Prognostic Significance of the Detection of Peripheral Blood CEACAM5mRNA-Positive Cells by Real-Time Polymerase Chain Reaction in Operable Colorectal Cancer

Colorectal cancer (CRC) represents approximately 10% of all cancer cases (1). Although curative resection is possible in 2 of 3 of patients at diagnosis, almost one third of them will eventually develop relapse and will die of the disease (2).

Dissemination of tumor cells is an obligatory step toward progression to metastatic disease. Disseminated CRC cells can be identified in various body compartments using immunophenotyping and reverse transcription-polymerase chain reaction (RT-PCR) techniques (3, 4). In patients with early-stage CRC, the detection of malignant cells in the bone marrow, the peritoneal lavage, and the regional lymph nodes is associated with poor survival (5, 6)

Detection and quantification of circulating tumor cells CTCs in blood samples could serve as a unique and easy diagnostic tool to determine prognosis and monitor treatment effectiveness (7). Quantitative real-time RT-PCR (RT-qPCR) has been shown to provide the sensitivity and the practicability that is necessary to detect rare CTCs in patients' blood. (8) In CRC, the most frequently analyzed marker is the carcinoembryonic antigen (CEA). The initial clinical studies have analyzed the usefulness of CEACAM5mRNA for the detection of CTCs in blood samples of patients with CRC, but none of them have quantified the PCR product (4, 9–11). Nevertheless, a low level of CEACAM5mRNA expression has been detected in normal subjects, suggesting an illegitimate expression of the CEACAM5 gene in hematopoietic cells (12, 13) whereas other studies have failed to show significant differences of CEACAM5mRNA expression between cancer patients and healthy individuals (14). These studies have been conducted using sets of primers that amplify a splice variant of CEACAM1 expressed in normal white blood cells (WBC), in which an “intron” sequence replaces part of the exon 10 (12), thus explaining the findings of CEACAM5mRNA expression in normal blood samples (13) or in patients with inflammatory bowel disease (14) and in cultured WBCs after induction with G-CSF (3). On the basis of the aforementioned results, the overall usefulness of CEACAM5 as a PCR-based tumor cell detection marker remains questionable.

The aim of this study was to investigate the presence of CEACAM5mRNA-positive CTCs using a new RT-qPCR assay, which is designed to avoid the amplification of the spice variant expressed in normal WBCs and to determine their prognostic significance in patients with operable (stages II–III) CRC.

Since January 1998, peripheral blood (15 mL in EDTA) was obtained from consecutive patients with CRC at the time of the first visit (usually 3–8 weeks after surgery) in the Department of Medical Oncology of the University Hospital of Heraklion, on a routine basis after obtaining the patient's consent. Totally, blood samples were obtained from 269 patients with stage II–III CRC. Four patients were excluded from the analysis due to incomplete clinical data (1 case) or inadequate follow-up (3 cases). In this analysis, 265 patients with operable CRC (stages II–III) with a minimum follow-up of 24 months were included. Histopathologic data were obtained from the pathology reports and included the extramural vascular invasion status, tumor differentiation, maximum depth of invasion (pT), and lymph node involvement (pN). For the evaluation of the analytical sensitivity and specificity of the method, blood samples from 96 healthy blood donors, 15 patients with colonic adenomas, 21 patients with inflammatory bowel disease (IBD), and 100 patients with metastatic CRC (mCRC) were analyzed. All blood samples were obtained at the middle-of-vein puncture after the first 5 mL of blood were discarded to avoid contamination with epithelial cells from the skin. Adjuvant chemotherapy was administered in 257 (97%) patients, and in 102 of the cases (39%) an oxaliplatin-based regimen was used. Blood samples from 104 patients were also obtained within 1 month after the completion of the treatment. Patients were followed with clinical examination and laboratory tests every 3 months for the first 3 years and every 6 months for the next 2 years, and yearly thereafter. Chest and abdominal pelvic computed tomography was done yearly for the first 3 years and subsequently at the discretion of the physician. The study has been approved by the Ethics and Scientific Committees of our Institution, and all patients and donors signed informed consent.

Peripheral blood mononuclear cells (PBMC) were obtained by gradient density centrifugation and RNA extraction was done as previously described (7). RNA concentration was measured using the NanoDrop (Thermo Scientific) equipment. Amplification of the β-actin as internal reference gene was done to verify the RNA integrity. RNA prepared from the Lovo and ARH-77 CRC and leukemic cell lines, respectively, was used as positive and negative controls.

The SuperScript III PlatinumTranscriptase (Invitrogen) was used to prepare cDNA from 5 μg of total RNA, according to the manufacturer protocol. Also, 1 μg of colonic commercial RNA control (Stratagene) was used in serial dilutions (1:10) to verify the efficacy and reproducibility of the reaction.

The RT-qPCR reaction for β-actin and CEACAM5 was carried out using 2.5 μL of cDNA template and 6.25 μL of TaqMan Universal Master Mix (AB; Applied Biosystems), 1.25 μL– of specific primers (3 μmol/L), and 0.5 μL of hydrolysis probe (2.5 μmol/L) for each gene and adjusted with DPEC water to a final volume of 12.5 μL/reaction. Quantification of gene expression was performed using the ABI Prism 7900HT Sequence Detection System (AB). All experiments were performed in triplicates. Only triplicates with a standard deviation of the quantification cycle (C_q) value less than 0.25 were accepted.

Quantification was based on an external calibration curve that was obtained by using external standard cDNAs. Total RNA was prepared from 1 × 10⁶ Lovo cells. cDNA synthesis of serial dilutions of this RNA, corresponding to 1–10⁵ Lovo cells, was also analyzed in each run. This calibration curve was created by plotting the number of Lovo cells corresponding to each external standard cDNA versus the value of its C_q. The number of CEACAM5mRNA-positive cells for all of the tested samples was expressed as Lovo cell equivalents/5 μg of total RNA according to the external standard calibration curve. In addition, to determine the number of Lovo cells that could be recovered, we added decreasing number of Lovo cells (from 10³ to 1) in 10⁶ negative PBMCs in 5 different experiments. The SDS 2.3 software was used for the analysis of the results.

The primers and probe sets were designed using Primer Express 2.0 Software (AB). The β-actin primers and probe set have been previously described (15) whereas those for CEACAM5 (Supplementary Fig. 1) were designed according to the Ref Seq NM_004363.2 (http://www.ncbi.nlm.nih.gov/LocusLink). To avoid the amplification of the splice variant, which is expressed in normal WBCs, the probe was designed to spread across the junction of exon 9 with exon 10. BLAST search confirmed the specificity of this set of primers, as no homology to pseudogenes, splicing variants, or unexpected targets was found. As CEACAM1 (Ref Seq NC_001024912.1) is approximately 90% homologous with CEACAM5, we designed primers and probes for this family member and explored the specificity of the assay both in HL60 leukemic cell line and in PBMCs of a healthy volunteer (HV) for the expression of CEACAM1 (16, 17) and the MCF7 cell lines for that of CEACAM5 (18). Finally, genomic DNA contamination was excluded, as no RNA transcripts could be detected in each analyzed sample in the absence of reverse transcriptase.

The aim of this study was to explore the prognostic significance of CEACAM5mRNA-positive CTCs in patients with stage II–III CRC. Disease-free survival (DFS) and overall survival (OS) were calculated from the day of surgery to the first documented relapse or death, respectively. Relapse was defined as the finding of metastatic disease, local recurrence, or a second primary tumor. The evaluation for the presence of CEACAM5mRNA-positive cells was done blindly to clinical data. The potential association between baseline characteristics, relapse, and CEACAM5mRNA-positive cells was compared with the 2-sided Fisher exact test for categorical variables. The association of risk factors with time-to-event endpoints was analyzed with the log-rank test, and the Kaplan–Meier method was used to plot the corresponding DFS and OS curves. Univariate and multivariate Cox proportional hazards regression models with hazard ratios and 95% CIs were used to assess the association between potential prognostic factor and DFS or OS. Statistical significance was set at P ≤ 0.05.

cDNA obtained from Lovo cells was analyzed to determine the analytical sensitivity and linearity of the RT-qPCR assay. The calibration curves of these experiments proved the linearity of the assay over the entire quantification range (1–10⁵ Lovo cells) with a correlation coefficient of 0.994 (Supplementary Fig. 2). The RT-qPCR efficiency expressed as E = 10^(−1/slope)−1 (19) was 0.97 ± 0.029 SD [coefficient of variation (CV): 3.0%; n = 18) and the mean value of slope and intercept were 3.516 ± 0.11 (CV: 3.1; n = 18) and 30.28 ± 0.78 (CV: 2.6%; n = 18 experiments), respectively. The limit of detection (LOD) of the assay was found to correspond to 0.7 Lovo cell equivalents/5 μg of RNA (LOD = 3.3 SD/slope, where SD is the standard deviation of the C_q for 1 Lovo cell equivalent; refs. 8, 20).

For the determination of the repeatability of the assay, cDNA samples corresponding to 1 to 10.000 Lovo cells were assayed in the same run in 6 parallel determinations, using the same calibration curve (21). The SDs of the C_q values and the CVs for Lovo cells are shown in Supplementary Table 1, which shows the repeatability of the assay. The results obtained from the same RNA, which was frozen in aliquots, and analyzed over a period of 6 weeks in 12 different experiments done in different days confirmed the reproducibility of the assay (22). In addition, as summarized in Supplementary Table 2, in 5 different experiments we were able to recover 10 Lovo cells in 10⁶ normal PBMCs (1:10⁵).

Only CEMCAM1 was amplified in HL60 cells, whereas only CEACAM5 was amplified in MCF7 cells with the use of conventional PCR using 100 ng of genomic DNA/reaction (Supplemental Fig. 3A). These results were also confirmed in the RT-qPCR starting from 5 μg of RNA for each gene (Supplemental Fig. 3B). In addition, sequencing of the RT-qPCR product confirmed the specificity of the assay (Supplemental Fig. 3C).

The incidence of detection of CEACAM5mRNA-positive cells in the controls and patients is presented in Table 1. According to the cutoff point (>0.7 Lovo cells equivalents/5 μg of RNA), only 2 (2.2%) samples from the 96 healthy volunteers (HV) tested positive, suggesting a specificity rate of 97.8%. Table 1 also indicates that 1 (4.6%) patient with IBD but none with colonic adenomas had detectable CEACAM5mRNA-positive cells. In contrast, CEACAM5mRNA-positive cells could be detected in 98 (37%) and 44 (44%) patients with stage II–III and IV CRC, respectively (P = 0.4; Table 1). The median number of CEACAM5mRNA-positive CTCs was 1.72 (range, 0.72–26.5) and 4.18 (range, 0.94–218.2) in patients with stage II–II and metastatic disease, respectively (P = 0.4; Table 1). There was no significant association for the detection of CEACAM5mRNA-positive CTCs with any of the examined clinicopathologic features in patients with stage II–III disease (Table 2). In contrast, in patients with metastatic disease, liver involvement was associated with significantly higher CEACAM5mRNA-positive CTCs detection (52% vs. 31% in patients with or without liver metastasis; P = 0.021; Table 3).

After a median follow-up period of 34.1 months (range, 24.2–122.4 months), 57 (22%) relapses were observed among the 265 patients with operable CRC. All relapsed patients had distant metastases and 7 also presented local relapse. The incidence of relapses was significantly higher in patients with CEACAM5mRNA-positive CTCs than in patients without (37% vs. 12%; P < 0.001; Table 4). The median DFS has not been reached, but it was significantly higher in patients without detectable circulating CEACAM5mRNA-positive cells compared with patients with (P < 0.001; Fig. 1A). In addition, the probability of 3-year relapse-free survival was 83% and 58% in patients with and without detectable CEACAM5mRNA-positive CTCs, respectively (P = 0.001). Moreover, patients with undetectable CEACAM5mRNA-positive CTCs before and after the administration of the adjuvant treatment (n = 54) experienced significantly increase DFS in comparison with those (n = 20) with detectable CEACAM5mRNA-positive CTCs before and after adjuvant treatment (P = 0.017) or with those (n = 9) with detectable CEACAM5mRNA-positive CTCs only after the completion of the adjuvant treatment (P = 0.005; Fig. 1C). Also, in the patients with metastatic disease, detection of CEACAM5mRNA-positive CTCs was correlated with significantly decreased progression-free survival (median = PFS 6.6 and 9.2 months in patients with and without detectable CEACAM5mRNA-positive CTC, respectively; P = 0.043).

During the follow-up period, 45 of 265 patients (17%) died because of disease progression (Table 4). Twenty-four (24%) deaths occurred in patients with detectable CEACAM5mRNA-positive CTCs and 21 (12%) in patients without (P = 0.017; Table 4). The median OS of patients without detectable CEACAM5mRNA-positive CTCs was significantly higher than that of patients with detectable CEACAM5mRNA-positive CTCs (P = 0.025; Fig. 1B). The 3- and 5-year probability of survival was 79% and 57%, respectively, for CEACAM5mRNA-positive patients in comparison with 88% and 78%, respectively, for CEACAM5mRNA-negative patients. Also, patients with undetectable CEACAM5mRNA-positive CTCs before and after the administration of the adjuvant treatment (n = 54) experienced significantly increased OS in comparison with those (n = 20) with detectable CEACAM5mRNA-positive CTCs before and after adjuvant treatment (P = 0.001) or with those (n = 9) with detectable CEACAM5mRNA-positive CTCs only after the completion of the adjuvant treatment (P = 0.001; Fig. 1D). Similarly, in patients with metastatic disease, the detection of CEACAM5mRNA-positive CTCs was associated with decreased survival (21.6 months vs. 17.3 months; P = 0.047).

The univariate analysis showed that CEACAM5mRNA-positive patients presented higher risk of progression than CEACAM5mRNA-negative patients (HR = 3.4; P < 0.001; Table 5). Also, disease stage (III vs. II), tumor differentiation (poor vs. well/moderate), and presence of neural invasion and cancerous vessels emboli (CVE) were associated with decreased DFS (Table 5). Multivariate analysis confirmed that the detection of CEACAM5mRNA-positive CTCs was independent prognostic factor for decreased DFS (HR = 3.4; 95% CI: 2.0–5.9; P < 0.001; Table 6). Moreover, stage III (HR = 3.0; 95% CI: 1.5–5.9; P = 0.002) and the presence of poorly differentiated tumors (HR = 1.9; 95% CI: 1.1–3.3; P = 0.022) also emerged as independent prognostic factors associated with decreased DFS (Table 6).

Similarly, patients with detectable CEACAM5mRNA-positive CTCs presented higher risk of death than patients without CEACAM5mRNA-positive cells (HR = 1.9; P = 0.028; Table 5). Disease stage, tumor differentiation, and presence of CVE were also significantly associated with OS. The multivariate analysis revealed that stage III (HR = 1.9; P = 0.048) and the presence of CVE (HR = 2.4; P = 0.008) were independent prognostic factors associated with decreased survival (Table 6).

The aims of this study was to develop a reliable and reproducible RT-qPCR assay, using CEACAM5mRNA as a detection marker for CTC quantification in patients with CRC and to evaluate their detection as a prognostic factor. In breast cancer, after “curative resection,” the rates of CTC detection are decreased after few weeks in a majority of the cases and the persistence of CTCs after have prognostic significance. (23) For this reason, we decided to obtain the blood samples just before the initiation of adjuvant chemotherapy (for stage III or high-risk stage II patients) or follow-up.

The results of our study should not be compared with those of previous studies in which PCR-based assays have been used for the detection of CEACAM5mRNA-positive cells. Most of these studies used qualitative assays (4, 6, 11, 24) and, most likely, included “false-positive” results due to the amplification of a splicing variant of CEACAM1, which is expressed in WBCs (12). On the basis of this assumption, we designed a specific set of primers and probe that did not amplify this variant and we proved the high specificity and efficiency of the method. Subsequently, we observed a significant decreased DFS (P < 0.001) and OS (P = 0.017) in patients with operable CRC and detectable CEACAM5mRNA-positive cells, whereas the detection of CEACAM5mRNA-positive cells was revealed as an independent prognostic factor for decreased DFS (HR = 3.4; P < 0.001). Also, a trend for increased risk of death (HR = 1.9; P = 0.065) was recorded that did not reach statistical significance probably due to the limited number of events occurring during the short follow-up period. In contrast, the presence of CVE was revealed as an independent prognostic factor for decreased OS in agreement with several previous reports (25, 26). In addition, in the subset of patients with available specimen at the completion of the adjuvant chemotherapy, the persistence of CEACAM5mRNA-positive cells was a strong prognostic factor of decreased DFS and OS. This preliminary finding provides a rational for the study of whether the detection CEACAM5mRNA-positive cells could be used as a surrogate marker for the efficacy of the adjuvant treatment.

There was no significant difference in the percentage of patients with detectable CEACAM5mRNA-positive cells between stage II/III and IV patients, even though the higher actual number of CTCs in stage IV patients. Of notice that within stage IV patients, those with liver involvement had a significantly higher percentage of CEACAM5mRNA-positive cells (P = 0.021). These results are in agreement with those reported in studies using cell-based techniques. Indeed, it has been reported that by using the CellSearch platform (Veridex LLC), ≥2 CTCs/7.5 mL blood could be detected only in patients with malignant epithelial tumors (27) whereas the CTC detection rate in CRC patients was 36.2% and the quantification of CTCs could be a valuable prognostic factor (28). A multicenter prospective study showed that CTCs could be detected in 60% of patients before treatment and was the strongest predictive factor for PFS and OS and that liver involvement was associated with significantly higher percentage of CTCs detection (29). Another interpretation of these results could be that none of the methods used since now are capable of detecting all the CTCs in circulation, and preliminary results from studies using advanced techniques indicated that CTCs could be detected in almost all patients with CRC (30).

Also, there was no difference in the percentage of patients with detectable CTCs between stage II (39%) and stage III (36%). The fact that 81% of the stage II patients were “high” risk stage II patients may partially explain this finding, as these patients had similar prognosis with those with stage IIIa or even IIIb disease. Also, the multivariate analysis revealed that the detection of CEACAM5mRNA-positive cells was prognostic factor for decreased DFS independent of stage (Table 6), whereas Sastre et al. reported almost identical detection rates in stage II (20.7%) and III (24.1%) patients (28). Moreover, it has been shown that CK-19mRNA–positive cells could be detected in 31% of breast cancer patients with negative axillary lymph nodes (31), indicating that dissemination of tumoral cells could be independent from lymph nodes infiltration.

However, as the presence of CTCs is a necessary (although not sufficient) step in the metastatic process, these cells probably define a subset of tumor cells that may lead to the development of metastasis. Metastatic cells might present very different molecular profiles from those of the primary tumor (30). Therefore, it is conceivable that the identification of CTCs might be used as a molecular signature of the metastatic ability of a specific tumor over time (32). The isolation and molecular characterization of CEACAM5mRNA-positive CTCs could address these important considerations and may allow for noninvasive genotyping of cancer cells, which could be monitored at specific therapeutic decision-making points during the patient treatment period (33).

Because of the weaknesses in the design of this study (not uniformally treated patients, lack of a validation patients' cohort), the results should be evaluated with caution. Future studies are warranted to evaluate whether the monitoring of CEACAM5mRNA-positive CTCs during the treatment of patients with early or metastatic disease could be of any clinical relevance influencing the therapeutic decision.

No potential conflicts of interest were disclosed.

This work was partly supported by research grants from the Cretan Association for Biomedical Research (CABR). Z.S. is a recipient of a CABR postdoctoral fellowship.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.