1. Academic Validation
  2. Construction of a prognostic model with histone modification-related genes and identification of potential drugs in pancreatic cancer

Construction of a prognostic model with histone modification-related genes and identification of potential drugs in pancreatic cancer

  • Cancer Cell Int. 2021 Jun 5;21(1):291. doi: 10.1186/s12935-021-01928-6.
Yuan Chen  # 1 Ruiyuan Xu  # 1 Rexiati Ruze 1 Jinshou Yang 1 Huanyu Wang 1 Jianlu Song 1 Lei You 1 Chengcheng Wang 2 Yupei Zhao 3
Affiliations

Affiliations

  • 1 Department of General Surgery, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Peking Union Medical College, Beijing, 100023, People's Republic of China.
  • 2 Department of General Surgery, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Peking Union Medical College, Beijing, 100023, People's Republic of China. wangchengcheng3638@163.com.
  • 3 Department of General Surgery, State Key Laboratory of Complex Severe and Rare Diseases, Peking Union Medical College Hospital, Chinese Academy of Medical Sciences, Peking Union Medical College, Beijing, 100023, People's Republic of China. zhao8028@263.net.
  • # Contributed equally.
Abstract

Background: Pancreatic Cancer (PC) is a highly fatal and aggressive disease with its incidence and mortality quite discouraging. An effective prediction model is urgently needed for the accurate assessment of patients' prognosis to assist clinical decision-making.

Methods: Gene expression data and clinicopathological data of the samples were acquired from The Cancer Genome Atlas (TCGA), Genotype-Tissue Expression (GTEx), and Gene Expression Omnibus (GEO) databases. Differential expressed genes (DEGs) analysis, univariate COX regression analysis, least absolute shrinkage and selection operator (LASSO) regression analysis, random forest screening and multivariate COX regression analysis were applied to construct the risk signature. The effectiveness and independence of the model were validated by time-dependent receiver operating characteristic (ROC) curve, Kaplan-Meier (KM) survival analysis and survival point graph in training set, test set, TCGA entire set and GSE57495 set. The validity of the core gene was verified by immunohistochemistry and our own independent cohort. Meanwhile, functional enrichment analysis of DEGs between the high and low risk groups revealed the potential biological pathways. Finally, CMap database and drug sensitivity assay were utilized to identify potential small molecular drugs as the risk model-related treatments for PC patients.

Results: Four histone modification-related genes were identified to establish the risk signature, including CBX8, CENPT, DPY30 and PADI1. The predictive performance of risk signature was validated in training set, test set, TCGA entire set and GSE57495 set, with the areas under ROC curve (AUCs) for 3-year survival were 0.773, 0.729, 0.775 and 0.770 respectively. Furthermore, KM survival analysis, univariate and multivariate COX regression analysis proved it as an independent prognostic factor. Mechanically, functional enrichment analysis showed that the poor prognosis of high-risk population was related to the metabolic disorders caused by inadequate Insulin secretion, which was fueled by neuroendocrine aberration. Lastly, a cluster of small molecule drugs were identified with significant potentiality in treating PC patients.

Conclusions: Based on a histone modification-related gene signature, our model can serve as a reliable prognosis assessment tool and help to optimize the treatment for PC patients. Meanwhile, a cluster of small molecule drugs were also identified with significant potentiality in treating PC patients.

Keywords

CBX8; CENPT; DPY30; Histone modification; PADI1; Pancreatic cancer; Potential drugs; Prognostic model.

Figures
Products