Journal: Journal of proteome research
This study addresses the challenge of differentiating gastric cancers (GCs) from acid peptic diseases (APDs) using a noninvasive, serum-based assay.
Researchers developed a 22-minute mass spectrometry Multiple Reaction Monitoring (MRM) test to quantify a panel of serum proteins in 135 treatment-naive patients with GC, APDs, and healthy controls.
A novel deep neural network (DNN) scoring system, combined with SHAP for model interpretability, was used to classify samples.
The assay showed minimal technical issues and demonstrated high accuracy, including:
- AUROC of 0.95
- Precision over 0.90 in distinguishing GCs from APDs and controls
Protein levels measured by this assay were comparable in sensitivity to ELISA measurements of key proteins.
This liquid biopsy Laboratory Developed Test (LDT) offers promise as a prediagnostic screening tool to help guide clinical decisions for early GC detection.