A three-lncRNA signature of pretreatment biopsies predicts pathological response and outcome in esophageal squamous cell carcinoma with neoadjuvant chemoradiotherapy

Background: Current strategies are insufficient to predict pathologically complete response (pCR) for esophageal squamous cell carcinomas (ESCCs) before treatment. Here, we aim to develop a novel long noncoding RNA (lncRNA) signature for pCR and outcome prediction of ESCCs through a multicenter analysis for a Chinese population.

Methods: Differentially expressed lncRNAs (DELs) between pCRs and less than pCR (<pCR) in the pretreated cancer biopsies were identified from 28 cases in Guangzhou cohort and verified from 30 cases in Beijing discovery cohort. Then a prediction model was built through Fisher's linear discriminant analysis (FLDA) of 67 cases in Beijing training cohort. Then an internal cohort and an integrated external cohort (Zhengzhou and Anyang cohorts) were used to validate the predictive accuracy. The prognostic value of this signature was also evaluated.

Results: Twelve DELs were identified from Guangzhou cohort and six lncRNAs were verified. Then, a classifier of three lncRNAs (SCAT1, PRKAG2-AS1, and FLG-AS1) was established and achieved a high accuracy with an area under the receiver operating characteristic curve (AUC) of 0.952 in the training cohort, which was well validated in the internal validation cohort and external cohort with the AUCs of 0.856 and 0.817, respectively. Furthermore, the predictive score was identified as the only independent predictor for pCR. Patients with high discriminant score showed a significantly longer overall and relapse-free survival (P < .05).

Conclusions: We developed the first and applicable three-lncRNA signature of pCR and outcome prediction, which is robust and reproducible in multicenter cohorts for ESCCs with nCRT.