A signature of saliva-derived exosomal small RNAs as predicting biomarker for esophageal carcinoma: a multicenter prospective study

Abstract

Background: The tRNA-derived small RNAs (tsRNAs) are produced in a nuclease-dependent manner in responses to variety of stresses that are common in cancers. We focus on a cancer-enriched tsRNA signature to develop a salivary exosome-based non-invasive biomarker for human esophageal squamous cell carcinoma (ESCC).

Methods: Cancer-enriched small RNAs were identified by RNA sequencing of salivary exosomes obtained from ESCC patients (n = 3) and healthy controls (n = 3) in a pilot study and further validated in discovery cohort (n = 66). A multicenter prospective observational study was conducted in two ESCC high-incidence regions (n = 320 and 200, respectively) using the newly developed biomarker signature.

Results: The tsRNA (tRNA-GlyGCC-5) and a previously undocumented small RNA were specifically enriched in salivary exosomes of ESCC patients, ESCC tissues and ESCC cells. The bi-signature composed of these small RNAs was able to discriminate ESCC patients from the controls with high sensitivity (90.50%) and specificity (94.20%). Based on the bi-signature Risk Score for Prognosis (RSP), patients with high-RSP have both shorter overall survival (OS) (HR 4.95, 95%CI 2.90-8.46) and progression-free survival (PFS) (HR 3.69, 95%CI 2.24-6.10) than those with low-RSP. In addition, adjuvant therapy improved OS (HR 0.47, 95%CI 0.29-0.77) and PFS (HR 0.36, 95%CI 0.21-0.62) only for patients with high but not low RSP. These findings are consistent in both training and validation cohort.

Conclusions: The tsRNA-based signature not only has the potential for diagnosis and prognosis but also may serve as a pre-operative biomarker to select patients who would benefit from adjuvant therapy.

Trial registration: A prospective study of diagnosis biomarkers of esophageal squamous cell carcinoma, ChiCTR2000031507 . Registered 3 April 2016 - Retrospectively registered.

Keywords: Esophageal carcinoma; Liquid-biopsy signature; Pre-operative biomarker of adjuvant therapy; Sequencing of salivary extracellular vesicles; Transfer RNA-derived small RNA; tRNA-derived fragments.

Fig. 1

Flow diagrams showing the design of the pilot cohort, discovery cohort and the two patient cohorts. A The pilot cohort. B The discovery cohort. C The CHSUMC training cohort. D The ATH validation cohort

Fig. 2

Identification of cancer-enriched sesncRNAs in salivary exosomes of ESCC patients. A The heatmap showing the top five differently expressed small RNAs by small RNA-seq of salivary exosomes. B Confirmation of the differentially expressed salivary RNAs. C Effect of sesncRNAs on cell migration and invasion. All experiments were performed in biological triplicate. D-E The box and scatter plots of tRNA-GlyGCC-5 (left) and sRESE (right) in the CHSUMC cohort (D) and the ATH cohort (E). F The results of ROC analysis of sesncRNAs in the CHSUMC cohort. SEM (*p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by unpaired t-test)

Fig. 3

Performance of sesncRNAs for prognostication in CHSUMC and ATH cohorts. A-B and E-F Kaplan-Meier analysis shows that the OS and PFS were significantly longer in patients with low expression of tRNA-GlyGCC-5 (A and E) or sRESE (B and F) than those with high expression. C and G The sesncRNAs-based Risk Score for Prognosis (RSP) of each patient. Kaplan-Meier analysis shows that patients with low-RSP have longer OS (C) and PFS (G) than those with high-RSP in the CHSUMC cohort. D and H OS (D) and PFS (H) were significantly longer in patients with low RSP than those with high RSP in the ATH cohort. The p-values were calculated using the unadjusted log-rank test and hazard ratios (HR) using univariate Cox regression. CI, confidence interval

Fig. 4

Nomogram to predict the probability of survival of ESCC patients using the bi-sesncRNAs RSP. A ESCC survival nomogram. The calibration curve for predicting OS at 3 years in the CHSUMC (B) and ATH (C) cohort. Nomogram-predicted probability of survival is plotted on the x-axis; actual overall survival is plotted on the y-axis

Fig. 5

RSP for prognostication of survival and treatment prediction for postoperative therapy. A and B The benefit of postoperative therapy in patients classified as high-RSP in CHSUMC cohort. C and D The benefit of postoperative therapy in patients with low-RSP in CHSUMC cohort. E and F The benefit of postoperative therapy for patients with high-RSP in ATH cohort. G and H The benefit of postoperative therapy for patients with low-RSP in CHSUMC cohort. The p-values were calculated using the unadjusted log-rank test, and hazard ratios (HR) using a univariate Cox regression analysis. CI, confidence interval