by Christos Evangelou, MSc, PhD – Medical Writer and Editor
Amyotrophic lateral sclerosis, commonly known as ALS, is a rare and progressive neurological disorder that primarily affects motor neurons, the neuronal cells responsible for controlling voluntary muscle movement. Although the mechanisms underlying ALS are not fully understood, hexanucleotide repeat expansions in C9orf72, the gene encoding a protein expressed in motor neurons and other parts of the brain, have been linked to a familial (genetic) form of ALS. C9orf72 mutations are found in approximately 40% of patients with familial ALS.
A factor limiting our understanding of the role of C9orf72 mutations in ALS is the high clinical heterogeneity of the disease, even among patients carrying the same mutations in C9orf72. Understanding the role of C9orf72 in ALS pathogenesis may lead to novel preventive interventions and therapies for ALS, for which there is currently no cure.
In a recent study, researchers from the Euan MacDonald Centre for Motor Neuron Disease Research at the University of Edinburgh and the Institute of Medical Sciences at the University of Aberdeen used digital pathology and machine learning algorithms to systematically and quantitatively examine immunohistochemical markers of C9orf72-associated ALS.1 The study showed that cytoplasmic aggregation and nuclear accumulation of a protein called FUS and activation of the brain-resident inflammatory cells (microglia) were common features in ALS and predicted cognitive involvement in patients with C9orf72 mutations.
“Our work demonstrates that in patients with ALS, there are shared pathologies and pathways for which we already have existing drugs, for example, anti-inflammatory therapies,” said Dr. Jenna Gregory, pathologist and ALS researcher and corresponding author of the study. “Understanding how and when to intervene with these therapies is the next step,” she added. The study was published in the Journal of Pathology.
Approach: Using Digital Pathology to Determine Pathological Features in the Brain
Immune dysregulation and protein misfolding have been implicated in various neurological conditions, including ALS. For example, inflammation and protein misfolding can affect the severity of motor or cognitive symptoms in ALS. However, little progress has been made in targeting inflammation and proteostasis to prevent or treat motor or cognitive symptoms in patients with ALS.
“Understanding brain pathology at post-mortem in neurodegenerative diseases is crucial to developing our knowledge of the brain regions and cellular processes that underpin clinical syndromes in these diseases. However, human tissue has a great deal of heterogeneity, as does the clinical presentation of these diseases,” Dr. Gregory noted. “We are also necessarily subjected to our own biases of what we already know when we look at these tissues.”
She explained that, to circumvent clinical heterogeneity and identify histopathological features of ALS in an unbiased manner, they “looked at hundreds of images using non-biased digital pathology, extracting millions of pathological features.” The team used machine learning to assess which of these features, or which combinations of these features, are the best at distinguishing brains with motor neuron disease from brains that do not have the disease.1
Commenting on the novelty of this approach, Dr. Gregory said: “Our non-biased digital analysis and machine learning approach has not been done previously in a cohort of clinically deeply profiled cases of people with motor neuron disease. The combination of the associated clinical data and the extensive pathological profiling allowed us to make clinicopathological correlations of an unprecedented scale.”
Microglial Activation and Nuclear Accumulation of FUS Aggregates Are Markers of C9orf72-Associated ALS
The team used digital pathology to examine morphological and spatial markers of inflammation (i.e., glial activation) and protein misfolding in nearly 5,000 images of post-mortem brains from ten patients with familial ALS associated with C9orf72 hexanucleotide repeat expansions and ten healthy controls. They found that manually graded staining intensity was significantly correlated with the grading intensity determined through digital analysis. This finding suggests that digital image analysis is reliable for assessing morphological and spatial markers of inflammation and protein misfolding in post-mortem brain tissues stained using standard immunohistochemistry (IHC).
They then used a random forest model as a disease classifier to analyze digital brain images to identify IHC features associated with ALS status in individuals with C9orf72 mutations. This analysis showed that microglial activation was significantly associated with ALS status and language dysfunction.1 The IHC staining for CD68 (a microglial marker) and Iba1 (an inflammation marker) predicted disease status with high sensitivity (67% and 78%) and specificity (60% and 84%). The aggregation-prone RNA-binding protein FUS also predicted disease status, with 64% sensitivity and 69% specificity.1
Iba1 staining maintained its significant disease-predictive value across all brain subregions examined, although CD68 and FUS could act as disease classifiers only within certain subregions.
FUS Localization and Microglial Alterations Are Associated With Clinicopathological Features in ALS
Spatial analysis of FUS showed that its localization was a stronger predictor of ALS status than the FUS staining intensity. Specifically, cytoplasmic aggregation and nuclear accumulation of FUS were evident in the post-mortem brains of individuals with C9orf72-associated ALS.1
The team also found that microglial activation in the brains of individuals with ALS was significantly associated with aggregation of the DNA-binding protein TDP-43, which has been implicated in ALS pathology.1 Notably, CD68 staining intensity in the gray and white matter was significantly associated with language impairment, suggesting that microglial activation may play a role in language impairment in patients with ALS.
This unbiased approach facilitated the assessment and compilation of data on a scale that would not be possible without digital processing. Collectively, these data suggest that immune deregulation and proteostasis impairment may play an important role in the pathogenesis of ALS associated with C9orf72 mutations. The study also provides proof-of-concept evidence to support the use of digital pathology to better understand the mechanisms underlying ALS. “We identified that key pathological features included the cytoplasmic aggregation and nuclear accumulation of a protein called FUS and activation of the brain-resident inflammatory cells (microglia),” said Dr. Gregory. “Indeed, these features also predicted whether these patients had cognitive involvement,” she added.
Looking Ahead
Commenting on their next steps, Dr. Gregory noted: “Having identified the pathological correlates of disease state in these cases, we now intend to molecularly profile these cases to identify molecular pathways that could be used as therapeutic targets. We will also attempt to understand how these hallmarks of disease can be used as biomarkers to help us to identify patients at risk. Ultimately, we hope that these studies will result in clinical trials to improve the therapeutic outlook for these patients.”
References
- Rifai OM, Longden J, O’Shaughnessy J, et al. Random forest modelling demonstrates microglial and protein misfolding features to be key phenotypic markers in C9orf72-ALS. J Pathol. September 2022. doi:10.1002/path.6008
