Development and validation of a hypoxia-related gene signature to predict overall survival in early-stage lung adenocarcinoma patients
Abstract
Background: Patients with early-stage lung adenocarcinoma(LUAD)exhibit significant heterogeneity in overall survival. The current tumour-node-metastasis staging system is insufficient to provide precise prediction for prognosis.
Methods: We quantified the levels of various hallmarks of cancer and identified hypoxia as the primary risk factor for overall survival in early-stage LUAD. Different bioinformatic and statistical methods were combined to construct a robust hypoxia-related gene signature for prognosis. Furthermore, a decision tree and a nomgram were constructed based on the gene signature and clinicopathological features to improve risk statification and quntify risk assessment for individual patients.
Results: The hypoxia-related gene signature discriminated high-risk patients at an early stage in our investigated cohorts. Survival analyses demonstrated that our gene signature served as an independent risk factor for overall survival. The decision tree identified risk subgroups powerfully. and the nomogram exhibited high accuracy.
Conclusions:Our study might contribute to the optimization of risk stratification for survival and personalized management of early-stage LUAD
Keywords: early-stage lung adenocarcinoma, gene signature, hypoxia, prognosis, therapeutic resistance
Introduction
Lung adenocarcinoma (LUAD) is the most common subtype of non-small cell lung cancer (NSCLC).1 Currently, treatment decisions for individual LUAD patients are based mainly on patient and cancer-specific factors, such as tumour-node-metastasis (TNM) staging and differentiation grade. However, the predictive power and accuracy for prognosis are often insufficient. Thus, reliable predictors that can accurately estimate prognosis would bring tremendous value in guiding the management of LUAD.2 For example, better classification of early-stage LUAD after surgery should be used because several large randomized studies suggested that most patients who were sectioned as pathological stage I (p-stage I) and received adjuvant therapy showed no overall survival benefit among unselected patients.3,4 The 5-year overall survival remains unfavourable in patients with p-stage I, with a rate ranging from 73% in Ia to 58% in Ib.5 Therefore, in addition to traditional strategies, there is an urgent need to seek more accurate predictors for early-stage LUAD to discriminate high-risk subsets that could benefit from systemic treatment.
Hypoxia, or lack of oxygen, is a feature of most solid tumours.6 The hypoxic environment in tumours is a result of an imbalance between decreased oxygen supply and increased oxygen demand, which is widely considered to be associated with resistance to therapies, advanced aggressiveness and poor clinical outcomes.7-9 Although several studies have indicated that intratumoural hypoxia and increased hypoxiainducible factor I-alpha (HIF1A) expression are firmly associated with cancer progreesion and poor survival in lung adenocarcinoma,10-12 no hypoxia-based method is available that can be used to identify high-risk patients in early stages.
In this study, we not only identified hypoxia among the various hallmarks of cancer as a dominant risk factor for overall survival in relatively early-stage (p-stage I and II) LUAD but also combined different methods to screen for robust biomarkers and establish a hypoxia-related gene signature for prognosis. In addition, we validated the prognostic value of the gene signature in four independent cohorts. Finally, an integrated model based on the gene signature and clinicopathological features was developed to improve the predictive power and accuracy.
