 20%reduction in brain atrophy compared with patients receiving placebo. However, further studies are needed, in order to cladribine’s effect on brain atrophy rates, be fully elucidated.

38MS is an evolving disease, now considered of both inflammatory and neurodegenerative nature. Axonal injury and loss accounting for brain atrophy may be either acute (i.e. due to inflammation) or chronic/late due to pathogenic mechanisms primed by the preceding inflammation and later perpetuating with disease progression. Brain atrophy occurs as early as CIS, progresses faster than it does in healthy adults, and is the best predictor of future disability, physical and cognitive. It is widely accepted to be a valid, sensitive and reproducible measure of neuroprotection in MS research studies and therapeutic trials.

39There is now a variety of approved DMDs, with secondary neuroprotective properties, and an even greater number of novel compounds, in various stages of development and investigation. A firm belief remains that for a therapy to be effective in delaying the disease progression, its impact on axon and neuronal survival needs to be monitored. Conventional MRI findings (T1-hypotensive or T2 hypertensive lesion load) have already shown their limits for monitoring the disease burden and progression in MS patients. Newly introduced sophisticated imaging methods hold promise for the future of the clinical surveillance of the disease. Trials incorporating brain atrophy in their endpoints are providing accumulating evidence that rises substantial hopes for treating neurodegeneration in the near future.