Meta-Analysis Reveals Both the Promises and the Challenges of Clinical Metabolomics

Simple Summary A highly desirable approach to diagnose a human disease quickly and easily is to identify a chemical, protein, or antibody in a biofluid like blood or urine that is uniquely associated with the disease state and can serve as a diagnostic biomarker. Metabolomics is the study of the set of metabolites or small molecular-weight compounds found in cells, tissues, organs, and organisms. Thus, metabolomics is being widely applied to a variety of human diseases including various cancers to identify potential metabolite biomarkers for disease diagnosis and personalized medicine. Pancreatic ductal adenocarcinoma (PDAC) is the third-leading cause of cancer-related death and has the lowest five-year survival rate primarily due to the lack of an early diagnosis. A total of 24 metabolomics studies have been reported in the scientific literature that aimed to identify diagnostic biomarkers for PDAC. We analyzed the outcomes from these 24 studies in detail and observed a high level of inconsistencies in the identified metabolites that we attributed to a variety of experimental factors. Despite this negative outcome, we did identify a set of 10 metabolites that were consistently detected by several clinical studies and have the potential to serve as PDAC biomarkers and warrant further investigation. Clinical metabolomics is a rapidly expanding field focused on identifying molecular biomarkers to aid in the efficient diagnosis and treatment of human diseases. Variations in study design, metabolomics methodologies, and investigator protocols raise serious concerns about the accuracy and reproducibility of these potential biomarkers. The explosive growth of the field has led to the recent availability of numerous replicate clinical studies, which permits an evaluation of the consistency of biomarkers identified across multiple metabolomics projects. Pancreatic ductal adenocarcinoma (PDAC) is the third-leading cause of cancer-related death and has the lowest five-year survival rate primarily due to the lack of an early diagnosis and the limited treatment options. Accordingly, PDAC has been a popular target of clinical metabolomics studies. We compiled 24 PDAC metabolomics studies from the scientific literature for a detailed meta-analysis. A consistent identification across these multiple studies allowed for the validation of potential clinical biomarkers of PDAC while also highlighting variations in study protocols that may explain poor reproducibility. Our meta-analysis identified 10 metabolites that may serve as PDAC biomarkers and warrant further investigation. However, 87% of the 655 metabolites identified as potential biomarkers were identified in single studies. Differences in cohort size and demographics, p-value choice, fold-change significance, sample type, handling and storage, data collection, and analysis were all factors that likely contributed to this apparently large false positive rate. Our meta-analysis demonstrated the need for consistent experimental design and normalized practices to accurately leverage clinical metabolomics data for reliable and reproducible biomarker discovery.