Abstract
Abstract
Background
A large number of patients with stage II/III colorectal cancer (CRC) have a high recurrence rate after radical resection. We aimed to develop a novel tool to stratify patients with different recurrence-risk for optimizing decision-making in post-operative surveillance and therapeutic regimens.
Methods
We retrospectively enrolled four independent cohorts from the Gene Expression Omnibus and 66 CRC tissues from our hospital. The initial signature discovery was conducted in GSE143985 (n = 91). This was followed by independent validation of this signature in GSE17536 (n = 111), GSE29621 (n = 40), and GSE92921 (n = 59). Further experimental validation using qRT-PCR assays (n = 66) was performed to ensure the robustness and clinical feasible of this signature.
Results
We developed a novel recurrence-related signature consisting of six genes. This signature was validated to be significantly associated with dismal recurrence-free survival in five cohorts GSE143985 (HR: 4.296 [2.612–7.065], P < 0.0001), GSE17536 (HR: 2.354 [1.662–3.334], P < 0.0001), GSE29621 (HR: 3.934 [1.622–9.539], P = 0.0024), GSE92921 (HR: 7.080 [2.011–24.924], P = 0.0023), and qPCR assays (HR: 3.654 [2.217–6.020], P < 0.0001). This signature was also proven to be an independent recurrent factor. More importantly, this signature displayed excellent discrimination and calibration in predicting the recurrence-risk at 1–5 years, with most AUCs were above 0.9, average C-index for the five cohorts was 0.8795, and near-perfect calibration.
Conclusions
We discovered and experimental validated a novel gene signature with stable and powerful performance for identifying patients at high recurrence-risk in stage II/III CRC.
Funder
National Natural Science Foundation of China
Henan Province Young and Middle‐Aged Health Science and Technology Innovation Talent Project
Henan Provincial Health Commission Joint Youth Project
Publisher
Springer Science and Business Media LLC
Subject
Cancer Research,Genetics,Oncology
Cited by
29 articles.
订阅此论文施引文献
订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献