A novel RNA sequencing-based risk score model to predict papillary thyroid carcinoma recurrence

Abstract

The papillary thyroid carcinoma (PTC) usually shows an excellent prognosis. But some patients suffer recurrence after treatment. Recent progress in RNA sequencing (RNA-Seq) allows us to explore whole-transcriptomic gene expression profiles to develop RNA-seq based predictive model for stratifying the risk of recurrence of PTC. RNA-seq and clinical data from The Cancer Genome Atlas thyroid carcinoma cohort were divided chronologically into a training cohort (before 2011, n\u2009=\u2009240) and a validation cohort (after 2011, n\u2009=\u2009239). A risk score model was developed in training cohort using univariate Cox analysis followed by stepwise multivariate Cox analysis, and assessed in the validation cohort. Univariate and multivariate analyses identified five independent predictive genes (TOP2A, RP11-180M15.7, RP11-635N19.1, PROSER3, and TMEM139) significantly (p\u2009<\u20090.05) associated with recurrence-free survival of PTC. The proposed risk score model defined by these five genes was able to divide patients into high-risk and low-risk groups with significantly different recurrence risk in both training cohort [hazard ratio (HR)\u20096.62, 95% confidence interval (CI)\u20093.16–13.86] and validation cohort [HR\u20093.40, 95% CI1.29–8.94). Furthermore, the model is independent of clinicopathologic factors and demonstrated better predictive performance than other clinical covariates in PTC patients with no distant metastasis. Our results indicate that TOP2A, RP11-180M15.7, RP11-635N19.1, PROSER3, and TMEM139 are potential independent predictive genes for the recurrence of PTC and the proposed RNA sequencing-based model has significant predictive value and may improve recurrence risk stratification for early-stage PTC patients.