Molecular Subtyping and Prognostic Prediction in Pancreatic Cancer Based on Mitophagy-Related Genes

Sep 22, 2026

Upload by Finn

Pancreatic cancer shows poor prognosis, motivating development of mitophagy-centered biomarkers. This study identified differentially expressed genes between pancreatic cancer and normal tissues from TCGA and GTEx, then intersected them with mitophagy-related genes curated from Reactome, GO, and KEGG to obtain DeMRGs. Consensus clustering defined molecular subtypes, and univariate Cox plus LASSO regression built a prognostic model and nomogram. Validation using TCGA and independent GEO cohorts assessed survival prediction and calibration, while immune infiltration and drug sensitivity were explored.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages16

Reading Time40 min

AccessFree to download

Abstract

Background

Methods

Study cohorts and data sources

Results

Conclusions

Introduction

Materials and Methods

Ivyspring International Publisher InternationalJournal of Medical Sciences 2026; 23(2): 620-635. doi: 10.7150/ijms.121350 Research Paper Molecular Subtyping and Prognostic Prediction in Pancreatic Cancer Based on Mitophagy-Related Genes Yunlong Cai1, Taohua Yue1, Yongchen Ma1, Guanyi Liu1, Jixin Zhang2,􀀍, Long Rong1,􀀍 1. Endoscopy Center, Peking University First Hospital, Beijing 100034, China. 2. Pathology Department, Peking University First Hospital, Beijing 100034, China. 􀀍 Corresponding authors: Long Rong, E-mail: [ronglong_8@163.com](ronglong_8@163.com), Tel: 010-83572437. Jixin Zhang, E-mail: [187297127@qq.com](187297127@qq.com). © The author(s). This is an open access article distributed under the terms of the Creative Commons Attribution License ([https://creativecommons.org/licenses/by/4.0/](https://creativecommons.org/licenses/by/4.0/)). See [https://ivyspring.com/terms](https://ivyspring.com/terms) for full terms and conditions. Received: 2025.07.10; Accepted: 2026.01.05; Published: 2026.01.14 Abstract Background: Pancreatic cancer (PaC) is characterized by poor prognosis. This study aimed to identify mitophagy-related clusters and develop a prognostic model for PaC. Methods: Differentially expressed genes (DEGs) between PaC and normal tissues were identified from the TCGA and GTEx cohorts. Mitophagy-related genes (MRGs) were sourced from Reactome, GO, and KEGG databases. The intersection of DEGs and MRGs identified differentially expressed MRGs (DeMRGs). Consensus clustering identified PaC subtypes based on DeMRG expression. Univariate Cox analysis was used to find prognosis-related genes, and LASSO regression analysis was employed to develop the prognostic model. A nomogram was constructed to predict survival probabilities. Results: A total of 7,240 DEGs were identified between PaC tissues and normal controls. From these, 12 DeMRGs were identified, and consensus clustering revealed three distinct molecular clusters. A prognostic model based on six significant genes (PAPPA, NBPF12, CXCL11, CKLF-CMTM1, CCDC6, AHNAK) was developed using LASSO regression analysis. This model demonstrated good predictive performance for overall survival in the TCGA cohort, with AUC values of 0.78, 0.74, and 0.82 for 1-, 2-, and 3-year survival in the training set, and 0.73, 0.82, and 0.73 in the validation set. External validation in independent GEO cohorts demonstrated moderate predictive performance. The nomogram demonstrated good calibration and accuracy in predicting survival. Significant correlations were found between the risk model and immune cell infiltration. High-risk patients showed higher sensitivity todasatinib and staurosporine. Conclusions: The study identified mitophagy-related molecular clusters and developed a prognostic model for PaC. This model may help predict overall survival and guide personalized treatment strategies for PaC patients. Keywords: pancreatic cancer, mitophagy, prognosis, gene expression analysis Introduction Pancreatic cancer (PaC) is a highly lethal malignancy originating in the tissues of the pancreas, characterized by a poor prognosis, with a 5-year relative survival rate of 12% for diagnoses made between 2012 and 2018 in the United States [1] . Despite being the sixth leading cause of cancer-related deaths worldwide[2], PaC often evades early detection due to the absence of distinctive clinical symptoms and its aggressive nature [3] . The current therapeutic approaches, primarily systemic chemotherapy and surgical resection, are limited and largely ineffective, contributing to high mortality rates [4] . These challenges underscore the critical need for novel prognostic markers and tailored treatment strategies to improve clinical outcomes for PaC patients. Mitophagy, a selective form of autophagy targeting mitochondria for degradation, operates via ubiquitin-dependent (e.g., PINK1-PRKN pathway) and ubiquitin-independent (e.g., BNIP3L/NIX, FUNDC1) mechan

FAQs

What was the purpose of the study in pancreatic cancer?

To identify mitophagy-related clusters and develop a prognostic model to predict overall survival in pancreatic cancer.

How were mitophagy-related genes and DeMRGs defined?

Mitophagy-related genes were collected from Reactome, GO, and KEGG, while DeMRGs were identified by intersecting these with differentially expressed genes between pancreatic cancer and normal tissues from TCGA and GTEx.

Which modeling and validation strategy was used to build the prognostic tool?

Univariate Cox analysis identified prognosis-related genes, LASSO regression constructed the prognostic model, and a nomogram was used for survival prediction, with performance evaluated in training and validation sets including independent GEO cohorts.

Sep 22, 2026

Upload by Finn