Simultaneous detection of volatile and non-volatile metabolites in urine using UPLC-Q-Exactive Orbitrap-MS and HS-SPME/GC-HRMS: A promising strategy for improving the breast cancer diagnosis accuracy.

Breast cancer (BC) is the primary cause of cancer-related deaths in women. Currently, the discovery of biomarkers primarily relies on single platform, which might overlook other potential biomarkers and lead to inaccurate diagnoses. This study aims to: (1) expand the detection range of biomarkers through multiple analytical techniques, thereby improving the accuracy of BC diagnosis, and (2) analyze the metabolic pathways of the biomarkers to explore the metabolic mechanisms underlying BC. Urine samples from BC patients and healthy controls were analyzed using two techniques: Ultra-high performance liquid chromatography combined with Quadrupole-Exactive-Orbitrap mass spectrometry (UPLC-Q-Exactive Orbitrap-MS), and headspace solid-phase microextraction combined with gas chromatography-high resolution mass spectrometry (HS-SPME/GC-HRMS). Data from each platform was analyzed independently using both univariate and multivariate statistical approaches to identify candidate biomarkers. Subsequently, a mid-level data fusion approach was applied to integrate the candidate biomarkers identified by each platform. The fused data were used to construct orthogonal partial least squares discriminant analysis (OPLS-DA) models and random forest (RF) models, which were then compared against models based on individual platform. The fused RF and OPLS-DA models demonstrated enhanced diagnostic accuracy compared to the individual model. Integrating GC-HRMS and UPLC-Q-Exactive Orbitrap-MS achieved the best performance, with an AUC value of 0.967, sensitivity of 86.37 %, and specificity of 89.19 %. Metabolic pathway analysis revealed that 10 metabolic pathways exert an impact on BC. Four pathways-pyruvate metabolism, sulfur metabolism, taurine and hypotaurine metabolism, and tyrosine metabolism-were found to be associated with BC in both metabolomics and volatolomics studies, indicating that these pathways play pivotal roles in BC. This study confirmed the potential of merging multi-platforms to enhance the accuracy of BC diagnosis, offering new avenues for understanding the metabolic mechanisms of BC.