Even though a disease may seem histologically similar, their underlying progression pathway at molecular level can vary vastly. To offer optimal treatment options, the identification of the patient’s disease subtype would be a critical step. Here, we developed a MRM-based subtype identification technology, with its application to pancreatic ductal adenocarcinoma (PDAC) patients. PDAC is a disease with one of the lowest 5-year survival rate where more than 90% of the patients show cold response to surgery or chemotherapy emphasizing the need for tailoring appropriate therapeutic options.

LC-MRM-MS/MS experiments targeting 153 targets were conducted across three replicates of 129 PDAC patient tissue samples using 6495C Agilent triple quadruple mass spectrometer combined with a dual-nanoflow LC system for subtype prediction. Subtypes of these PDAC patients were previously characterized into 6 distinct subtypes through an extensive proteogeonomic analysis. Based on these characterized subtypes, subtype-specific signature peptide sets were selected from each of these subtype which sequences were stable isotope labeled (SIL). These peptides were purified, AAA (amino acid analysis)-MS quantified and mixed together to generate a PDAC subtype identification mixture consisting of 153 signature peptides. With optimized MRM conditions and amounts for each of these peptides, LC-MRM-MS/MS experiments were performed., Area ratios of the SIL and endogenous peptide were taken to quantify endogenous peptide amount of each subtype, then taken to further select key subtype signature peptides. With these key peptides, a PLS-DA (Partial least squares-discriminant analysis) model was built resulting an average of 88.9% prediction accuracy and AUC of 0.905 across all 6 subtypes.

Correlation between survival rates and the predicted subtypes is planned to be examined to assess the value of PDAC subtype identification technology (PDAC-SIT) in clinical trials in selecting drug candidates as a method of predictive enrichment strategy.