Recursive ensemble feature selection provides a robust mRNA expression signature for myalgic encephalomyelitis/chronic fatigue syndrome.
Metselaar, Paula I, Mendoza-Maldonado, Lucero, Li Yim, Andrew Yung Fong et al. · Scientific reports · 2021 · DOI
Quick Summary
Researchers found a set of 23 genes in blood cells that could reliably identify people with ME/CFS. By looking at both gene activity and chemical markers on genes, they developed a test that correctly identified ME/CFS patients about 92-97% of the time. This is important because there's currently no reliable blood test to diagnose ME/CFS.
Why It Matters
ME/CFS lacks a specific diagnostic test, making diagnosis difficult and delays treatment. This study demonstrates a potentially robust gene-expression-based biomarker signature that could eventually enable objective diagnosis. If validated in clinical settings, such a test could improve diagnosis accuracy and standardize patient identification for research and clinical care.
Observed Findings
A 23-gene mRNA signature distinguished ME/CFS patients from healthy controls with 92% accuracy in the discovery cohort.
The signature validated on an independent platform with 95% accuracy.
48 DNA methylation markers (CpGs) associated with these genes predicted disease status with 97% accuracy across four cohorts.
Ten of the 23 genes were functionally related to immune system dysfunction.
The recursive ensemble feature selection method outperformed other statistical approaches.
Inferred Conclusions
A gene expression signature in peripheral blood cells can reliably distinguish ME/CFS patients from healthy individuals.
DNA methylation patterns corroborate the mRNA-based signature and may represent an alternative diagnostic approach.
Immune system dysfunction, as reflected in these gene signatures, is a central feature of ME/CFS pathology.
Remaining Questions
Can this biomarker signature distinguish ME/CFS from other chronic illnesses with similar symptoms (fibromyalgia, long COVID, autoimmune conditions)?
Does the gene signature change over the disease course, and can it track disease severity or predict treatment response?
What This Study Does Not Prove
This study does not prove that these genes cause ME/CFS or that targeting them will treat the disease. The biomarker signature correlates with disease status but does not establish causation. Additionally, the study used archived data and requires prospective clinical validation before determining whether this test can be reliably used in clinical practice.