Microarray based diagnosis profits from better documentation of gene expression signatures

Area of Science:

• Biotechnology
• Bioinformatics
• Medical Diagnostics

Background:

• Microarray gene expression signatures offer potential for disease diagnosis and prognosis.
• Current documentation standards lack clarity, hindering external validation and clinical adoption.

Purpose of the Study:

• To evaluate the consistency of diagnoses between internal and external studies using microarray gene expression signatures.
• To demonstrate the impact of improved documentation on the reliability of these signatures.

Main Methods:

• Analysis of data from eight publicly available clinical microarray studies.
• Evaluation of classification consistency using documented information only.
• Comparison of diagnostic discrepancies before and after implementing a 'documentation by value' strategy.

Main Results:

• Significant discrepancies (median 18%, worst case 30%) observed between study-internal and study-external diagnoses.
• The proposed 'documentation by value' strategy, including quantitative preprocessing information, reduced the median discrepancy to 1%.

Conclusions:

• Current documentation practices for microarray signatures are insufficient for unambiguous application.
• Enhanced documentation, specifically 'documentation by value', is essential for reliable evaluation and clinical implementation of gene expression signatures.
• Improved documentation will accelerate the translation of microarray signatures into clinical practice.