Development and validation of a cancer stem cell-related signature for prognostic prediction in pancreatic ductal adenocarcinoma

Abstract

Background Cancer stem cells (CSCs) are crucial to the malignant behaviour and poor prognosis of pancreatic ductal adenocarcinoma (PDAC). In recent years, CSC biology has been widely studied, but practical prognostic signatures based on CSC-related genes have not been established or reported in PDAC. Methods A signature was developed and validated in seven independent PDAC datasets. The MTAB-6134 cohort was used as the training set, while one local Chinese cohort and five other public cohorts were used for external validation. CSC-related genes with credible prognostic roles were selected to form the signature, and their predictive performance was evaluated by Kaplan–Meier survival, receiver operating characteristic (ROC), and calibration curves. Correlation analysis was employed to clarify the potential biological characteristics of the gene signature. Results A robust signature comprising DCBLD2, GSDMD, PMAIP1, and PLOD2 was developed. It classified patients into high-risk and low-risk groups. High-risk patients had significantly shorter overall survival (OS) and disease-free survival (DFS) than low-risk patients. Calibration curves and Cox regression analysis demonstrated powerful predictive performance. ROC curves showed the better survival prediction by this model than other models. Functional analysis revealed a positive association between risk score and CSC markers. These results had cross-dataset compatibility. Impact This signature could help further improve the current TNM staging system and provide data for the development of novel personalized therapeutic strategies in the future.

Transcriptomic Profiling Identifies DCBLD2 as a Diagnostic and Prognostic Biomarker in Pancreatic Ductal Adenocarcinoma

“…These findings highlighted the clinical utility of DCBLD2 in liquid biopsy. Researchers recently reported two prognostic gene signatures constructed using DCBLD2 in PDAC (Raman et al, 2018;Feng et al, 2020). The prognostic accuracy of DCBLD2 was certainly inferior to that of gene models, because accumulating evidence had demonstrated that multi-gene signatures achieved higher prognostic accuracy compared with a single gene (Beane et al, 2009;Sim et al, 2017;Schmidt et al, 2018;Huang et al, 2020;Ahluwalia et al, 2021).…”

“…Apart from OS, we found that high expression of DCBLD2 was also indicative of short DFS. We previously reported that DCBLD2 was overexpressed in pancreatic cancer stem cells ( Feng et al, 2020 ), a small population of cancer cells with an indispensable role for tumor metastasis and recurrence ( Hermann et al, 2007 ; Simeone, 2008 ). These results suggested that DCBLD2 was potentially involved in the regulation of cell stemness and might serve as a promising therapeutic target for PDAC.…”

“…However, we found that the diagnostic accuracy of DCBLD2 was close to or no less than that of gene models. The diagnostic AUC values of DCBLD2 were 0.800, 0.874, 0.984, and 0.818 while those of a four gene signature containing DCBLD2 ( Feng et al, 2020 ) were 0.806, 0.881, 0.997, and 0.724 in GSE32676, GSE60979, GSE62165, and GSE71729 cohorts, respectively ( Supplementary Figure 3 ). High diagnostic accuracy of DCBLD2 might facilitate novel development of a single-marker qPCR approach for PDAC detection with a reduced cost and widespread applicability.…”

“…In gastric cancer, DCBLD2 is downregulated by epigenetic modification, and it exhibits a suppressive role in cancer cell proliferation and invasion ( Kim et al, 2008 ). In the context of PDAC, the elevated expression of DCBLD2 is correlated with poor clinical outcome ( Raman et al, 2018 ; Feng et al, 2020 ). However, the prognostic value of DCBLD2 was evaluated solely in microarray data ( Feng et al, 2020 ), which does not capture mRNA expression as accurately as RNA-Sequencing ( Zhao et al, 2014 ).…”

“…In the context of PDAC, the elevated expression of DCBLD2 is correlated with poor clinical outcome ( Raman et al, 2018 ; Feng et al, 2020 ). However, the prognostic value of DCBLD2 was evaluated solely in microarray data ( Feng et al, 2020 ), which does not capture mRNA expression as accurately as RNA-Sequencing ( Zhao et al, 2014 ). Compared with clinical predictors such as age and histological grade, the superiority of DCBLD2 expression for survival prediction is largely unclarified.…”