Metabolomic Research On Plasma Samples Of Non-Small Cell Lung Cancer Patients Response To Bevacizumab

Objective:Byusing 1H-NMR-based metabolomics research method,we detectedmetabolic changes of plasma ofhealthy volunteers and NSCLC patientswhowere treated by bevacizumab.Metabolomic data analysis was performed using PC A,PLS-DA,LSVM method to investigate differential metabolites and metabolic pathways and observe their relationship with clinical efficacy and finallyidentify the appropriate markers for screening patients,assessing and predicting treatmentefficacy,and guiding individualized treatment of anti-cancer drugs.Methods:We collected plasma from 30 healthy volunteers without serious medical illness and from 55 lung cancer patientsthen we detected the metabolites in these samples using nuclear magnetic resonance(NMR)technology.Seven kinds of 59 metabolites was validated through data base comparing from 176 plasma samples,the concentrations of metabolites were also confirmed.We identified the endogenous metabolite changes associated with bevacizumab treatmentbyusing unsupervised and supervised machine learning model analysis.To predict the efficacy of bevacizumab,we trained a total of 14 classifiers and screened them for optimization and validation of the best-performing Linear Support Vector Machine(LSVM)classifier.Results:PC A and PLS-DA were used to perform multivariate data analysis on metabolite data,a separatetest was used to filtarate potential biomarkers.Potential biomarkers associated with the diagnosis of non-small-cell lung cancer include 16 small molecule metabolites such as citrate,glucose,and serine;differential metabolites between the pre-treatment and post-treatment groups include glutamine,taurine,lysine and other 9 metabolites;The potential markers between patients with good effect(PR+MR)after 2 cycles of bevacizumab and patients with poor efficacy(SD)are taurine,glutamate,leucine,trimethylamine,urea and so on,furthermore,the concentration of glutamine,leucine,and urea metabolites in the former is closer to the healthy group,suggesting that metabolic disorders in patients treated with bevacizumab have been decreased;the potential markers between patients with good effect after 4 cycles of bevacizumab and patients with poor efficacy are methyl-histidine,arginine,alanine,propylene glycol,glutamate,lysine,and the like.Metabolomics pathway analysis showed that top alteredpathways between the healthy and non-small-cell lung cancer patients group mainly include the metabolism of some energy substances,such as the TCA cycle,ketone body synthesis and decomposition,etc.;amino acid-related metabolism,such as alanine,aspartate and glutamate metabolism,protein synthesis-related metabolic pathways such as synthesis of aminoacyl-tRNA,after treatment,the most relevant alter pathways were some amino-realted pathways,including Alanine,aspartate and glutamate metabolism,Taurine and hypotaurine metabolism,Arginine and proline metabolism,Aminoacyl-tRNA biosynthesis.By training the linear SVM model with pre-treatment plasma data of known therapeutic effects,we obtained a classifier that can be used to predict the efficacy of bevacizumab in combination with chemotherapy.The accuracy rate in the training set was 84.62%(AUC=0.881).Accuracy of accuracy is 75%(AUC=0.75),which shows the linear SVM model was feasible.Conclusion:In summary,this study analyzed clinical plasma before and after chemotherapy and bevacizumab treatment in patients with NSCLC though metabolomic methods,identified some potential biomarkers for the diagnosis of non-small cell lung cancer,and assessed the treated efficacy of bevacizumab treatment.The differences in metabolic pathways were analyzed.The LSVM classifier can be used to predict the efficacy of bevacizumab was trained and tested.Thisresearchcan provide a new way of predicting efficacy for the clinical bevacizumabtreatment.