Brain Magnetic Resonance Imaging (MRI) in Spinal Muscular Atrophy: A Scoping Review

Inclusion criteria

The inclusion criteria used to select the articles for the review were based on the population and concept elements reported below.

Literature search

Published literature on brain imaging in 5q SMA was identified through database searches up to April 2022. Selected keywords were combined to create search strategies that were adjusted for each database. Articles were searched in PubMed/MEDLINE and Ovid EMBASE + EMBASE Classic from 1991–2022. To identify relevant articles, Medical Subject Headings (MeSH terms indexed on PubMed) and key terms were used such as ‘spinal muscular atrophy’, ‘Werdnig Hoffmann’, ‘Kugelberg-Welander’, ‘Dubowitz disease’, which were combined with acronyms for neuroimaging modalities such as ‘MRI’, ‘fRMI’, ‘DTI’, and ‘DWI’. Language was restricted to English and French, however it is possible that some articles in French were missed due to the use of English search terms. Retrieved literature was screened and all relevant articles were used for this review. The reference section of relevant articles was manually searched to identify additional articles of relevance.

Selection of studies

Articles were initially screened by two reviewers (NM and MO) based on titles and abstracts according to the population and concept elements previously described (data on brain magnetic resonance imaging) in SMA types 1 to 4. Full-text peer-reviewed articles and conference abstracts in English or French were selected. The articles were examined by two authors (NM and MO) and eligibility for inclusion was performed independently.

Data extraction and presentation

All relevant data to inform the scoping review objectives and questions were extracted and are summarized in Table 1. Study characteristics related to publication (reference), population (SMA type, sample size, sex, age), and study design were first extracted. The main findings of each article relevant to brain MRI were extracted as well as other clinical findings. Risk of bias assessment was performed by one certified author (MO) using the American Academy of Neurology framework [11].

Search results

The initial search conducted in July 2021 identified 891 articles across the two databases that described neuroimaging findings in SMA. A total of 137 duplications were removed and after screening by title and abstract a total of 15 articles were included in the full-text screening. Twelve articles were included in the final review (Fig. 1). All articles included patients either genetically diagnosed and confirmed to have 5q SMA or diagnosed by muscle biopsy. An updated literature search was performed in April 2022, identifying no additional articles of interest.

Fig. 1

Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) Flow Diagram of the study.

SMA type 0

Three class IV studies reported on five patients with SMA type 0, showing progressive changes with prolonged survival. All infants had molecular diagnosis confirmation and fetal onset of symptoms. In one case report, no abnormalities were detected on brain MRI of a newborn despite a clinical diagnosis of moderate hypoxic ischemic encephalopathy (HIE) and undergoing therapeutic hypothermia. This infant was extubated and died at 11 days of age [12]. Two studies described a pattern of progressive atrophy and loss of white matter observed by MRI with prolonged survival in infants who remained on permanent ventilation [13, 14]. In a case series of three children under 3 years old, all with one copy of SMN2 and no documented hypoxic ischemic encephalopathy (HIE) at birth, Mendonça and colleagues initially described at 2 months supratentorial atrophy, a widening sulcus and ventricles, and a tapered corpus callosum. Imaging follow-up at 11 months (patient 3), 1 year (patient 1), and 3 years (patient 2) documented a marked progression of supratentorial and hippocampal atrophy and severe reduction in white matter [13]. Moreover, in a longitudinal MRI assessment as a neonate, at 7 months, and at 2 years of age without documented neonatal HIE, Maeda et. al observed progressive central nervous system (CNS) atrophy of the cerebral cortex, subcortical white matter, thalamus, and basal ganglia [14]. In contrast, Mendonça et. al reported no abnormalities in the thalamus and basal ganglia [13].

SMA type 1

Four class IV studies reported on four patients with SMA type 1. A case report of a 7-year-old patient described diffuse cerebral atrophy, dilation of the 3rd and lateral ventricles, and significant white matter atrophy as well as a hypoplastic corpus callosum [15]. T2-weighted images revealed high intensity areas around the posterior horns of the lateral ventricles. Similarly, Ito et al. reported high intensity areas around the posterior horns of the lateral ventricles based on T2-weighted and FLAIR (fluid attenuated inversion recovery) images and found high intensity areas in anterolateral portions of the bilateral thalami [16]. Moreover, lesions in the thalamus and cerebellum were observed in one patient [17]; however, the cerebellum and brainstem were normal in another patient [15]. No brain abnormalities were reported in another case of a newborn with type 1 SMA; yet postmortem examination revealed neurodegeneration in cortical and subcortical brain structures [18].

SMA type 2

Only one class IV case report of brain MRI in an infant with type 2 SMA was identified [19]. Delayed myelination with high intensity areas around the posterior horns of the lateral were found, however the cerebral hemisphere, cerebellum, and brainstem were normal.

SMA type 3 and 4

One class IV and two class III studies reported on 29 patients with SMA type 3 & 4. In a case series of four young children with type 3 SMA patients and one case report on two-year old, no structural brain abnormalities were observed on MRI [20, 21]. A larger prospective case control study of 25 type 3 and type 4 adult SMA patients described a greater cortical volume in those with type 4 SMA compared to type 3 SMA in the left motor cortex. Moreover, they found an increase in grey matter density in the bilateral motor cortex and extra-motor regions in both SMA types compared to control [22]. No white matter abnormalities were observed in either type 3 or 4. Using a similar case-control group, a significantly smaller total volume of cerebellar lobes was observed in SMA patients compared to controls. In particular, grey matter in one lobule was significantly smaller in volume compared to controls [23].

SMA type 0

Common neuropathological features marked in subcortical brain structures including the thalami, basal ganglia and cerebellum showed to be variably affected. The involvement of the thalami, substantia nigra and cuneate nuclei have been well documented in autopsy examinations [7, 24, 25], and the most severe cases of type 0 SMA additionally showed severe and diffuse CNS atrophy [13, 14], distinct from changes secondary to HIE. Imaging follow up in four type 0 patients (less than 3 years of age) highlighted widespread and progressive abnormalities over time: supratentorial brain atrophy with white matter reduction, severe hippocampal atrophy, and a thinned corpus callosum. Given that the SMN protein is highly expressed during fetal development up until early infancy in various brain structures of the nervous system [26], it could be hypothesized these significant abnormalities may be related to the reduction of SMN protein observed during this critical time period. In contrast, Vincent & McSweeney found no brain pathology in an infant of SMA type 0 in the neonatal period [12]. Most of the reported structural changes were observed over time, and they may be secondary to non SMN pathways in a chronically ill child.

SMA type 1 and 2

High signal intensity areas were observed around the lateral ventricles [15, 16, 19] and the thalami [16]. These could reflect unmyelinated areas or dysmyelination. A previous study found significant post-mortem cell death in the developing telencephalon at both the dorsal and ventral side of lateral ventricles [27]. It is not clear the type of cells described by the authors, but it is known that the ventricles are surrounded by ventricular and subventricular zones in which reside proliferating cells that contribute to white matter repopulation [28]. Based on the process of positive reinforcement in exuberance selection during development, the observed change in white matter is likely a consequence of a lack of myelin maturation during development. Lower SMN levels lead to motor neuron degeneration, which can lead to an absence of positive reinforcement in those connections and subsequent lack of myelin maturation. It would be less plausible for these observed changes to be caused by an active demyelination process. However, more research is needed. Abnormal development of the corpus callosum was also observed in two individuals [15, 19]. The corpus callosum is critical for inter-hemispheric communication and shows consistent reduced integrity in another motor neuron disorder, amyotrophic lateral sclerosis [29]. It is also important to note that all children were genetically confirmed for 5q SMA except for one child with SMA type 1 who underwent a muscle biopsy for diagnosis [15]. The authors performed additional diagnostic procedures and made longitudinal assessments of the child to further support this diagnosis. Nonetheless, without genetic confirmation of SMA, findings from this report should be interpreted with caution of potential misclassification bias.

In contrast to type 1 findings of significant abnormalities in cortical and subcortical structures, one case report of an infant with type 2 SMA revealed a normal brainstem, cerebral hemisphere and cerebellum besides white matter abnormalities [19]. The nature of type 1 SMA being the most severe may help explain these contrasting findings, however autopsy examinations in two other individuals with type 2 SMA revealed a reduced number of neurons in the primary motor cortex and fewer myelinated fibers in the spinal-pyramidal tract [30]. These differences could also be attributed to timing of the imaging, where some chronic changes could be missed in younger children.

SMA type 3 and 4

In later onset SMA (types 3 and 4) evidence of cerebellar pathology and changes in grey matter density in motor areas were identified in the largest adult cohorts of SMA patients [22, 23]. Increased cortical volume in the motor cortex of SMA type 4 compared to type 3 potentially suggests a reorganization of the grey matter in the motor cortex and an adaptive reorganization mechanism with progressing lower motor neuron degeneration [22]. Moreover, cerebellar function has historically been related to movement and balance and more recently in non-motor tasks such as sensorimotor control [31]. Cerebellar grey matter volume was found to be significantly reduced [23]. The three cerebellar lobules affected share the final target of proprioceptive output which is closely tied to the control of movement. In a recent cerebellar study in mice, researchers found significantly lower network activity in the cerebellum suggesting neuronal dysfunction that may follow volume loss [32].

These anomalies prompt inquiries about the potential impact of neurodegenerative and neurodevelopmental processes on the brain. SMN protein plays a crucial role in establishing the nervous system during embryonic and early postnatal stages, as well as maintaining motor neurons. Thus, it is plausible that the observed neurodegeneration may result from impaired neurodevelopmental processes. In studies of severely affected children with early-onset SMA, progressive degeneration across brain structures has been observed. In adults with late-onset SMA, cross-sectional examinations show an increase in gray matter and a reduction in cerebellar volume, indicating compensatory and degenerative processes. Since the SMN protein is reduced across the entire spectrum of the disease, both neurodevelopmental and neurodegenerative effects possibly overlap and contribute to the observed changes in the brain.

The clinical consequences of the brain abnormalities remain uncertain, with limited and variable data available from reported cases. The more widespread CNS atrophy seen in the most severe end of the clinical spectrum suggests a potential correlative relationship between disease severity and brain involvement. Interestingly, key motor regions such as thalamus, basal ganglia, cerebellum are often affected. In individuals with later-onset SMA, we observed different patterns of brain involvement, specifically increases in grey matter volumes in the motor cortex in patients with type 4 SMA compared to type 3 [22] as well as decreases in cerebellar volume [23]. The increase in grey matter volume suggests a compensatory process, and it was found to be positively correlated with longer disease duration and slower progression rates seen in type 4 SMA individuals. With cerebellar imaging findings, no significant correlations were found between clinical features or scales.

We provide a comprehensive overview of the available evidence of structural brain involvement detected by MRI across the SMA spectrum. A previous review published in 2021 had reported MRI findings only in SMA type 0/1 individuals [8] censoring their search to 2019. They identified three studies of brain MRI, all included in our review, while we identified an additional four studies in this population. They also concluded that several structures, including the thalami, basal ganglia, temporal and frontal cortices, hippocampi, and cerebellum were variably affected, with changes observed over time in more severely affected children.

Limitations

There are several limitations inherent to the literature reviewed in this scoping review. Case reports are particularly vulnerable to publication bias. Though we had reports of normal brain MRI findings in this scoping review, the overall limited number of studies identified may have been affected by publication bias.

Another limitation is the methodological quality of the papers. Only three studies out of twelve disclosed the type of MRI sequence (T1-weighted, T2-FLAIR, T2-weighted) used to capture the brain abnormalities. The type of sequence used is particularly important as different sequences are better suited to detect different structural abnormalities. It is also important to note the variability that comes with reading and analyzing MRI images. While many reports appear to have been obtained clinically by the site’s radiologist, no studies included in this review described the level of expertise of the radiologist analyzing the images. As such, it is possible that some abnormalities were missed or misinterpreted by some and not others. A standardized protocol for image acquisition and reporting should be used to ensure consistency and reliability. The small sample sizes, limited number of studies, and heterogeneity of neuropathological features present in some patients and not others, hampers the interpretation and generalizability of the findings.