Identification of the long non-coding RNA POU3F3 in plasma as a novel biomarker for diagnosis of esophageal squamous cell carcinoma

Abstract

Background: Recent studies have demonstrated that long non-coding RNAs (lncRNAs) were present in the blood of cancer patients and have shown great potential as powerful and non-invasive tumor markers. However, little is known about the value of lncRNAs in the diagnosis of esophageal squamous cell carcinoma (ESCC). We hypothesized that ESCC-related lncRNAs might be released into the circulation during tumor initiation and could be utilized to detect and monitor ESCC.

Methods: Ten lncRNAs (HOTAIR, AFAP1-AS1, POU3F3, HNF1A-AS1, 91H, PlncRNA1, SPRY4-IT1, ENST00000435885.1, XLOC_013104 and ENST00000547963.1) which previously found to be differently expressed in esophageal cancer were selected as candidate targets for subsequent circulating lncRNA assay. A four-stage exploratory study was conducted to test the hypothesis: (1) optimization of detected method to accurately and reproducibly measure ESCC-related lncRNAs in plasma and serum; (2) evaluation of the stability of circulating lncRNAs in human plasma or serum; (3) exploration the origin of ESCC-related lncRNAs in vitro and in vivo; (4) evaluation the diagnostic power of circulating lncRNAs for ESCC.

Results: ESCC-related lncRNAs were detectable and stable in plasma of cancer patients, and derived largely from ESCC tumor cells. Furthermore, plasma levels of POU3F3, HNF1A-AS1 and SPRY4-IT1 were significantly higher in ESCC patients compared with normal controls. By receiver operating characteristic curve (ROC) analysis, among the three lncRNAs investigated, plasma POU3F3 provided the highest diagnostic performance for detection of ESCC (the area under the ROC curve (AUC), 0.842; p < 0.001; sensitivity, 72.8%; specificity, 89.4%). Moreover, use of POU3F3 and SCCA in combination could provide a more effective diagnosis performance (AUC, 0.926, p < 0.001, sensitivity, 85.7%; specificity, 81.4%). Most importantly, this combination was effective to detect ESCC at an early stage (80.8%).

Conclusions: Plasma POU3F3 could serve as a potential biomarker for diagnosis of ESCC, and the combination of POU3F3 and SCCA was more efficient for ESCC detection, in particular for early tumor screening.

Figure 1

Validation of ESCC-related lncRNAs expression in ESCC tissues. △Ct method was used to calculate lncRNA expression, which was normalized to GAPDH, and smaller ΔCt value indicated higher expression. Horizontal lines inside the box plots represent the median, boxes represent the interquartile range, and error bars represent 97.5th and 2.5th percentiles. *P < 0.05.

Figure 2

ESCC-related lncRNAs were detectable in plasma of tumor patients and normal control. (A-C) Plasma levels of POU3F3 (A), HNF1A-AS1 (B), and SPRY4-IT1 (C) were measured in 24 clinical samples and 24 cancer free controls. Horizontal bars indicate median and interquartile range. Circulating lncRNAs expressions were calculated using △Ct method. Statistical differences were analyzed using Mann–Whitney test. *p < 0.001.

Figure 3

Stability of plasma ESCC-related lncRNAs. POU3F3, HNF1A-AS1, and SPRY4-IT1 shown no significant degradation when plasma was treated with prolonged room temperature incubation time (A), or multiple freeze-thaw cycles (B), or low (pH = 1) or high (pH = 13) pH solution (C), or RNase A digestion (D). Data presented as raw Ct value, and each bars represented the mean (SD) (n = 3).

Figure 4

Correlation of ESCC-related lncRNAs expression between plasma and serum. Left panel, linear correlation plot of POU3F3 (A), HNF1A-AS1 (B) and SPRY4-IT1 (C) level from plasma (y-axis) versus from serum (x-axis). There was a high correlation comparing the indicated lncRNAs levels between plasma and serum. Spearman’s rank analysis was used to identify the correlation of lncRNAs levels between plasma and serum. Right panel, Bland-Altman plot of the difference between plasma and serum ESCC-related lncRNAs level (y-axis) versus their average (x-axis). Horizontal solid lines in the middle represent the mean difference. Upper and lower solid lines represent the limits of agreement (95% confidence intervals).

Figure 5

Origin of circulating lncRNAs. (A) Quantitative real time polymerase chain reaction (qPCR) was used to measure ESCC-related lncRNAs expression in five esophageal cells. GAPDH was used as a normalization control. (B) ESCC-related lncRNAs could be secreted into the cell culture medium. Data presented as relative lncRNAs fold change. (C) ESCC-related lncRNAs could enter into the circulation of xenograft-bearing mice. (D) Spearman’s rank correlation scatter plot of ESCC-related lncRNAs levels in tumor samples and plasma. Data were presented as △Ct values normalized to GAPDH. (E) ESCC-related lncRNAs expressions were significantly declined in post-operative samples compared with that in pre-operative samples. (F) Effect of delayed blood centrifugation on plasma ESCC-related lncRNAs expressions. At room temperature, ESCC-related lncRNAs levels of unfiltered plasma was slightly increased from 0 h to 6 h, but then significantly decreased at 24 h when compared to 6 h. *p < 0.05.

Figure 6

Evaluation of plasma lncRNAs for detection of ESCC. Receiver operating characteristics (ROC) curves were drawn with the data of plasma lncRNAs from 147 patients with ESCC and 123 healthy controls. (A-D) ROC-AUC for detecting ESCC from healthy controls (POU3F3, 0.842, p < 0.001; HNF1A-AS1, 0.781, p < 0.001; SPRY4-IT1, 0.800, p < 0.001; SCCA, 0.784, p < 0.001).

Figure 7

ROC corves to compare the diagnostic performance of SCCA, POU3F3 , and a combination of SCCA and POU3F3 , to discriminate ESCC from normal controls. The combination of SCCA and POU3F3 provided a more powerful differential diagnosis compared with SCCA and POU3F3 alone (SCCA+ POU3F3, 0.926, p < 0.001; POU3F3, 0.842, p < 0.001; SCCA, 0.784, p < 0.001).