一种新的基于A-I RNA编辑预测乳腺癌预后模型的建立和验证

Establishment and validation of a new prognostic model for breast cancer based on A-I RNA editing

目的:开发一种基于腺苷到肌苷的脱氨基(A-to-I RNA编辑,ATIRE)的预后模型用于改善乳腺癌的个体化治疗。方法:首先使用单变量Cox回归分析来获得训练集中与总生存(OS)相关的ATIRE位点,然后进行最小绝对收缩和选择算子(LASSO)回归算法来确定最佳预后ATIRE位点,进行多因素Cox比例风险回归分析建立风险模型,纳入ATIRE风险评分和临床病理学特征变量构建预后列线图,绘制校准曲线并计算一致性指数以评价模型的预测概率与实际的一致性,通过决策曲线分析(DCA)评价该模型的临床收益价值。结果:确定18个预后位点用来构建预后模型,并生成ATIRE风险评分。高风险评分患者的中位生存时间显著缩短,列线图在预测乳腺癌的OS概率方面表现良好。校准曲线表现出优异的一致性,决策曲线显示其具有更高的净收益。结论:分析ATIRE事件在预测乳腺癌生存中的作用,基于AITRE的预后模型可以帮助临床医师更好进行临床决策。

Objective:To develop a prognostic model based on adenosine-to-inosine deamination(A-to-I RNA editing,ATIRE)to improve individualized treatment of breast cancer.Methods:Univariate Cox regression analysis was first used to obtain ATIRE loci associated with overall survival(OS)in the training set,followed by a least absolute shrinkage and selection operator(LASSO)regression algorithm to determine the best prognostic AIRE loci,a multivariate Cox proportional risk regression analysis to build a risk model,incorporating AIRE risk scores and clinico-pathological characteristics variables to construct A prognostic nomogram was constructed,calibration curves were plotted and consistency indices were calculated to evaluate the agreement between the predicted probability of the model and the actual,and the clinical benefit value of the model was evaluated by decision curve analysis(DCA).Results:Eighteen prognostic loci were finally identified for constructing prognostic models and generating ATIRE risk scores.Patients with high risk scores had significantly shorter median survival times,and nomogram performed well in predicting the probability of OS in breast cancer.The calibration curves showed excellent agree-ment and the decision curves showed a higher net benefit.Conclusion:To analyze the role of ATIRE events in predicting breast cancer survival,AITRE-based modal prognostic phenotype can help clinicians to make better clinical decisions.