Potential Metabolite Markers of Multiple Myeloma

E-mail: wenzhou@csu.edu.cn

Background: Metabolism reprogramming is one of ten features in cancer. It is well known that metabolites in tumor microenvironment contribute to the survival and proliferation of cancer cells. Currently, a lack of detailed information about the metabolites profiling in bone marrow microenvironment limits us to understand the roles of metabolites associated with multiple myeloma(MM) and its diagnosis and treatment. Here we report a serum untargeted metabolomics study of MM patients, together with healthy donors(HD), with the aim of discovering metabolite markers associated with MM.

Materials and Methods: Gas chromatography-time-of-flight mass spectrometry (GC-TOFMS)-based metabolomics was used to analyze 140 serum subjects, including 81 bone marrow subjects(22 HD, 59 MM patients) and 59 peripheral blood subjects(27 HD, 32 MM patients). The bone marrow subjects were divided into training set(11 HD, 32 MM patients) and testing set(11 HD, 27 MM patients). SIMCA-14.1 software package was used to visualize the metabolite alterations between MM patient and HD through Principal component analysis (PCA) and orthogonal projection to latent structures discriminant analysis (OPLS-DA). Both the T-test and the receiver operating characteristic curve(ROC) analysis were performed by SPSS software. Metabolites in serum with higher fold change(FC) and variable importance in the projection(VIP) value(VIP > 1.5, P < 0.05 and FC > 1.5, P < 0.05, FDR < 0.05) were considered as biomarker candidates.

Results: A total of 117 and 123 metabolites were annotated from the detected spectral features in bone marrow serum subjects derived from training set and testing set, respectively. Based on multivariate statistical analysis(PCA and OPLS-DA) and univariate statistical analysis(T-test), a panel of 6 and 10 metabolites were identified as differential metabolites(VIP > 1.5, P < 0.05 and FC > 1.5, P < 0.05, FDR < 0.05) between MM patients and HD in training set and testing set, respectively, among of which 5 metabolites were found significantly altered in both sets. Creatinine and glycine were significantly elevated in MM patients compared with HD, while fatty acid consists of palmitic acid, petroselinic acid and stearic acida were found decreased in MM patients compared with HD. ROC analysis of these 5 metabolites resulted in an area under the receiver operating characteristic curve (AUC) of 0.922(95% confidence interval=0.748-1) in the training set and 0.923(95% confidence interval=0.853-1) in the testing set. Furthermore, the diagnostic potential of the metabolite signatures was assessed in peripheral blood subjects. Consistent with bone marrow subjects, metabolite signatures were significantly changed(VIP > 1.5, P < 0.05 and FC > 1.5, P < 0.05, FDR < 0.05) in peripheral blood subjects derived from MM patients compared with HD. The AUC of this metabolites signatures was 0.901(95% confidence interval=0.748-1) in peripheral blood subjects, implying that this panel of metabolites could be of potential clinical significance for the diagnosis of MM.

Conclusion: We conclude that a panel of 5 metabolites, including creatinine, glycine, palmitic acid, petroselinic acid and stearic acid, in serum has great potential in discriminating MM patient from HD. This metabolite signatures provides a novel and promising molecular diagnostic approach for the detection of MM.