Sample Pooling and Inflammation Linked to the False Selection of Biomarkers for Neurodegenerative Diseases in Top–Down Proteomics: A Pilot Study

Proteomic technologies have been recently adapted to the new field of clinical proteomics. The origin of errors and biases has been well-identified in the pre-analytical steps, leading to the measurement of clinical analytes. One possible source of inadequacy in clinical proteomics is linked to sample pooling. This practice is usually related to low sample availability, variability, experiment time/cost. In this study, we first asked whether sample pooling in top–down proteomics is suitable to obtain a relevant biological average. Our second objective was to identify inflammatory biomarkers of outlier samples in our population of Creutzfeldt-Jakob disease patients. Our results demonstrated that, in a proteomics study, sample pooling as well as the inflammation status was an important source of errors: missed detection of biomarkers and false identification of others. Pooled samples were not equivalent to the average of biological values. In addition, this procedure reduced the statistical value of the identified biomarkers due to a stabilization of their standard deviation and rendered outlier samples difficult to detect. We identified serum amyloid A as a candidate biomarker of outlier samples. The presence of this protein, which could be explained by inflammatory processes, induced major modifications in the sample profiles.