Screening and diagnosis of triple negative breast cancer based on rapid metabolic fingerprinting by conductive polymer spray ionization mass spectrometry and machine learning

Abstract

We present the use of conductive spray polymer ionization mass spectrometry (CPSI-MS) combined with machine learning (ML) to rapidly gain the metabolic fingerprint from 1 μl liquid extraction from the biopsied tissue of triple-negative breast cancer (TNBC) in China. The 76 discriminative metabolite markers are verified at the primary carcinoma site and can also be successfully tracked in the serum. The Lasso classifier featured with 15- and 22-metabolites detected by CPSI-MS achieve a sensitivity of 88.8% for rapid serum screening and a specificity of 91.1% for tissue diagnosis, respectively. Finally, the expression levels of their corresponding upstream enzymes and transporters have been initially confirmed. In general, CPSI-MS/ML serves as a cost-effective tool for the rapid screening, diagnosis, and precise characterization for the TNBC metabolism reprogramming in the clinical practice.

Recent Contributions of Mass Spectrometry-Based “Omics” in the Studies of Breast Cancer

“…A large number of statistical tools are used for these omics studies, which are listed by the names. Pathway analysis, clustering analysis (e.g., Hierarchical Clustering, K-means), network analysis (e.g., protein-protein interaction networks), functional annotation and gene ontology analysis, pathway enrichment analysis (e.g., GO enrichment, KEGG pathway enrichment), , Bayesian analysis, cox proportional-hazards model, nonparametric statistics (e.g., Wilcoxon Rank-Sum Test), multivariate analysis ,, (e.g., partial least squares, principal component regression), weighted gene co-expression network analysis (WGCNA), gene set variation analysis (GSVA), time-series analysis, copy number variation analysis (CNV), functional regulatory network inference co-expression analysis, gene regulatory network inference, functional genomic screening, comparative genomics analysis, allelic specific expression analysis (ASE), network topology analysis, survival regression analysis (e.g., Cox PH model), metagenomic functional annotation, time-series omics analysis, and hidden Markov models (HMMs) in genomics and oroteomics. For machine learning-based omics analysis a variety of machine learning techniques, e.g., random forest, support vector machines, and feature selection, are also employed. , Bioinformatics software is used in conjunction with the variety of recent MS-based techniques, and applications in breast cancer have been outlined in various reviews published recently.…”

“…Song (2022) developed a MS-based method for rapid metabolomic fingerprint generations for screening and diagnosis of TNBC . Lipidomic markers for TNBC detection are attempted.…”

“…A total of 76 metabolite markers with expression levels of their corresponding upstream enzymes and transporters were verified to predict this subtype. 30 Immunohistochemistry-assisted laser ablation inductively coupled plasma mass spectrometry is used for quantitative analysis of mapping biomarkers. As a new tool for visualizing tissue sections for vital markers like MUC1 and HER2, Seuma (2008) employed metal-tagged antibodies that were subjected to laser ablation for viewing the proteins in multiple line-rastered tissue sections.…”

“…75 Song (2022) developed a rapid metabolomic fingerprint generation for screening and diagnosis of TNBC. 30 A few molecules and proteins from different independent studies have been reported to affect the treatment responses of this subtype. Zhou (2023) identified OTUD1, a chemosensitization factor for doxorubicin exerting its effect by modulating P16 expression.…”

“…The authors suggested that this approach with the machine learning algorithms developed could serve as a rapid, costeffective diagnostic tool for detection of TNBC specific metabolites. 30 Huang (2022) developed a fast screening of serum metabolic fingerprints through nanoparticle-enhanced laser desorption/ ionization mass spectrometry (NPELDI) to distinguish BC from nonBC. A 7 metabolic signature consisting of 3,4-dihydroxy benzylamine, uracil, histidine, dehydrophenylalanine, thymine glyceric acid, and nicotinamide could distinguish the disease, of which the first 4 were upregulated and the last 2 were downregulated in healthy donors (and in benign tumors).…”