Biomarkers Screening For Treatment And Prognosis Of Pancreatic Cancer Based On PDX Model

Pancreatic cancer is highly malignant,its incidence and death increased rapidly,5-year survival rate of about 5%.Pancreatic cancer is highly heterogeneous,and different patient takes different mutation signature and biological behavior.Each treatment approach including surgery can only benefit a small part of the patients.Same disease using the same program in the treatment of pancreatic cancer is completely impractical.So it is very important to screen out biomarkers that predict the efficacy of pancreatic cancer surgery and drug therapy.Patient Derived Xenografts(PDX)is an animal model that retains the drug sensitivity and pathological features and genetic characteristics of the primary tumors from patients,which can truly reflect the heterogeneity of the patient's tumors.So PDX is a good platform for biomarker screening and therapeutic studies.In this study,we first established a large pancreatic cancer PDX model bank and collected the features of PDX modeling.Then the features of PDX modeling were integrated with the patient's clinical data and omic data to screening out the biomarkers of the prognosis of pancreatic cancer.We conducted drug screening in the same PDX model to verify the ability of the biomarkers to predict the sensitivity of chemotherapy.Finally,this study explored the establishment of humanized pancreatic cancer PDX model and the expression of PD-1 /PD-L1 in pancreatic cancer and its influencing factors were evaluated.Also we conducted biomarkers screening for PD-1/PD-L1 antibodies treatment in pancreatic cancer.The following are major findings.1.In this study,71 cases of pancreatic cancer were modeling for PDX and 36 PDXs were established as PDX bank covering all staging of patients.The stability of th pathological and genetic characteristics of xenografts was identified.Through the correlation analysis of clinicopathological characteristics,it was found that successful PDX modeling was associated with the malignant biological behavior of pancreatic cancer,especially metastatic potential.2.In this study,biomarkers were selected by the integration of t the features of PDX modeling,the clinical information and the results of exon sequencing and RNA sequencing.We can use these biomarkers to divide pancreatic cancer into three different subtypes with different treatment and prognosis.By carrying out therapeutic studies in the PDX model,we found that BRCA and MMR gene mutation could predict the sensitivity of gemcitabine in some patients.3.In this study,the expression of PD-1 / PD-L1 and MLH1 / MSH2 expression and clinicopathological features were analyzed in clinical samples.It was found that the expression of MSH2 in pancreatic cancer was negatively correlated with PD-L1 expression.Then we establish two humanized pancreatic cancer PDX model with MSH2 mutation and MMR gene wildtype,respectly.The expression of PD-L1 in PDX with MSH2 mutations in pancreatic cancer was higher than another one and TILs were more abundant.MSH2 can be used as a biomarker to predict the efficacy of PD-1 / PD-L1 antibodies.In summary,the features of PDX modeling reflect the malignant biological behavior,predicting the DFS and OS of pancreatic cancer patients.biomarkers which can predict the DFS and OS were selected by the integration of t the features of PDX modeling,the clinical information and omic data.Combined with drug screening in PDX,we found that BRCA and MMR gene mutation can predict pancreatic cancer chemotherapy sensitivity.Finally,the clinical samples and humanized PDX model were used to explore the possibility of predicting the efficacy of PD-1 / PD-L1 by MSH2 mutation.