Detection of lymph node metastasis of oesophageal cancer by RT-nested PCR for SCC antigen gene mRNA

With recent development in molecular biology, reverse transcriptase polymerase chain reaction (RT-PCR) has been applied to detect occult lymph node metastasis, but there have been few reports concerning oesophageal cancer. The objective of this study is to investigate the usefulness of the squamous cell carcinoma (SCC) antigen gene as a marker with RT-nested PCR to detect occult lymph node metastases of oesophageal cancer. The SCC antigen has been widely used as a serum tumour marker and was reported as a target gene to detect tumour cells in peripheral blood in cervical cancer. In this study, 620 lymph nodes from 14 oesophageal cancer patients were analysed. The results of RT-nested PCR were compared with that of pathological and immunohistochemical examinations. In the test of sensitivity, the RT-nested PCR detected 101of SCC antigen producing cells in 107peripheral blood mononucleocytes and was not found in 43 control lymph nodes. The pathological examination, immunohistochemical examination and the RT-nested PCR detected 36, 45 and 65 nodes respectively. The RT-nested PCR detected statistically more lymph nodes than the pathological or immunohistochemical examination. The sensitivity and specificity seem higher in squamous cell carcinoma cases. The SCC antigen gene is one of the more useful markers for RT-nested PCR to detect occult lymph node metastases of oesophageal cancer. © 2000 Cancer Research Campaign

Recent developments in molecular biology have produced new methods to detect smaller numbers of tumour cells. Many kinds of molecular markers to detect occult tumour cells are being sought (Pelkey et al, 1996;Raj et al, 1998). It is still controversial whether occult metastases found by recent techniques have an impact on patients' prognosis or not. As for oesophageal cancer, Izbicki et al (1997) and Luketich et al (1998) reported that microlymph node metastasis indicates poor prognosis. On the other hand, Glickman et al (1999) do not recommend extensive lymph node sectioning with keratin immunohistochemistry because occult lymph node metastasis was not an independent poor prognostic feature in their study.
The squamous cell carcinoma (SCC) antigen has been widely applied as a serum tumour marker of SCC of the uterine cervix, oesophagus, head and neck, and non-small-cell lung cancer and some other organs (Kato, 1992). Recently, Stenman et al (1997) succeeded in detecting tumour cells in the blood of patients with cervical SCC by RT-nested PCR for SCC antigen gene mRNA.
In this study, we compared the result of the RT-nested PCR for the SCC antigen mRNA with that of the pathological and the immunohistochemical examination for cytokeratins to investigate the usefulness of the SCC antigen gene mRNA expression as a marker to detect micrometastases of oesophageal cancer.

Lymph node and tissue samples
A total of 620 lymph nodes were obtained from 14 patients with oesophageal carcinoma who underwent oesophagectomy at Kyoto University Hospital (Table 1). Twelve patients had SCC, one patient had adenocarcinoma originating from the Barrett's epithelium and one patient had basaloid squamous carcinoma. Three pN0 patients were included in this study. Normal and malignant oesophageal tissues were taken from surgical specimens immediately after resection and stored in liquid nitrogen until the extraction of RNA. Each lymph node was serially numbered and then cut into two pieces. One half of a lymph node was formalin-fixed and subjected to the routine pathological and immunohistochemical examination. The other half was snap-frozen in liquid nitrogen and kept in liquid nitrogen until the extraction of RNA. We used disposable surgical scalpels to cut lymph nodes to avoid crosscontamination. As controls, lymph nodes were obtained from breast cancer, gastric cancer, colon cancer and rectal cancer patients. Each control node proved to be negative for metastasis by pathological examination and each primary tumour was confirmed to be negative for the SCC antigen gene expression by RT-nested PCR.
Detection of lymph node metastasis of oesophageal cancer by RT-nested PCR for SCC antigen gene mRNA

Cell line and normal peripheral blood mononucleocytes
The oesophageal cancer cell line, KYSE 273 (Shimada et al, 1992) that was established from an oesophageal SCC in our laboratory, was used in this study. This cell line produces SCC antigen in its culture medium (Shimada, 1999). Normal peripheral blood mononucleocytes (PBMN) were harvested from the heparinized blood of healthy volunteers using Lymphoprep (Nycomed Pharma, Oslo, Norway) gradient centrifugation according to the manufacturer's instructions.

Serial dilution of carcinoma cells
Serial dilution of KYSE 273 cells with normal PBMN was carried out to confirm the sensitivity of this RT-nested PCR assay. Known numbers of KYSE 273 cells were added to 1 × 10 7 normal PBMN, to give a ratio ranging from 1 : 10 2 to 1 : 10 7 , then total RNA was extracted.

RNA extraction
Total RNA was extracted from cell lines, PBMN, oesophageal tissues or lymph nodes using the acid guanidinium thiocyanate-phenol-chloroform extraction method (Chomczynski and Sacchi, 1987). Extracted RNA was dissolved in DEPC (diethylpyrocarbonate)-treated water and subjected to reverse transcription.

Oligonucleotide primers
Primers were designed from the published sequence of human SCC antigen cDNA (Suminami, 1991). To avoid false-positive results caused by genomic DNA contamination, primers were selected which spanned introns ( Figure 1). The outer primers produce a PCR fragment of 1083 bp. The nested primers produce a PCR fragment of 261 bp. The quality of RNA and cDNA synthesis was ascertained by amplification of glyceraldehyde phosphate dehydrogenase (GAPDH) gene as the internal control. The primer sequences for GAPDH primers were as follows: 5′-TGGTATCGTGGAAGGACTCATGAC-3′ and 5′-ATGCCAGT-GAGCTTCCCGTTCAGC-3′.

RT-nested PCR
Reverse transcription was performed using a First Strand cDNA Synthesis kit (Pharmacia Biotech, Uppsala, Sweden) according to the manufacturer's instructions. The random hexamer was used as the RT primers. After RT, cDNA corresponding to 1 µg of RNA was used as the PCR template. The first-round PCR was performed in 50 µl reaction mix containing 20 mM Tris-HCl (pH 8.4), 50 mM potassium chloride (KCl), 2.0 mM magnesium chloride (MgCl 2 ), 1 µM each of the two primers (A and B), 0.2 mM dNTP and 1 U Taq. polymerase (Gibco-BRL, Grand Island, NY, USA). After initial heating at 94°C for 5 min, 35 cycles of PCR were carried out (94°C for 30 s, 50°C for 30 s and 72°C). For the nested PCR, 3 µl aliquot of the first-round PCR product was used as the template. The nested PCR was carried out in 50 µl reaction mix containing 20 mM Tris-HCl (pH 8.4), 50 mM KCl, 1.5 mM MgCl 2 , 1 µM each of the two primers (C and D), 0.2 mM dNTP and 1 U Taq. polymerase. After initial heating at 94°C for 5 min, 30 cycles of PCR were carried out (94°C for 30 s, 60°C for 30 s and 72°C). For analysis, 10 µl of reaction product was run on a 2% agarose gel followed by SYBR green (Molecular Probe, Eugene, OR, USA) staining. The specificity of the PCR product was confirmed by restriction enzyme digestion pattern (Eae 1) and sequencing. The 261 bp RT-nested PCR products generated from KYSE273 cells were cloned into the pCR II plasmid vector using the TA cloning system (Invitrogen, Carlsbad, CA, USA) and sequenced by ALF autosequencer (Pharmacia Biotech). A Gen Bank-Update new sequences library nucleotide database search demonstrated that the sequence was specific to the SCCA1.
We took special care to avoid cross-contaminations. For example, we used filtertips of micropipettes in every procedure; preparation of PCR reagent and electrophoresis of PCR product were performed in separate places and a negative control, with no template cDNA, was run with every PCR to exclude carry-over contaminations. The results of RT-nested PCR were compared with that of pathological and immunohistochemical examination in a blind fashion.

Immunohistochemistry
Paraffin-embedded sections were stained for epithelial cytokeratins by the streptavidin-peroxidase method. We used anticytokeratin antibody because the antibody for the SCC antigen is not commercially available. The antibody, AE1/AE3 (Dako, Kyoto, Japan), has been used for detection of occult tumour cells in lymph nodes (Greenson et al, 1994;Jeffers et al, 1994) and is known to be negative in normal lymph nodes. Sections were dewaxed, rehydrated and blocked endogenous peroxidase by 0.3% hydrogen peroxide in methanol for 30 min. Antigens were reactivated by microwave treatment. Cell membrane permeability was enhanced by 0.2% Triton X in phosphate-buffered saline (PBS) for 20 min. Non-specific antigen was blocked by 10% normal horse serum for 30 min at room temperature. Monoclonal murine antibody for epithelial cytokeratins, AE1/AE3 (Dako, Kyoto, Japan), in PBS containing 1% bovine serum albumin (BSA) was applied and sections were incubated 4°C overnight. Slides were washed 6 times in PBS and then incubated with biotinylated horse antimouse antibody for 40 min at room temperature. Biotinylated second antibody was detected by Vectastain Elite ABC Kit (Vector Laboratories, Burlingame, CA, USA) according to the manufacturer's instructions. The slides were then counterstained by haematoxylin. A lymph node with metastatic oesophageal SCC was included in all series as the positive control and a negative control run by exclusion of the primary antibody. All slides were evaluated in a blind fashion by two observers and all stained cells were evaluated by a pathologist. We excluded solitary stained cells without nuclear atypism and cell clusters that were located apparently out of the lymph node tissue.

Statistical analysis
Wilcoxon signed-rank test was carried out to examine the difference among the histological examination, the immunohistochemical examination and the RT-nested PCR.

Sensitivity and specificity
Serial dilution experiments showed that our RT-nested PCR assay detected ten KYSE273 cells per 10 7 of PBMN ( Figure 2). We did not detect SCC antigen mRNA expression in 43 of the control lymph nodes from nine patients even when each PCR cycle was increased to 40 cycles (data not shown). Any non-specific band was not detected from RNA templates without RT (Figure 2).

Clinical samples
We detected SCC antigen gene expression in all normal and malignant oesophageal tissues analysed in this study (data not shown). The result from all patients is shown in Table 2. Pathological examination, immunohistochemical examination and RT-nested PCR detected 36, 45 and 65 lymph nodes respectively. Twenty-six lymph nodes were positive only by RT-nested PCR. There were six metastases detected by immunohistochemistry but not by RT-nested PCR. In these six nodes, three were from a basaloid squamous carcinoma patient, two were from an adenocarcinoma patient and one node was from a SCC patient. RT-nested PCR detected significantly more lymph nodes than histological examination (P < 0.0001 by Wilcoxon signed-rank test) and immunohistochemical examination (P = 0.0004). The immunohistochemical examination either detected statistically more nodes than the pathological examination (P = 0.0027).  Suminami et al (1991) In 12 squamous cell carcinoma cases, the RT-nested PCR detected all of pathologically positive nodes and all but one immunohistochemically positive node (Table 3). Both sensitivity and specificity of the RT-nested PCR were higher in SCC cases (98.2% and 99.8% respectively) than those in an adenocarcinoma case (84.6% and 90.9% respectively).
The influence of detected micrometastases on each patient is summarized in Table 4 (International Union Against Cancer, 1997). The RT-nested PCR for the SCC mRNA found micrometastases in 13 cases out of 14 cases whereas immunohistochemical examination detected five cases. In all pN0 patients, only RTnested PCR revealed positive nodes near the primary tumour in each case. In three cases, nos 2, 3 and 14, RT-nested PCR detected tumour dissemination beyond regional lymph nodes (Figure 3) (Japanese Society for Esophageal Diseases, 1992). Especially in case 14, who was pN0 by pathological and immunohistochemical examinations, RT-nested PCR indicated the spread of cancer cells toward the para-aortic lymph node via the perigastric lymph node.
In case 10, a poorly differentiated adenocarcinoma, RT-nested PCR detected eight nodes out of nine pathologically positive nodes and two more positive nodes that were not detected pathologically. In contrast to case 10, in case 7, a basaloid squamous, RT-nested PCR for SCC antigen gene did not detect any positive nodes out of 51 dissected nodes.    The pN classification by the RT-nested PCR advanced in three cases and dropped in a basaloid SCC case. The pM classification advanced in three cases by the RT-nested PCR. Finally, the pTNM stage advanced in five cases and dropped in a basaloid SCC case by the RT-nested PCR.
The average micrometastases were 1.9 nodes a case. There were four cases that had more micrometastases than the average. We are prospectively following the impact of micrometastases on the prognosis.
The SCC antigen was isolated from a metastatic, cervical SCC (Kato and Torigoe, 1977) and has been widely applied as a serum tumour marker of SCC (Kato, 1992). On the other hand, it also exists in normal squamous epithelium (Maruo et al, 1985;Matsuda et al, 1990). The SCC antigen gene was cloned in 1991 and was identified to be a member of ovalbumin family of serine protease inhibitors (Suminami, 1991) and it was revealed to be located on the locus at 18q21.3 and to contain a tandem array of two genes, SCCA1 and SCCA2 (Schneider et al, 1995). These two genes are thought to code for the neutral and acidic forms of the protein respectively. It is thought that the neutral form of the SCC antigen exists in both normal and malignant squamous epithelium. On the other hand, the acidic form is the dominant form of the serum SCC and it is thought to be secreted by malignant cells. However, the sensitivity of the serum SCC antigen is 28.6-37% (Mealy et al, 1996;Nakamura et al, 1998;Quillien et al, 1998) in oesophageal cancer. Therefore the acidic form may not always be produced in oesophageal cancer cells. We think the SCCA1 gene is suitable for the RT-PCR target because it already exists in squamous cells.
Our RT-nested PCR for the SCC antigen gene more efficiently detected micrometastases of oesophageal cancer than the immunohistochemical staining for the cytokeratins (AE1/AE3). It is theoretically possible that the RT-nested PCR overlooks micrometastases when small numbers of cancer cells are included exclusively in the other half of the lymph node. The only lymph node from a SCC case that was negative by RT-nested PCR but positive by immunohistochemical examination (Table 3) may be the case. Though this problem may occur equally with each method, RT-nested PCR found 25 micrometastases that were not detected by immunohistochemical examination. Generally, in such cases where a small number of micrometastatic cells are found, RT-nested PCR which can analyse a lymph node en bloc which has an advantage.
Clinical application of these sensitive but laborious methods is still a problem. Indeed it is not practical to examine every dissected lymph node by RT-PCR. In order to use these methods practically, it may be better to focus some key lymph nodes, for example, recurrent nerve chain (Malassagne et al, 1997) or perigastric nodes or, as Stenman et al (1997) indicated, these methods can be applicable to detect circulating tumour cells. These highly sensitive methods can provide new clues to select patients at risk or to investigate the way of tumour spread.
We detected SCC antigen gene expression in an adenocarcinoma case. This means, although sensitivity in adenocarcinoma seems lower than that in SCC, even adenocarcinomas may express SCC antigen mRNA in the oesophagus. Though the expression of the SCC antigen mRNA in adenocarcinomas has not been fully investigated, some previous reports support our findings. For example, Ueda et al (1984) reported that the SCC antigen (referred to as TA-4) was detected in some adenocarcinomas of the uterine cervix and less frequently in those of the endometrium. Crombach et al (1989) showed elevated cytosolic SCC antigen protein in adenocarcinomas of the uterine cervix in comparison with other gynaecological adenocarcinomas. Takeshima et al (1992) detected mRNA of SCC antigen in the columnar cells of the cervix but not in the endometrium. Considering these findings, combined with our observations, we expect that we can detect occult metastases of an oesophageal adenocarcinoma using SCC antigen gene expression. However, because we have experienced only two cases other than SCCs, further study will be needed for the application of SCC antigen gene expression for detecting oesophageal adenocarcinoma micrometastasis.
On the other hand, in a basaloid squamous carcinoma case, our RT-nested PCR did not detect any metastases. The primary tumour of this case had a small amount of SCC; most of the tumour was basaloid carcinoma cells. The lymph node metastases of this case exclusively consisted of these basaloid carcinoma cells. This may be the reason our RT-nested PCR detected SCC antigen expression in the primary tumour but not in lymph node metastases.
In conclusion, the SCC antigen gene is a useful target to find occult oesophageal SCC cells. The RT-nested PCR for the SCC antigen gene seems more sensitive in SCC cases. This unique characteristic of the SCC antigen gene differs from CEA or cytokeratins. As Stenman et al (1997) suggested in cervical cancer cases, this method will be applicable in other SCCs.