An Exploratory Study On Screening Of Spcific Peptides In Patients With NSCLC And EGFR-TKI Efficacy Related Peptides By Serum Mass Spectrum

Background and objective：1.The improved survival of patients with lungcancer depends on early diagnosis of lung cancer. However, the traditional diagnostictechniques have several limitations. Mass spectrometry (MS) has been applied as acore technology for cancer diagnosis in preliminary proteomic studies. This paperanalyzed differences in the serum peptide levels of patients with non-small-cell lungcancer (NSCLC) and healthy individuals using matrix-assisted laserdesorption/ionization (MALDI)-time-of-flight (TOF)-MS. A NSCLC serumclassification model was then established.2. EGFR gene mutations are the importantmeasurements of advanced NSCLC individualized treatments, for their curative effectprediction of epidermal growth factor receptor-tyrosine kinase inhibitor(EGFR-TKI).Unfortunately, a considerable amount of patients cannot detect the EGFR genemutations due to the limitation of all kinds of objective conditions in the real clinicalwork. We studied the mass spectrometry analysis of serum specimens of theadvanced NSCLC patients treated with EGFR–TKI, using MALDI-TOF-MSexploratively,in order to find the predictable protein or polypeptide as the supplementor complement of EGFR gene detection.Methods：1. One hundred and thirty three cases of patients with NSCLC serumspecimens and132cases of healthy human serum specimens were randomly dividedinto two groups in accordance with the ratio of three to one without age and genderdifferences. The training group was used to establish the classification model, thisgroup included serum samples from100NSCLC cases and100healthy individuals.The test group for validating the proposed model was composed of the remainingserum samples from33NSCLC cases and32healthy individuals. Peptides wereextracted from the samples using magnetic bead-based immobilized metal ion affinity,and their mass spectra were obtained using an automated MALDI-TOF-MS system.The MS data from the training group was analyzed using the ClinproToolTM softwareto identify the individual peptide fragments and establish thclassification model. Thesensitivity and specificity of the model were verified by blind testing with the testgroup.2. Serum specimens were obtained from the advanced NSCLC patients beforethe treatment of EGFR–TKI, the effects were evaluated by imaging4weeks after divided into control group and disease progression group according to the curativeeffects.Peptides were extracted from the samples using magnetic bead-basedimmobilized metal ion affinity, and their mass spectra were obtained using anautomated MALDI-TOF-MS system. Then we found the different peptides betweencontrol group and disease progression group and established the classification model.The sensitivity and specificity of the model were verified by blind testing. thecorrelation between these modeling peptide and EGFR gene expression was analyzedwith rank and inspection.Results：1.Among the131different peptide peaks, ranging from m/z800Da to10,000Da,14peaks were significantly different in the NSCLC samples of thetraining group, as compared with the controls (P<0.000,001; AUC≥0.9); theseincluded2higher peaks and12lower peaks. The classification model was established,and the test group was verified for only3peptide peaks (7,478.59,2,271.44, and4,468.38Da), which were selected by the statistical software. Blind testing revealedthat the proposed method had100%sensitivity,96.9%specificity, and98.5%accuracy.2. A total of103patients who underwent EGFR-TKI treatment of NSCLCpatients was included in the statistics,they were divided into control group (51patients,CR+PR+SD) and disease progression group (52patients,PD) according to the curativeeffects. And more,the specimens were randomly divided into two groups with theratio of3:1. The training group was used to establish the classification model,thisgroup included serum samples from30control group cases and30progression groupcases. The test group for validating the proposed model was composed of theremaining serum samples from21control group cases and22progression group cases.There were125different peptide peaks, ranging from m/z800Da to10,000Da,in30samples of control group and30samples of disease progression group.8peaks weresignificantly higher in the control group, as compared with the disease progressiongroup (P<0.003,AUC≥0.8). The classification model was established according to6peptide peaks (2,660.79、3,883.91、3,891.06、4,644.26、7,776.19、9,307.12Da), whichwere selected by the statistical software. Blind testing revealed that the proposedmethod had76.2%(16/21) sensitivity,81.8%(18/22) specificity, and79.1%(34/43)accuracy.3peptide peaks (m/z2,660.79、4,644.26、9,307.12)were related to EGFRgene mutations.Conclusion：1.The study showed that the serum peptide levels were significantlydifferent between NSCLC patients and healthy individuals. A serum peptide-basedclassification of NSCLC patients was established using an automatedMALDI-TOF-MS system. This method demonstrated high sensitivity and specificityin a small-scale test. Future studies should test the proposed model through mass validation. The model could be compared or combined with traditional diagnosticmethods to establish novel techniques for the early diagnosis of patients with NSCLC.2. Application of MALDI-TOF-MS technique can select a group of serum peptides,these serum peptides expression is positively related to the efficacy of EGFR-TKI,and3peptides expressions (m/z2,660.79、4,644.26、9,307.12) are statistically related toEGFR gene state.They have the potential to be markers for predicting EGFR-TKIcurative effect.