Detection of mesothelioma from blood is challenging. Current proposed biomarkers mainly rely on single protein cancer products detected in serum or plasma. Here, we investigate if a mass spectrometry based multiplexed proteomic biomarker signature can improve accuracy of mesothelioma detection in serum.
We used targeted proteomics technology to investigate more than 400 serum samples from cohorts of mesothelioma and asbestos exposed donors from USA, Australia and Europe. Serum samples were processed for enrichment of N-linked glycoproteins on 96-well plates before peptide separation on ultra performance liquid chromatography (UPLC) followed by targeted analysis on a triple quadrupole type of mass spectrometer. The software Skyline was used for data visualization. Workflow for quantitative large scale data analysis was based on the software package MSstats.
We applied logistic regression models for a multiplexed signature of six peptides from six different proteins, including the biomarker mesothelin. In the receiver operating curve, signature had an area under the curve (AUC) of 0.76 in discriminating mesothelioma from asbestos exposed donors in a training set of 212 donors. AUC in a separated validation set of 193 donors was 0.72. In the validation set, AUC was 0.74 in separating mesothelioma early stages I/II from asbestos exposed. In comparison, single mesothelin peptide assessed by mass spectrometry discriminated mesothelioma from asbestos exposed with AUC of 0.71 in training and AUC of 0.64 in validation set, and mesothelioma early stages I/II were separated from asbestos exposed with AUC of 0.66 in the validation set.
Diagnostic strategies based on multiplexed targeted proteomics biomarkers bear potential of an increased accuracy for detection of mesothelioma in serum.
Clinical trial identification
Legal entity responsible for the study
James Thoracic Center, The Ohio State University Medical Center
All authors have declared no conflicts of interest.