Cell-Free RNA Content in Peripheral Blood as Potential Biomarkers for Detecting Circulating Tumor Cells in Non-Small Cell Lung Carcinoma

Circulating tumor cells (CTCs) have been implicated in tumor progression and prognosis. Techniques detecting CTCs in the peripheral blood of patients with non-small cell lung carcinoma (NSCLC) may help to identify individuals likely to benefit from early systemic treatment. However, the detection of CTCs with a single marker is challenging, owing to low specificity and sensitivity and due to the heterogeneity and rareness of CTCs. Herein, the probability of cell-free RNA content in the peripheral blood as a potential biomarker for detecting CTCs in cancer patients was investigated. An immunomagnetic enrichment of real-time reverse-transcription PCR (RT-PCR) technology for analysis of CTCs in NSCLC patients was also developed. The mRNA levels of four candidate genes, cytokeratin 7 (CK7), E74-like factor 3 (ELF3), epidermal growth factor receptor (EGFR), and erythropoietin-producing hepatocellular carcinoma receptor B4 (EphB4) that were significantly elevated in tumor tissues and peripheral blood mononuclear cells (PBMCs) were determined. The expression of CK7 and ELF3 in tumor tissues and EGFR in PBMCs was associated with lymph node metastasis (all p < 0.05). The expression of CK7 in PBMCs was correlated with age and EphB4 in PBMCs correlated with histopathological type, respectively (all p < 0.05). The expression of all four genes in tumor tissues and PBMCs was significantly correlated with the clinical stage (all p < 0.01). Survival analysis showed that the patients with enhanced expression of CK7, ELF3, EGFR, and EphB4 mRNA in PBMCs had poorer disease-free survival (DFS) and overall survival (OS) than those without (all p < 0.0001). The present study showed that this alteration of cell-free RNA content in peripheral blood might have clinical ramifications in the diagnosis and treatment of NSCLC patients.


Introduction
The continuous search for tumor markers has yielded several positive results [1]. Research on biomarkers playing a vital role in the early diagnosis and individualized treatments of the tumors is Real-time reverse-transcription PCR (RT-PCR) analysis of marker gene expression in non-small cell lung carcinoma (NSCLC) tissues and paired peripheral blood mononuclear cell (PBMC) specimens. Total amount of mRNA was normalized to GAPDH and the relative mRNA expressions of the samples (tumor vs. para-cancerous histological normal tissues (PCHNTs), or paired PBMCs vs. positive control) were calculated using 2 −ΔΔCt formula. (A) Tumor tissues vs. PCHNTs from 111 NSCLC patients. The mRNA expression of CK7, ELF3, EGFR and EphB4 in lung cancer tissues were significantly higher than that in PCHNTs (all p < 0.0001); T: Tumor tissues; P: PCHNTs; (B) PBMCs from 111 NSCLC patients vs. positive reference value (PRV). The mRNA expression of CK7, ELF3, EGFR and EphB4 in PBMCs from NSCLC patients were significantly higher than that in healthy controls, respectively (all p < 0.0001); P: PBMC; R: PRV. *: p ˂ 0.0001.

The mRNA Expression of Selected Markers in Paired PBMC Preparations
346 blood samples were taken from the recruited participants including 111 NSCLC patients, 115 benign pulmonary disease patients, and 120 healthy controls. No significant difference was observed in the age and gender among the different groups (p > 0.05 for both). specimens. Total amount of mRNA was normalized to GAPDH and the relative mRNA expressions of the samples (tumor vs. para-cancerous histological normal tissues (PCHNTs), or paired PBMCs vs. positive control) were calculated using 2 −∆∆Ct formula. (A) Tumor tissues vs. PCHNTs from 111 NSCLC patients. The mRNA expression of CK7, ELF3, EGFR and EphB4 in lung cancer tissues were significantly higher than that in PCHNTs (all p < 0.0001); T: Tumor tissues; P: PCHNTs; (B) PBMCs from 111 NSCLC patients vs. positive reference value (PRV). The mRNA expression of CK7, ELF3, EGFR and EphB4 in PBMCs from NSCLC patients were significantly higher than that in healthy controls, respectively (all p < 0.0001); P: PBMC; R: PRV. *: p < 0.0001.

The mRNA Expression of Selected Markers in Paired PBMC Preparations
346 blood samples were taken from the recruited participants including 111 NSCLC patients, 115 benign pulmonary disease patients, and 120 healthy controls. No significant difference was observed in the age and gender among the different groups (p > 0.05 for both).  The expression of the four putative markers identified in paired PBMCs was validated by real-time RT-PCR. We found that these four mRNAs were detectable in the majority of PBMC samples from 111 NSCLC patients. Four markers (CK7, ELF3, EGFR, and EphB4) were significantly elevated in PBMCs (all p < 0.0001) ( Figure 1B). However, no significant difference was observed in the expression of the four genes in PBMCs from 115 benign pulmonary disease patients as compared to 120 healthy controls (p > 0.05). Furthermore, the results indicated that these four mRNA levels in PBMCs was closely correlated with the clinical stage (p = 0.006, 0.003, 9.62 × 10 −5 , and 0.006, respectively) in NSCLC patients ( Table 2). The circulating CK7 mRNA was closely correlated with the patient's age revealed during the first examination (p = 0.038). The circulating EphB4 mRNA was closely correlated with the histopathological type (p = 0.032) while the circulating EGFR mRNA was closely correlated with the lymph node metastasis (p = 0.023) ( Table 2).

Receiver Operating Characteristics (ROC) Curve Analysis of the Four mRNAs as a Marker of CTCs
The areas under the ROC curves were 0.909 (0.853-0.965, 95.0% confidence interval (CI)) for CK7, 0.917 (0.863-0.971, 95.0% CI) for ELF3, 0.939 (0.893-0.986, 95.0% CI) for EGFR, and 0.894 (0.834-0.954, 95.0% CI) for EphB4 ( Figure 2). The cut-off values of CK7, ELF3, EGFR, and EphB4 were defined as a 2-fold increase of the expression compared with the positive control. The sensitivity of each marker was: CK7 (81.8%), ELF3 (83.3%), EGFR (87.8%), EphB4 (78.8%), whereas the specificity was: CK7 (100%), ELF3 (100%), EGFR (100%), and EphB4 (100%). The expression of at least one of these four markers (a combination of the four markers) was considered as CTC positive. Thus, we determined the effect of the combination of the four markers by constructing a ROC curve and fitting a logistic model with parameters for CK7, ELF3, EGFR and, EphB4. The area under the ROC curve was 0.848 (0.779-0.917, 95.0% CI); the sensitivity and specificity were 69.7% and 100%, respectively, when the optimal threshold was defined as a 2-fold increase of expression ( Figure 2). These results indicated that the detection of CTCs using the combination of four markers was not superior to the four markers individually.

Cell-Free RNA Content in Peripheral Blood Is Closely Associated with the Prognosis of NSCLC Patients
Next, we investigated the association of circulating CK7, ELF3, EGFR, and EphB4 mRNA with patients' survival. In both disease-free survival (DFS) and overall survival (OS) situations, the four circulating mRNAs were significantly associated with poor prognosis, short DFS, and low OS, respectively. The average durations of DFS and OS in patients with high level circulating CK7, ELF3, EGFR, and EphB4 mRNAs (tumor PBMCs vs. positive control: >2-fold) were significantly shorter than those of patients with normal or downregulation of the four mRNAs in the PBMCs (Figure 3). The average DFS in patients with high level circulating CK7, ELF3, EGFR, and EphB4 mRNAs was 17.6, 17.9, 18.4, and 17.7 months vs. 24.3, 24.2, 24.0, and 24.1 months in patients with a normal or low level of the four mRNAs, respectively. The average OS in patients with high level circulating CK7, ELF3, EGFR, and EphB4 mRNA was 22.7, 22.9, 23.1, 22.6 months vs. 24.3, 24.2, 24.0, and 24.4 months in patients with a normal or low level of the four mRNAs, respectively ( Table 3). All of the four circulating mRNAs, including CK7, ELF3, EGFR, and EphB4 mRNA, were strongly associated with NSCLC patients' prognosis; however, the association of the cell-free RNA content with patient survival was independent of the tumor stage ( Figure 3). In addition to the four circulating mRNAs, we observed that a number of other previously characterized clinical parameters were associated with patient survival (Table 3), including gender, smoking, and TNM stage. However, no significant correlation was found between a majority of the clinical characteristics such as age at diagnosis, lesion site, histopathological type, tumor size, differentiation, invasive depth, nodal metastasis, and distant metastasis with the patients' prognosis. In summary, these data indicate that in addition to a minority of clinical characteristics that have been shown to affect the prognosis of NSCLC patients, these four circulating mRNAs might serve as a biomarker to predict the DFS and OS of the patients.

Discussion
Metastasis is a multistep event. Some tumor cells acquire metastasis ability independent of the in situ tumor and then migrate or invade into the blood and lymph vessels to act as circulating tumor cells [15]. More than 90% of the tumor-related deaths arise out of the progression of hematogenously disseminated metastasis [16]. CTCs play a vital role in metastasis, and detection of CTCs can contribute towards the evaluation of progression, prognosis, and personal treatment [4,5]. However, because of low quantity and heterogeneity of CTCS, a unified detection method to count and classify CTCs has not yet been described [5,6]. Herein, we collected PBMCs from 111 NSCLC patients to detect the target genes' expression originating from epithelium instead of acquiring and examining the CTCs. The elevated levels of gene expression from epithelial origin indicate the existence of CTCs or circulating tumor nuclei in peripheral blood; these cannot be detected in the normal peripheral blood [17,18]. Lung cancer is a heterogeneous disease with respect to histological and biological characteristics [19]. Owing to the heterogeneity of the tumor cells and the limited reliability of single marker, we selected four genes of epithelial origin, described previously, for the analysis of gene expression [17,18]. These four markers cannot be detected in normal mononuclear cells by optimized experiments and are specific for the detection of circulating epithelial cells [12]. The analysis of CK7, ELF3, EGFR, and EphB4 mRNAs in PBMCs for the detection of CTCs have been reported in various epithelial cancers including NSCLC [11][12][13][14], respectively; however, a combined of these four markers in NSCLC has not yet been reported.
CK7, a star member in lung cancer, has been reported in primary or metastatic lung cancer, cavity liquid, and pleural mesothelioma. The frequency of CK7 upregulation in adenocarcinoma (ADC) was higher than that in squamous cell carcinoma (SCC); it can be used as a potential marker to distinguish the two histological types [19,20]. However, our results were inconsistent with the previous reports. We found that neither CK7 expression in tissues nor peripheral mononuclear cells was associated with histological types (all p > 0.05). CK7 mRNA was up-regulated in all 111 NSCLC patients, including 65 ADC and 46 SCC patients. The CK7 level in both tissues (p < 0.0001) and PBMCs (p = 0.006) was significantly correlated with the clinical stage. Moreover, circulating CK7 mRNA was significantly correlated with age during diagnosis (p = 0.038). NSCLC patients with low CK7 expression (fold < 2) have a long DFS while 92.5% (49/53) patients with high CK7 expression (fold ≥ 2) suffer progression, metastasis, or death. Therefore, CK7 overexpression in peripheral cells may be a good biomarker for predicting poor prognosis in NSCLC. CK7 is almost exclusively expressed in epithelial tissues, especially lung tissues. The expression of CK7 in lung cancer tissues is elevated and can be used to distinguish between primary lung cancer and lung metastasis of other cancers [21][22][23].
ELF3 is limited to cells of epithelial origin and is firstly found in lung tissues. Moreover, ELF3 is a transcriptional factor in the ETS-domain family and is known to regulate the expression of several growth-related genes, including angiopoietin 1, collagenase, as well as other transformed growth and invasion-related genes [24,25]. ELF3 may play a crucial role in lung tumorigenesis and progression. However, the additional functions of ELF3 in various kinds of tumors have frequently been studied for many years since its expression in lung cancer tissues and lung cancer cell lines was first discovered in 1997 [25]. ELF3 was demonstrated to possess dual functions in the transcriptional regulation of genes involved in squamous epithelial differentiation. ELF3 suppresses the basal keratin4 promoter activity while simultaneously activating the late differentiation linked small proline-rich protein 2A (SPRR2A) promoter in both esophageal and cervical epithelial cancer cell lines [26]. Nakarai et al. evaluated the mRNA expression of ELF3 and carcinoembryonic antigen (CEA) in the lymph node and the tissue from patients with colorectal cancer (CRC) and controls. The results showed that ELF3 may sufficiently assess the lymph node metastases of CRC [27]. In addition, ELF3 expression plays a critical role in lung tumorigenesis and is regulated by oncogenic protein kinase C (PKC) [28]. Our data indicate that ELF3 expression in NSCLC tissues is associated with lymph node metastasis and clinical stage; however, its expression in PBMCs is not correlated with lymph node metastasis. Moreover, the increased ELF3 expression correlated with poor prognosis. Hitherto, there is no report describing the association of ELF3 expression and metastasis of lung cancer. Hence, these results should be verified by a large sample analysis in the future studies.
EGFR is a biomarker used for the prediction of chemotherapy and targeted treatment. It is a representative marker and abnormal expression of EGFR influences the treatment and prognosis of lung cancer. When patients with lung cancer carry activating EGFR mutations, their first-line of treatment can be selected with EGFR-TKI, like gefitinib [29]. The prognosis of NSCLC patients with mutated EGFR was superior to that of the wild-type EGFR in NSCLC [30]. We found that high EGFR expression level was associated with short DFS and is an independent influencing factor for the prognosis in NSCLC (p < 0.0001). Also, it has been demonstrated that patients showed higher EGFR expression in SCC than non-SCC (p < 0.05) [31], and co-expression with insulin-like growth factor 1 receptor (IGF1R) was associated with poor survival. We found the EGFR expression did not correlate with the histological type (p > 0.05).
EphB4 plays a key role in numerous kinds of tumor, including lung cancer [32], esophageal cancer [33], pancreatic cancer [34], and gliomas [35]. As a member of receptor tyrosine kinases, EphB4 is frequently implicated in tumor pathogenesis. Moreover, the inhibition of some tyrosine kinases targets EphB4, and is deemed efficient treatment Ferguson et al. [36] found that EphB4 overexpression promotes cellular proliferation, colony formation, and motility while EphB4 inhibition reduces cell viability in vitro. The growth of the established tumors in mouse xenograft models can be halted by single-target strategy, and EphB4 may be a potential novel therapeutic target in lung cancer. EphB4 expression in tumors was increased significantly compared to the control (adjacent normal tissues), which was correlated with differentiation, lymph node metastasis, and TNM stage [32]. However, we found that EphB4 expression in PBMCs was not associated with the differentiation and lymph node metastasis, but correlated with the histological type and TNM stage. The survival curve of EphB4 was different from the other three genes. The patients with low EphB4 expression also underwent progression, metastasis, or death.

Patients
A total of 111 NSCLC patients who underwent curative surgery, without prior treatments, at Zhejiang Cancer Hospital (Hangzhou, China) from January 2010 to December 2011, were enrolled in this study. All the patients showed no history of other tumors by examination of a plain chest radiograph, CT scan, fiber optic bronchoscopy, and bone scan and were diagnosed with NSCLC by histopathology. The patients' medical records were reviewed to obtain data including age at diagnosis, sex, and smoking history (Table 1). Tumor stage was determined according to the American Joint Committee on Cancer (AJCC)/ Union for International Cancer Control (UICC) TNM tumor classification [37]. The mean age of the patients at tumor resection was 62 years (range 44-82 years); 79 (71.2%) were males and 32 (28.8%) were females. The tumor specimens and paired PCHNTs were collected at the time of surgery. The paired PCHNTs were obtained from the proximity of 5 cm from the tumor edge and was assessed microscopically for the presence of healthy cells and absence of dysplastic cells. The paired peripheral blood samples were collected from these patients during diagnosis before surgery, radiation, or chemotherapy. The tissue and blood specimens for the present study were procured and used after obtaining informed consent from all the participants. As a measure of prognosis, we analyzed the clinical data concerning DFS and OS. All recruited patients had been followed-up periodically until the due date. The mean follow-up duration was 29 months; the average survival time was 32 months (range, 13-52 months) for all patients.
In addition, peripheral blood mononuclear cells (PBMCs) from 115 benign pulmonary disease patients and 120 healthy subjects as normal controls were also included to examine the specificity of these markers. PBMCs were prepared as described below.
Informed consent was obtained from all individual participants included in the study. The research involved human participants. The study was approved by the Zhejiang Cancer Hospital Institutional Review Board (2012KYB035).

Sample Collection and Processing
Biopsies of NSCLC and paired PCHNTs were obtained from thoracic surgery patients. The peripheral blood samples were obtained from NSCLC patients. Peripheral blood samples from 115 benign pulmonary disease patients and 120 healthy subjects were provided by our hospital. Fresh tissue samples (<300 mg) were immediately ground into powder in liquid nitrogen. Total RNA was extracted from these samples using 1 mL TRIzol (Invitrogen, Carlsbad, CA, USA) and stored at −80 • C until further use. The peripheral blood samples were collected and processed as described previously [37]. In order to avoid contamination with skin epithelial cells, the first 2 mL blood was discarded while collecting the blood samples (7.5 mL). The PBMCs containing CTCs were prepared using Red Blood Cell Lysis Buffer (Solarbio, Beijing, China) and the blood samples were processed within 30 min. After centrifugation at 2000× g for 15 min, buffy coat was collected and washed once with phosphate buffered saline (PBS; containing 0.14 g/L KH 2 PO 4 , 9 g/L NaCl, 0.8 g/L Na 2 HPO 4 , pH 7.4). The cells were pelleted by centrifugation. Subsequently, the tumor cell enrichment using anti-EpCAM antibodies, MACS HEA MicroBeads ® , density gradient centrifugation, and OncoQuick ® plus (Miltenyi Biotec Inc.,Bergisch Gladbach, Germany) was performed as described previously [17,[38][39][40]. For each cytometric enrichment procedure, 15 mL peripheral blood was used.

RNA Extraction and cDNA Synthesis
Total RNA of fresh tissue was isolated with TRIzol (Invitrogen, Carlsbad, CA, USA). The total RNA of PBMCs was extracted using the MiRNeasy Mini Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. The concentration and purity of the total RNAs were assessed by Ultraviolet spectrophotometer, and the integrity of total RNAs was estimated by polyacrylamide electrophoresis (PAGE). cDNA synthesis was performed using PrimeScript TM RT reagent Kit (Takara, Otsu, Japan) for tissue RNA and PrimeScript TM miRNA cDNA Synthesis Kit (Takara, Otsu, Japan) for PBMCs RNA.

Real-Time RT-PCR
The mRNA levels of the genes of interest were estimated by real-time RT-PCR on Applied Biosystems 7500 (Foster City, CA, USA) using SYBR ® Premix Ex Taq TM II (Takara, Otsu, Japan). The sequences of primers were designed as described previously [17,18]. To prepare an artificial CTC model as a positive control, a trace amount of A549 human lung adenocarcinoma cells was mixed with peripheral blood samples (five A549 cells in 7.5 mL peripheral blood). Then, the total RNA was extracted from the mixture and used as the positive control in real-time RT-PCR. The total amount of mRNA was normalized to GAPDH, and the relative mRNA expressions of the samples (tumor vs. PCHNTs, or paired PBMCs vs. positive control) were calculated by 2 −∆∆Ct . The differential expression of the samples may be defined as the upregulation when the cut-off value was set at 2-fold.

Statistical Analysis
The SPSS 17.0 statistical software package was used for all the statistical analyses. The correlation between the level of the above described four mRNAs and the clinicopathological characteristics was analyzed by Person's chi-square test or Fisher's exact test. The survival curves and univariate analysis were generated by the Kaplan-Meier method and log-rank test. Multivariate analysis was performed by the Cox regression model and Wald test. p < 0.05 was considered statistically significant.

Conclusions
In present study, we selected CK7, ELF3, EGFR and EphB4, four epithelial origin markers, to evaluate their efficiency for detecting CTCs in NSCLC. The expression of CK7, ELF3, EGFR and EphB4 in NSCLC tissues and para PMBCs were both upregulated significantly. ROC curve analysis indicated that these four genes have high sensitivity and specificity for detecting CTCs. Moreover, upregulation of four genes in PMBCs are associated with tumor progression closely. The patients with high expression of these genes in PMBCs have poor survival. We speculated that CK7, ELF3, EGFR and EphB4 expression in PMBCs are appropriate markers for detecting CTCs and have potential in the evaluation of prognosis and monitoring of therapy in NSCLC.