Research On The Efficacy Prediction Model Constrution Of Neoadjuvant Chemotherapy For Breast Cancer Based On Serum Glycoprotein Glycoforms

Background: Breast cancer is a malignant tumor with high incidence and seriously endangers women’s health.At present,neoadjuvant chemotherapy has been widely used in the comprehensive treatment of breast cancer.However,it’s a difficult question to forecast the efficacy of neoadjuvant chemotherapy in current clinical research.Aberrant glycosylation occurs frequently during tumour genesis and progression,and tumor-associated glycosylation alterations play important roles in both tumor diagnosis and therapy.By using serum as samples,this study utilized lectin microarray technology to analysis the protein glycan structures of serums from healthy female volunteers(HV),breast cancer patients(BC)and breast cancer neoadjuvant chemotherapy patients(NCT).The aim of this study is to screen out differentially expressed glycan chain structure,investigate whether it can be used as detection and prognostic indexs for early efficacy prediction of neoadjuvant chemotherapy in breast cancer to judge whether it can be used as bases to maintain established treatment regimens.Methods: In this study,the protein glycan chain structure of serum samples from 30 HV,37 BC,and 144 NCT patients were analyzed by lectin microarray technology.Thereinto,144 cases NCT patients were divided into chemotherapy effective group(NCTa,92 cases)and chemotherapy ineffective group(NCTb,52 cases)according to Miller-Payne grading system for pathological reactions after neoadjuvant chemotherapy.The differentially expressed glycan chain structures were screened out by chip analysis,which were verified by lectin blotting experiments.Subsequently,the results of lectin microarray analysis from serum of 67 NCT patients after theirs first or second chemotherapy were used as the training set(42 NCTa;25 NCTb).The binary Logisitic stepwise logistic regression analysis method was applied to construct the NCT curative effect prediction Model(Model NCT).The results of lectin microarray analysis from serum of 44 NCT patients after theirs third chemotherapy were used as the validation set(26 NCTa;18 NCTb).Then,the predictive performance of Model NCT was evaluated through Receiver operating characteristic cure(ROC).Finally,serum samples from 20 patients after theirs first or second chemotherapy with unknown clinical efficacy were collected for blind testing to further evaluate the clinical application potential of Model NCT.Result: The results of lectin microarray showed that there were 17 kinds of lectin-recognized glycan chain structures with significant different in HV and BC group.Among them,the Normalized fluorescence signal intensity(NFIs)values of 8 lectins(EEL,LEL,LCA,MAL-I,SBA,PSA,WGA,PWM)that specifically recognize(Glc NAc)N,Fu Cα1-6Glc NAc and Galβ-1,4Glc NAc and so on were significantly up-regulated in BC group.Besides,the NFIs values of 9 lectins(ECA,STL,RCA120,SJA,PNA,PTL-I,PTL-II,MAL-II,AAL)that specifically recognize Galβ1-3Gal NAcα-Ser/Thr,Gal NAcα-1,3Gal,β-Gal and so on were significantly down-regulated in BC group.Following,lectin bloting assays were performed to verify ECA and SJA,the results of which were consistent with that of lectin microarrays.The results of lectin microarrays analysis also showed that there were significant differents in the glycan chain structures recognized by 17 lectins between NCTa and NCTb groups.Therein,the NFIs values of 3 lectins(STL,RCA120,GSL-I)that specifically recognize core(Glc NAc)of N-glycan,β-Gal,αGal NAc and so on were dramatically up-regulated in NCTb group;and the NFIs values of the rest 14 lectins(PHA-E,MAL-II,Jacalin,EEL,PWM,MAL-I,GNA,SBA,PTL-I,BS-I,DSA,PTL-II,BPL,LTL)that specifically recognize high-Mannose,(Gal NAc)n,Fucα1-2Galβ1-4Glc NAc and so on were markedly down-regulated in NCTb group.In the same way,lectin bloting assays were carried out to verify GNA and EEL,the results of which were consistent with that of lectin microarrays analysis,thereby demonstrated the accuracy and reliability of lectin microarrays.The Model NCT constructed by the binary Logisitic stepwise logistic regression analysis method involves three lectins including GNA,PWM and LTL.For single lectin GNA,the obtained parameter is 0.878 for area under the ROC curve(AUC),0.857 for sensitivity and0.760 for specificity;For PWM,the obtained parameter is 0.771 for AUC,0.714 for sensitivity and 0.600 for specificity;For LTL,the obtained parameter is 0.762 for AUC,0.738 for sensitivity and 0.600 for specificity;While for Model NCT,the obtained parameter is 0.920 for AUC,0.857 for sensitivity and 0.880 for specificity.It follows that the predictive abilities of a single lectin are all obviously lower than that of Model NCT.Meanwhile,in validation group,Model NCT also showed excellent ability for efficacy prediction with an AUC of 0.861,a sensitivity of 0.731,and a specificity of 0.778,which can correctly predict 19 of the 26 NCTa cases and 14 of the 18 NCTb cases.In addition,serum samples from 20 NCT patients with unknown clinical efficacy were collected for blind testing.After detected by lectin microarrays in turns,the acquired analysis data were put into Model NCT to caculate,the results of which were compared with the final clinical outcomes.The results indicate that Model NCT could correctly predict 8 of the 9 NCTa cases and 9 of the 11 NCTb cases,with an accuracy rate high to 85%.Conclusion: For breast cancer patients that receive neoadjuvant chemotherapy,Model NCT constructed based on the glycan structures of serum glycoproteins has excellent ability to predict patients’ late-stage clinical efficacy and to evaluate theirs prognosis.The Model NCT construted by us is hopeful to be applied in clinical practice with the advantages to fight for treatment window,save medical cost and adjust personalized treatment options for patients who are predicted to be ineffective in chemotherapy,thereby provides clinical reference values.