Establishment and validation of a prognostic model for resectable lung squamous cell carcinoma patients based on miRNA and clinical features
Objective To investigate the prognostic performance of miRNA combined with clinical features in patients with resectable lung squamous cell carcinoma(LUSC).Methods The miRNA sequencing data of 310 patients with stageⅠ-Ⅲ resectable LUSC were downloaded from the TCGA database to establish a training cohort.Tissue samples from 134 patients with stageⅠ-ⅢLUSC in our hospital in the past 5 years were retrospectively collected and miRNA sequencing was performed to establish a validation cohort.All analysis and figure plot were performed by R software or Graphpad PRISM 5.0.Results Compared with normal tissues,patients with stage Ⅰ toⅢoperable lung squamous cell carcinoma showed a large number of miRNAs with high or low expression,in which the number of significantly high expression miRNAs(45)was much more than that of significantly low expression miRNAs(6).Enrichment analysis revealed a number of tumor-related functions or pathways,including PI3K-Akt,MAPK,Ras pathway,RHO GTPase cycle,and nervous system functions.Univariate Cox regression analysis showed that tumor stage and location were significant prognostic factors.Multi-factor analysis showed that site was an independent factor affecting prognosis.Univariate Cox regression analysis of mirnas found that 7 mirnas were prognostic factors.Multivariate Cox regression analysis showed that miR-374a,miR-378c,miR-4638,miR-4746,miR-656,miR-99b and location were independent factors affecting prognosis.These factors were used to establish a Cox regression model for individual prognostic prediction.Subsequent internal and external validation showed that the prediction performance of the model from 1 to 5 years was good.The maximum AUC of the internal validation came from the prediction of the third year,and the maximum AUC of the external validation came from the prediction of the fifth year,and the AUC of the internal and external validation reached 0.77 and 0.75 respectively.Conclusion The individual prognosis prediction model for LUSC was successfully established based on miRNA and clinicopathological factors,which may be helpful for prognosis prediction and treatment strategy selection.
NETL
NSTL
万方数据
