Assessment of a 44 gene classifier for the evaluation of chronic fatigue syndrome from peripheral blood mononuclear cell gene expression.

Observed Findings

• A 44-gene classifier successfully distinguished CFS patients from healthy controls in the original derivation dataset
• When applied to a separate, blinded validation dataset from a new population, the same classifier correctly identified only ~67% of both CFS and healthy samples
• Individual genes performed poorly as standalone classifiers compared to the combined 44-gene metric in the original dataset
• High rates of misclassification occurred in the validation population despite successful performance in the derivation population
• Many previously published reporter genes for CFS appear to be study-specific rather than universally applicable

Inferred Conclusions

• Gene expression profiles differ between CFS patients and healthy controls, but these differences may be population-specific or study-dependent
• A combined multi-gene approach performs better than individual genes in the original population, but this advantage does not translate across different populations
• Current gene expression biomarkers are not yet suitable for clinical diagnosis of ME/CFS due to poor generalizability
• New validation strategies and larger, more diverse population studies are needed before gene-based diagnostics can be used in clinical practice

Remaining Questions