Whole blood transcriptome analysis in amyotrophic lateral sclerosis: A biomarker study

Area of Science:

• Genomics
• Neuroscience
• Molecular Biology

Background:

• Amyotrophic lateral sclerosis (ALS) pathogenesis and diagnostic markers remain poorly understood.
• Gene expression analysis offers a pathway to identify differentially regulated genes and potential biomarkers for ALS.
• Current diagnostic challenges necessitate advanced molecular approaches to elucidate disease mechanisms.

Purpose of the Study:

• To identify differentially expressed transcripts in amyotrophic lateral sclerosis (ALS) using a two-stage transcriptome-wide study.
• To develop and validate machine learning models for discriminating ALS patients from healthy controls.
• To assess the utility of whole blood transcriptome profiles in differentiating ALS from clinically similar conditions.

Main Methods:

• A two-stage transcriptome-wide analysis involving 397 ALS patients and 645 control subjects.
• Application of three distinct classification models: support vector machines, nearest shrunken centroids, and LASSO.
• Validation of model performance in distinguishing ALS from controls and from mimic syndromes.

Main Results:

• Identification of 2,943 differentially expressed transcripts, primarily associated with RNA binding and intracellular transport.
• High accuracy achieved in discriminating ALS patients from control subjects using validated machine learning models.
• Significantly reduced accuracy when attempting to differentiate ALS from diseases with similar clinical presentations; survival prediction was not feasible.

Conclusions:

• Whole blood transcriptome profiles can illuminate biological processes implicated in amyotrophic lateral sclerosis (ALS).
• Distinguishing ALS from mimic syndromes using transcriptome profiles presents a significant challenge, even after accounting for batch effects.
• Further research is needed to refine diagnostic and prognostic biomarkers for ALS.