Clinical and experimental studies regarding the expression and diagnostic value of carcinoembryonic antigen-related cell adhesion molecule 1 in non-small-cell lung cancer

A total of 69 serum samples were included in this study, including 35 samples that were collected from NSCLC patients before surgery (average age: 60, range: 34 to 78 years; samples with any other disease outside the lungs were excluded) and 34 samples that were collected from sex- and age-matched healthy volunteers who passed all of the routine medical examinations (e.g., blood tests, X-ray, and computerised tomography test) without abnormal results. Informed consent was obtained from all of the patients who participated in the study, which was conducted with the approval of the Ethics Committee of the Scientific and Ethical Committee of Shanghai Jiao Tong University in accordance with the Helsinki declaration of 1975 (as revised in 1983). All the patients were diagnosed with cancer for the first time, and none previously received chemotherapy or radiation therapy.

Briefly, the samples were maintained at room temperature for approximately 30 minutes and then centrifuged at 1,300 g at 4°C for 20 minutes. The serum was then collected, divided into aliquots and frozen at -80°C until analysis. We also collected 21 paired (tumour and normal mucosa) tissue samples in parallel from the same 35 NSCLC patients mentioned above to assay by quantitative real-time polymerase chain reaction (qRT-PCR). During the surgical removal of each tumour, an adjacent section of normal mucosa was also removed for normal background tissue following pathological confirmation that it was free from tumour deposits. Tissues were obtained and snap frozen in liquid nitrogen. Additionally, 13 of 21 pairs of tissues were analysed for CEACAM1 isoforms: 6 cases of adenocarcinoma, 5 cases of squamous cell carcinoma, 1 case of poorly differentiated carcinoma, and 1 case of a lymphoepithelioma-like carcinoma. Detailed clinical and pathological information of the samples can be found in Additional file 1: Table S1.

Serum CEACAM1 was analysed with enzyme-linked immunosorbent assay (ELISA) kits (RayBiotech, Atlanta, GA, USA) according to the manufacturer's instructions. Briefly, a 96-well microplate was precoated with anti-human CEACAM1, which recognises the extracellular domain (a.a. 35-428). Before use, all of the reagents and samples were brought to room temperature (18-25°C). The standard dilution series ranged from 20.58 to 15,000 pg/ml. First, 100 μl of each standard or serum sample (1:100 prediluted) was added to the appropriate wells and incubated for 2.5 hours at 24°C with gentle shaking. After discarding the solution and washing 4 times, 100 μl of prepared biotinylated anti-human CEACAM1 antibody was added to each well and incubated for 1 hour. After washing away unbound biotinylated antibody, 100 μl of horseradish peroxidase (HRP)-conjugated streptavidin was pipetted into the wells and incubated for 45 minutes, and 100 μl of 3,3’ ,5,5’-Tetramethylbenzidine (TMB) one-step substrate reagent was added after 5 washes. Subsequently, 50 μl of stop solution was added to each well, and the plate was immediately read at 450 nm.

In addition, we assayed carcinoembryonic antigen (CEA; ARCHITECT i2000 SR, Abbott, Chicago, IL, USA) and neuron-specific enolase (NSE; cobas e601 Roche, Basel, N.A., Switzerland) in all of the serum samples. The CEA and NSE cut-off values were 5.0 and 17 ng/ml, respectively. Tumour markers with serum values higher than the cut-off were classified as positive, while those with values lower than the cut-off were negative.

To determine the expression and location of CEACAM1, a total of 21 specimens were stained using immunohistochemistry method, which was performed on paraffin-embedded specimens with the mouse monoclonal anti-CEACAM1 antibody (Abcam; 29H2, Cambridge, UK, dilution: 1:75). Briefly, the sections were dewaxed, and endogenous peroxidase was blocked by immersing the slides in a 3% solution of hydrogen peroxide in methanol for 10 minutes followed by antigen retrieval. The slides were microwaved for 10 minutes in 10 mmol/L citrate buffer, pH 6.0. After washing 3 times in 1 mol/l phosphate-buffered saline (PBS; pH 7.4) for 5 minutes, the sections were blocked with normal rabbit serum in a humidity chamber for 30 minutes at room temperature. The excess serum was rinsed off with 1 mol/l PBS, and the sections were incubated with primary antibodies in a humidity chamber overnight at 4°C. The next morning, sections were rinsed with PBS before incubation with the biotinylated second antibody in a humidity chamber for 40 minutes at 37°C. After rinsing with PBS, the streptavidin–peroxidase complex reagent (StrepABComplex/HRP Duet, DAKO, Glostrup, Denmark) was added. Diaminobenzidine and hydrogen peroxide were used for visualisation. For negative controls, the primary antibody was replaced with PBS.

All the 21 NSCLC were categorized into a high expression group (i.e., ≥ 66% positive tumour cells) and a low expression group (i.e., < 66% positive tumour cells) according to the percentage of positive tumour cells [23, 29, 30] (Figure 1). The high or low classifications were independently assigned by two experienced pathologists and consensus was achieved after discussion.

Total RNA was isolated using the TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s instructions. The purity and concentration of RNA were determined with a spectrophotometer at 260 nm. Complementary DNA (cDNA) was generated by the PrimeScript® RT reagent Kit (Takara Biotechnology, Otsu, Shiga, Japan). GAPDH was used as an internal control for each sample. Quantitative real-time PCR was performed using specific primers (Table 1). Briefly, 1 μg of total RNA was denatured for 5 minutes at 70°C and cooled for 5 minutes on ice. Reverse transcriptase (RT) was added to a total volume of 20 μl, and reverse transcription was performed for 15 minutes at 37°C followed by 5 seconds at 85°C, according to the protocol recommended by the manufacturer. The synthesised cDNA was either used immediately for PCR amplification or stored at -20°C for further analysis. Quantitative PCR was performed in a total reaction volume of 20 μl per capillary in the LightCycler format for 40 cycles (denaturation: 30 seconds at 95°C, annealing: 20 seconds at 60°C, and extension: 15 seconds at 72°C). After PCR amplification, melting curve analysis was performed to verify the specificity of the test.

The reverse transcription PCR primers as reported by Wang et al. [31] and Gaur et al. [32] (Table 1) were designed to distinguish the 408 bp (CEACAM1-L) and 355 bp (CEACAM1-S) CEACAM1 isoforms. The PCR was performed in a total volume of 50 μl containing 0.25 μl of TaKaRa Ex Taq (5 U/μl), 5 μl of 10× Ex Taq Buffer (Mg²⁺ Plus), 4 μl of dNTP mixture, 2 μl of cDNA template, 1 μl of each forward and reverse primer and ddH₂O. After pre-denaturation at 94°C for 5 minutes, the PCR cycling conditions for CEACAM1 and GAPDH were as follows: 30 cycles of 94°C for 1 minute, 60°C for 30 seconds, and 72°C for 30 seconds. The reaction was performed with an Eppendorf thermal cycler (Eppendorf, Hamburg, Germany). At the end of the reaction, the mixtures were loaded onto a 2% agarose gel and stained with ethidium bromide prior to examination under UV light.

L-form CEACAM1 and S-form CEACAM1 levels were represented as integral optical density (IOD) values with Image Pro Plus V6.0 for Windows (Media Cybernetics, Inc., Rockville, MD, USA). Briefly, after intensity rectification, IODs were obtained as the ratio of sum optical density (OD) to the sum area, which is proportional to the quantity of RNA. Most of the data were not normally distributed. Thus, they were expressed as a median or a range. The Mann–Whitney and Kruskal–Wallis tests were used to determine the significance of two independent groups and various groups, respectively. Nonparametric receiver operating characteristic (ROC) curves in which the value for sensitivity is plotted against the false-positive rate (1-specificity) were generated to assess the diagnostic accuracy of serum CEACAM1. Receiver operating characteristic (ROC) curves are measured to test whether the area under the curve (AUC) of the ROC exceeds 0.5. If not, no further assessment of the diagnostic test is warranted. Statistical significance in this study was set at P < 0.05, and all of the reported P values are 2-sided. All of the analyses were performed with SPSS v.16 for Windows (SPSS Inc., Chicago, IL, USA) or SigmaPlot V. 12 for Windows (Systat Software Inc., San Jose, CA, USA).

The clinical and pathological characteristics of patients are shown in Table 2. The median serum CEACAM1 level was significantly higher in patients with NSCLC compared with normal healthy controls (P < 0.0001; Figure 2A). For patients with NSCLC, the median CEACAM1 level was 544.79 ng/ml (range: 381.30 ~ 968.13 ng/ml), and for normal controls, the median was 386.20 ng/ml (range: 226.80 ~ 490.11 ng/ml). Patients who were at an early stage of disease (stage I and II disease) showed significantly higher CEACAM1 levels than patients in stage III and IV (P = 0.016; Table 2). Moreover, serum CEACAM1 levels were significantly lower in female patients than in male patients (Table 2).

In multivariable logistic regression analysis, CEACAM1 levels significantly predicted NSCLC vs. normal control (OR: 1.052; 95% CI: 1.022 ~ 1.083; P < 0.001) when adjusted for age and gender effects. The ability of serum CEACAM1, CEA and NSE to predict NSCLC was analysed by nonparametric ROC analyses. When used to distinguish NSCLC from normal healthy individuals, the AUCs for serum CEACAM1, CEA and NSE were 0.96 (95% CI: 0.9148 ~ 0.9995; P < 0.001), 0.91 (95% CI: 0.8454 ~ 0.9773; P < 0.001) and 0.98 (95% CI: 0.9302 ~ 1.023; P < 0.001), respectively. Using the cut-off level of 440.3 ng/ml (according to the Youden index), serum CEACAM1 had a sensitivity of 97%, a specificity of 82%, a positive predictive value of 70%, and a negative predictive value of 95% (Figure 2B, Additional file 2: Table S2). Therefore, according to the AUC of the ROC curve, CEACAM1 was better than CEA but did not exceed NSE.

In the clinic, the cut off values for CEA and NSE were set to 5 and 17 ng/ml, respectively. The corresponding diagnostic accuracy of CEACAM1, CEA and NSE is summarised in Additional file 2: Table S2. Generally, CEACAM1 had a significantly higher sensitivity (97%) than CEA (29%, P < 0.05) and NSE (20%, P < 0.05). However, CEA (97%) and NSE (97%) had better specificity than CEACAM1 (82%), not significantly (P > 0.05). The negative predictive value of CEACAM1 (95%) was higher compared with CEA (57%) and NSE (54%), but the positive predictive value of CEACAM1 (70%) was lower than CEA (91%) and NSE (88%), which is likely a result of the low sensitivity of CEA and NSE.

By immunohistochemical staining, we found that the CEACAM1 expression was restricted to neoplastic epithelium in all the specimens of 21 patients (Figure 1A). No CEACAM1 expression was found in normal cells adjacent to the tumours or in the negative controls (Figure 1B). Tumours of 17 patients (81%) were classified as high expression (Figure 1A, i.e., ≥66% positive tumour cells), and specimens of 4 patients (19%) were classified as low expression (Figure 1B, i.e., <66% positive tumour cells).

The 2^-△Ct (-△∆Ct = Ct,_GAPDH-Ct,_CEACAM1) method was employed. Although 15 of 21 subjects showed higher CEACAM1 mRNA levels in tumours compared to adjacent tumour-free tissues, no significant differences were found between the mRNA expression of CEACAM1 in tumours and normal tissues by the Wilcoxon signed-rank test for 2 related samples (Figure 3A). Further studies of CEACAM1 mRNA levels and the patient clinical and pathological characteristics were shown in Table 3. CEACAM1 mRNA levels were significantly higher in male patients than in female patients (P = 0.028), which was consistent with the serum protein levels (Table 2). CEACAM1 mRNA levels also showed a significant negative correlation with tumour invasive extension (P = 0.039), which is in accordance with the serum protein levels. In addition, patients with adenocarcinoma showed higher CEACAM1 mRNA levels than squamous cell carcinoma or other types (P = 0.003). No other significant association was found between clinical characteristics and CEACAM1 mRNA levels.

To determine whether the expression of CEACAM1-L and CEACAM1-S are altered in primary NSCLC, we analysed 13 pairs of primary tumour and normal lung tissue specimens with the same PCR primers reported by Gaur et al. [32] and Wang et al. [31]. The forward primer is located in exon 6, and the reverse primer is located within the 3’ untranslated region. Thus, the primers can amplify a 408 bp fragment (CEACAM1-L) or a 355 bp fragment (CEACAM1-S) simultaneously by inclusion or exclusion of exon 7. As shown in Figure 3B, CEACAM1-S was predominantly expressed in lung tumours tissues, whereas the normal tissues predominantly expressed CEACAM1-L. In total, 12 of 13 lung tumours had a CEACAM1-S/CEACAM1-L (S: L) ratio greater than 1 (S-form > L-form), whereas normal lung tissue did not. Further statistical data were consistent with this finding. The CEACAM1-S and the CEACAM1-S/CEACAM1-L (S: L) ratio was significantly higher in tumours than in normal tissues (P = 0.023 for CEACAM1-S and 0.016 for the CEACAM1-S/CEACAM1-L (S: L) ratio; Figure 3C and 3D, Table 4). However, the expression of CEACAM1-L did not show a marked difference between tumour and normal tissue (Figure 3C, Table 4).