Correlation Of Pancreatic Cancer Subtypes And Clinical And CT Features Based On Metabolomics

CHAPTER ONE Correlation between metabolic subtypes and clinical features and CT features of pancreatic cancerObjective: Based on liquid chromatography-mass spectrometry（LC-MS）metabolomics,our group found that 82 patients with pancreatic ductal adenocarcinoma（PDAC）could be classified into two metabolic subtypes based on their serum metabolic characteristics,and the two subtypes were named Subtype A and Subtype B in this study.The correlation between the subtypes and clinical and CT features was explored by analyzing the differences between the clinical and CT features of Subtype A and Subtype B.Methods: In the early stage of our research group,LC-MS metabolomics technology was used to perform non-targeted metabolomic analysis of serum samples from 82 cases of pancreatic cancer to obtain the original data,and found that 82 patients could be classified into two subtypes based on serum metabolic characteristics using principal component analysis（PCA）,which were named as Subtype A（n=51）and Subtype B（n=31）in this study,univariate analysis was performed on the clinical and CT characteristics of patients with two metabolic subtypes,including gender,age,diabetes,hypertension,recent weight loss,jaundice,serum carbohydrate antigen 19-9（CA19-9）,carcinoembryonic antigen（CEA）,cholesterol,triglycerides,tumor differentiation,tumor size,location,shape,cystic-solid,main pancreatic duct dilatation,common bile duct dilatation,distal pancreatic atrophy,degree of enhancement,lymphatic invasion,vascular invasion,distant metastases,to find out the correlation of metabolic subtypes with clinical and CT features.Based on the clinical and CT characteristics of the two subtypes,multivariate logistic regression was used to explore the independent influencing factors for distinguishing the two subtypes.SPSS 26.0 was used for statistical software,t-test was used to compare measurement data between the two groups,and χ2 test was used to compare count data.Results:（1）The results of univariate analysis showed that the proportion of diabetes mellitus was significantly lower in Subtype A than in Subtype B（P<0.05）,and the proportion of distant metastases was significantly higher in Subtype A than in Subtype B（P<0.05）.The remaining gender,age,hypertension,recent weight loss,jaundice,CA19-9,CEA,cholesterol,triglycerides,degree of differentiation,tumor size,site,shape,cystic solidity,main pancreatic duct dilatation,common bile duct dilatation,distal pancreatic atrophy,degree of enhancement,lymphatic invasion,and vascular invasion were not statistically different（P>0.05）.（2）Multi-factor logistic regression results combining clinical and CT features showed that distant metastasis was an independent influencing factor to distinguish the two subtypes of Subtype A and Subtype B（P<0.05,OR=4.879）,and Subtype A had a greater tendency of distant metastasis compared with Subtype B.Conclusions:（1）The tendency of distant metastasis was higher in Subtype A compared with Subtype B,suggesting that Subtype A is associated with worse tumor biological behavior.（2）Different metabolic subtypes of pancreatic cancer are associated with different distant metastases,suggesting that in-depth research on the microscopic metabolic differences between the two metabolic subtypes is helpful for diagnosing distant metastasis of pancreatic cancer at the metabolic level.CHAPTER TWODetermination of metabolic characteristics of two subtypes and correlation of differential metabolites and distant metastasisObjective: To screen the differential metabolites and differential metabolic pathways of pancreatic cancer Subtype A and Subtype B based on LC-MS metabolomics technology,analyze the heterogeneity of the two metabolic subtypes at the serum metabolic level,and explore the correlation between micro-metabolic features and macroscopic distant metastasis.Methods: Selected 82 patients with pancreatic cancer,which were analyzed by the previous LC-MS metabolomic technique in our group and could be divided into two subtypes,Subtype A（n=51）and Subtype B（n=31）according to their serum metabolic characteristics,and combined univariate and multivariate statistical methods to perform preliminary screening of metabolites differing between the two subtypes.Univariate analysis included t-test and fold change analysis（FC Analysis）,and multivariate statistical analysis included partial least squares discriminant analysis（PLS-DA）and orthogonal partial least squares discriminant analysis（OPLS-DA）to initially screen for differential metabolites with projection variable importance value（VIP）>1 and P<0.05.Clinical confounders were included using binary logistic regression,and differential metabolites influenced by clinical confounders were excluded.The metabolites enriched in each of the two subtypes were ranked according to VIP values,and the differential metabolic pathways between the two subtypes were analyzed with KEGG pathway enrichment to determine the metabolic characteristics of the two subtypes,and the two subtypes were named as such.Combined with the first part,Subtype A type is more prone to distant metastasis than Subtype B type.Spearman correlation analysis was performed between the differential metabolites of the two subtypes and distant metastasis,and then the pancreatic cancer group with metastasis（n=21）and those without the metastatic group（n=61）was the dependent variable,with clinical confounding factors as independent variables and each differential metabolite as independent variables,binary logistic regression analysis was performed separately to find out the relationship between small molecule metabolites and macroscopic pancreatic cancer distant metastasis correlation.Results:（1）In this study,after univariate and multivariate analysis,a total of 51 serum differential metabolites of the two subtypes were initially screened by VIP>1 andP<0.05,and a total of 6 metabolites disturbed by clinical confounding factors were excluded using binary logistic regression,leaving 45 metabolites.（2）Combining the different metabolic pathways derived from the KEGG pathway analysis of the two subtypes and the enrichment of the respective different metabolites of the two subtypes,Subtype A was named as bile acid salt elevated type,which showed a significant enrichment of bile acid salt;Subtype B was named as mixed type,which showed a mixed enrichment of glycerophospholipids,carboxylic acids,fatty acyl groups and amino acid derivatives.In combination with the first part of this study it is clear that the bile acid salt elevated type is more inclined to distant metastasis.（3）The correlation analysis of 45 differential metabolites and distant metastasis found that there were 12 metabolites significantly correlated with distant metastasis,3of which were bile salts,and the correlation value of bile salts（r value）is relatively large,including: taurochenodeoxycholate（P<0.001,r=0.443）,glycochenodeoxycholate（P<0.001,r=0.417）,taurocholate（P<0.001,r=0.385）.The results of binary logistic regression of three bile salts and distant metastasis of pancreatic cancer found that taurochenodeoxycholate（P=0.003;OR=3.20）,glycochenodeoxycholate（P=0.007;OR=3.67）,and taurocholate（P=0.005;OR=2.17）were risk factors for distant metastasis of pancreatic cancer and had a positive effect on distant metastasis of pancreatic cancer.Conclusions:（1）The metabolic characteristics of patients with bile acid salt elevated type and mixed types are highly variable,and the combined first part Subtype A type is more inclined to distant metastasis,suggesting that serum bile acid salts levels of patients may be correlated with distant metastasis status.（2）In this study,taurochenodeoxycholate,glycochenodeoxycholate,and taurocholate were significantly positively correlated with distant metastasis,suggesting that the preoperative detection of elevated levels of these three serum bile salts in patients is helpful in helping clinical diagnosis of distant metastasis in patients,and also establishes a basis for clinical assessment of prognosis and selection of targeted therapeutic agents for patients with different metabolic characteristics.（3）In this study,we found that the difference in bile acid salt metabolism between the two subtypes was only the most significant metabolic difference between the two subtypes.In addition to bile acid salts,there are still many other types of differential metabolites between the two subtypes,partly associated with distant metastasis and partly not,suggesting that there are still other differences in metabolic characteristics between the two subtypes,but the reciprocal pattern of these differential metabolites has not been found in this study,and the influencing factors leading to typing may be more than distant metastasis,which need to be further explored in the future.