Vulvar cancer is a rare disease, representing about 4% of all female genital tract tumors. Its incidence is currently rising, possibly due to the increasing life span of the female population as well as increasing prevalence of human papillomavirus infections.1 About 200 new cases are diagnosed every year in Poland. Around 40% of patients with vulvar cancer are diagnosed at an advanced stage (stage III or IV). The main reason for this is a delay in seeking medical attention by patients themselves or misdiagnosis. Radical surgery is commonly recognized as the optimal treatment of early and locally advanced cases of vulvar cancer.1
The most common histology of vulvar cancer is squamous cell carcinoma (about 90% of lesions). Vulvar carcinoma spreads primarily by local expansion and via the lymphatic system. Lymphatic mapping begins in the vulva, then drains to the superficial inguinal nodes above the cribriform fascia and finally progresses to the deep left and/or right inguinal or femoral nodes. An important prognostic factor for vulvar cancer, as in all malignancies, is the stage of disease at the time of diagnosis. Of patients with negative pathologic status of groin nodes, approximately 90% have a survival rate of 5 years compared to approximately 30% for those with 3 or more unilateral nodes involved.2
Carbonic anhydrase 9 (CA9, originally named MN) was first described by Pastorekova et al.3 as a 58/54 kDa plasma membrane protein detected in HeLa cervical carcinoma cells with a specific monoclonal antibody (MAb), M75. Expression of the MN protein was correlated with the tumorigenic phenotype of somatic cell hybrids between HeLa cells and normal fibroblasts and was linked with human tumors.4 Its cDNA sequence and intron–exon map were subsequently elucidated.5, 6 MN protein was found to contain a carbonic anhydrase (CA) domain, to bind zinc and to possess CA activity.5 At that time, MN was the ninth member of the CA family and was therefore renamed “CA9”. Later, a cDNA of the renal cell carcinoma antigen G250 was shown to be homologous to CA9 cDNA.7, 8 CAs are ubiquitous metalloenzymes that catalyze a reversible hydration of carbon dioxide to carbonic acid and play a fundamental role in physiologic processes that involve ion transport and acid-base balance, such as respiration, digestion, bone resorption, renal acidification, etc.9
CA9 is a tumor-associated CA isoenzyme present in several types of tumor, such as renal,10, 11 cervical,12, 13, 14 esophageal,15 colorectal,16 lung17 and breast18 carcinomas, but not in the corresponding normal tissues. Normal tissues expressing CA9 identified so far include gastric, intestinal and gallbladder epithelia.19, 20 Weak natural expression is also detectable in the crypt areas of colonic mucosa and in pancreatic tubular epithelium.21 CA9 presence in those normal tissues does not appear to interfere with its value as a tumor marker. Sequence analysis of stomach cDNA showed no difference between normal and tumor tissue–derived cDNAs, indicating that association of CA9 with oncogenesis does not depend on mutations.19
Although the exact significance of CA9 in carcinogenesis and metastasis has not been fully elucidated, experiments indicate its functional significance in the regulation of cell proliferation and/or differentiation and in intercellular adhesion.22, 23, 24 Expression of CA9 is strongly inducible by increased cell density and decreased level of oxygen, i.e., by the conditions that are consistently linked with tumor growth.25, 26, 27 Moreover, inactivating mutations of the VHL tumor-suppressor gene result in abnormal CA9 expression in renal cell carcinomas, apparently via a constitutively activated hypoxic pathway.28
Hypoxia is significantly associated with poor prognosis, aggressive tumor phenotype and resistance to anticancer therapy.29 Expression of CA9 is directly driven by hypoxia-inducible factor-1 (HIF-1), exhibits a clear hypoxia-related pattern of intratumoral distribution and is linked with poor prognosis in several tumor types, as reviewed by Potter and Harris.30 In hypoxic tumor cells, CA9 appears to participate in the maintenance of neutral intracellular pH and in extracellular acidification, thus contributing to the survival and generation of the microenvironment conducive to tumor growth and invasion.31, 32
All these features make CA9 a promising diagnostic and prognostic marker of different solid tumors, including the gynecologic malignancies. We have focused on RT-PCR-based evaluation of CA9 expression in vulvar carcinomas and related regional lymph nodes, with an intention to examine its potential application in the detection of micrometastases.
Material and methods
We have studied material from 20 patients with microscopically confirmed vulvar squamous cancer of general status of I-II (WHO) and clinical stage T1-2, N0-2, M0 in the Maria Skłodowska-Curie Memorial Cancer Center and Institute of Oncology. Only patients with no prior treatment for vulvar cancer or any other malignancies were enrolled. The study was approved by the Independent Ethics Committee of the Cancer Center and the Institute of Oncology.
Lymph node identification
All patients were administered a Tc99m isotope of technetium (activity 1.2 mCi) in the direct vicinity of the vulvar tumor 24 hr before surgery. To visualize the sentinel lymph node, a 2.5% solution of Patentblau V was injected transcutaneously in the direct vicinity of the vulvar tumor 10 min before the skin incision. Inguinal lymph node excision was performed as the first step of the procedure. Sentinel nodes were identified with a hand-held gamma-camera equipped with a radiation meter (Navigator Gamma Positioning System, Radiation Monitoring Devises, Inc., Watertown, MA) in the following sequence of measurements: (i) above the skin of the inguinal area, (ii) after skin incision identifying the sentinel nodes in vivo. After removal of the sentinels (either one or multiple nodes over which we detected the highest isotope accumulation), the radioactivity of the wound itself was measured and noted as the so-called background uptake. If no isotope accumulation was detected within the entire inguinal node area, the nodes stained with Patentblau V dye were removed and referred for histopathologic examination as the sentinels. Surgery was continued with radical excision of the entire group of inguinal lymph nodes and the vulva.
Tissue samples and cultured cells
Tissue samples included 36 specimens of the primary site (17 analyzed by RT-PCR and 19 by immunohistochemistry for CA9 detection), 77 specimens of secondary sites (38 sentinel and 39 inguinal lymph nodes, left and right) and an additional 14 samples of the surgically removed margin. Tissue samples were divided into 2 parts: one part was examined by conventional histology, the second part was stored at –80°C until used in RT-PCR. Cultured cells from the epidermoid vulvar carcinoma cell line A431 (Deutsche Sammlung von Zellkulturen und Mikroorganismen, Braunschweig, Germany) served as a positive RT-PCR control. Lymph nodes evaluated as cancer-negative by both RT-PCR and histopathology were used as negative controls.
Twenty operative samples including vulva, sentinel and inguinal lymph nodes (left and right) were formalin-fixed and paraffin-embedded. Sections (4 μm) were routinely stained by hematoxylin and eosin. Antihuman cytokeratin MAb clone MNF 116 (Dako, Copenhagen, Denmark), which binds keratins 5, 6, 17 and 19, was used for the identification of cells of epithelial origin in lymph node sections. Nineteen primary tumors and some lymph node tissue sections were immunostained by the biotin-streptavidin complex method (Immunotech, Marseille, France) with M75 MAb (diluted 1:50) specific for the N-terminal proteoglycan region of CA9.3, 22, 33 The staining signal was visualized with the DAB Chromogen Kit (Immunotech), and sections were counterstained with hematoxylin. CA9 staining was scaled on the basis of both intensity and extent to 1, 2 or 3, corresponding to weak-focal, moderate and strong-diffuse reaction, respectively.
mRNA extraction and cDNA synthesis
Whole-tissue samples were pulverized on liquid nitrogen using the Microdismembrator II (Braun Biotech, Aylesbury, UK). Total RNA was isolated from approximately 200 mg of each pulverized sample with a NucleoSpin RNA L kit, which includes DNase I (Macherey-Nagel, Duren, Germany). RNA integrity was examined by denaturing agarose gel electrophoresis. Reverse transcription was performed on 4 μg of total RNA. RNA was incubated for 10 min with oligo(dT)12–18 at 65°C. After 10 min at room temperature, reaction reagents were added (SuperScript, Invitrogen, Carlsbad, CA) and incubated at 37°C for 1 hr. cDNA was extracted with phenol-chloroform, precipitated with isopropanol and dissolved in 20 μl of water. The quality of cDNA preparations was controlled by RT-PCR of glyceraldehyde-3-phosphate dehydrogenase.
Each sample (3 μl) was taken from a total 20 μl of cDNA and used as a template in a nested PCR. Specific oligonucleotide primers were designed on the basis of previously published data: outer sense ACTGCTGCTTCTGATGCCTGT, outer antisense AAAGGCGGTGCTGAGGT, nested sense GGGACAAAGAAGGGGATGAC and nested antisense AGTTCTGGGAGCGGCGGGA.11, 34 PCR was performed in a Perkin-Elmer (Foster City, CA) thermocycler (GeneAmp PCR system 240) using Taq polymerase in a 25 μl mix containing PCR buffer with (NH4)2SO4 (MBI Fermentas, Vilnius, Lithuania) and 10 pmol of each primer. Conditions for the first round of PCR were as follows: 35 cycles at 94°C for 1 min, 57°C for 2 min and 72°C for 1 min. After the first round of PCR, 0.8 μl of the mixture served as a template for the second round, consisting of 30 cycles at 94°C for 1 min, 58°C for 2 min and 72°C for 1 min. Both PCRs were preceded by 4 min at 95°C and followed by 7 min at 72°C. Nested primers produced a PCR fragment of 187 bp. For each nested PCR, a set of control samples included one positive control, i.e., cDNA from the A431 cell line, and 2 negative controls, represented by a sample with no template and a sample with placental DNA. RT-PCR products were electrophoresed on a 2% agarose gel.
We first examined primary tissue specimens (n = 19) for the presence of CA9 protein by immunohistochemistry. All 19 specimens were positive, with the staining intensity ranging from weak to strong and the distribution of signal extending from focal to diffuse (Table I). Staining was mostly membranous (Fig. 1a,b), with only a few cases of cytoplasmic positivity (Fig. 1c). Subsequent immunohistochemical inspection of several lymph nodes in some cases revealed scattered cells with cytoplasmic positivity for CA9. In one specimen, a cluster of carcinoma cells with strong membrane CA9 staining was seen (data not shown). Some CA9-positive specimens showed positivity for the cytokeratins, suggesting that the lymph nodes did contain carcinoma cells (Table I). These findings indicated that detection of CA9-expressing cells in lymph node specimens is feasible. However, the immunohistochemical method is impractical for detection of micrometastases because it requires systematic examination of large numbers of slides corresponding to whole lymph nodes.
Table I. Results of Histopathologic and RT-PCR Examinations for CA9 of Tissue Specimens from 20 Vulvar Cancer Patients
With the purpose of developing a sensitive and effective CA9 detection method suitable for routine tissue examination, we elaborated a nested RT-PCR that can be performed on RNA isolated from an entire tissue or its representative portion. The outer PCR antisense primer was designed to span the splice junction between the first and second exons of the CA9 gene, and this allowed RNA-specific evaluation of CA9 gene transcription. This approach eliminated all possible RT-PCR false-positives that could arise from trace amounts of genomic DNA. Indeed, we did not observe any positive signal from PCR of the CA9 gene performed on placental DNA as a template. There are no published data on CA9 pseudogene(s) existing, but this control excluded putative CA9 pseudogene detection. Moreover, using this nested RT-PCR, all 17 primary tumor specimens as well as the A431 vulvar cancer cell line showed a specific positive signal (Table I). This was not surprising given the capability of detecting CA9 with immunohistochemistry, which is much less sensitive than RT-PCR.
However, lymph nodes might contain micrometastases of a minimal number of CA9-positive cancer cells; and in such cases, sensitivity of the detection method would be a crucial parameter. To evaluate this parameter in our setting, we determined the sensitivity of the nested RT-PCR assay. A lymph node that was negative by histopathology, cytokeratin immunostaining and RT PCR was used as a background. Seven equal amounts of RNA (elutions of 500 μl, each obtained from a 200 mg tissue portion) were extracted from this node. Decreasing amounts of RNA isolated from A431 cells corresponding to 105, 104, 103, 102, 101 and 100 cells were separately added to those 500 μl RNA portions. Then, 4 μg aliquots of all RNA dilutions (including the one with RNase-free water instead of A431 cells, “zero”) were reverse-transcribed and used as templates in nested PCR for CA9 detection. The results shown in Figure 2 indicate that our RT-PCR assay is highly sensitive.
Representative RT-PCR results of all samples from one patient are shown in Figure 3. In total, we tested 77 lymph nodes from 20 vulvar cancer patients (i.e., 38 sentinel and 39 inguinal lymph node samples). In 58 samples, the results obtained by RT-PCR for CA9 fully correlated with those obtained by standard hematoxylin-eosin staining, giving 49 negative and 9 positive results. The additional 19 positive PCR results accounted for 28% of the total number of 68 histologically negative lymph nodes. No specimens were positive by pathologic examination and negative by RT-PCR; thus, RT-PCR gave no false-negative results. Among 48 RT-PCR-negative lymph node specimens, there was only one cytokeratin-positive case, but this discrepancy could be due to different tissue portions examined by these methods (Table I). Comparison of immunohistochemistry with MNF 116 antibody and RT-PCR data analyzed simultaneously by both methods revealed 77% concordance of the results. The remaining 23% of specimens were negative by immunohistochemistry and positive by RT-PCR. There were no false-negatives for RT-PCR compared to immunohistochemistry for cytokeratins.
In 38 groins, one positive inguinal lymph node with a negative sentinel node was observed in the same groin by RT-PCR. By conventional examination, these 2 lymph nodes were evaluated as cancer-negative. Interestingly, 10 of 14 controls from the surgical margin of vulvar skin were RT-PCR-positive (data not shown). The histopathologic, RT-PCR and staining results of all of the specimens examined are summarized in Table I.
Detection of malignant spread at the earliest stage, when metastatic disease is clinically indistinguishable from localized disease, is highly desirable. RT-PCR-based assays have several advantages over traditional methods in detecting metastases. Above all, they are highly sensitive at detecting tumor cells. Since metastasis cells may be well under way long before the tumor is detectable by standard techniques, staging of the lymph nodes by RT-PCR might be a good predictor of recurrence.35, 36 In several reports, RT-PCR has been shown to detect cancer- (tissue-) related marker expression in a subset of lymph nodes that were immunohistochemically negative. Moreover, patients whose lymph nodes were positive by RT-PCR but negative by conventional histopathologic analyses often developed metastatic disease 5 years after the resection (see Raj et al.35 and Ghossein et al.36 for references).
RT-PCR assays are based on the ability to detect expression of cancer- (or tissue-) specific mRNA in the total RNA isolated from tissue specimens. The rationale is that, in cancer patients, the secondary site contains the metastatic cells expressing the markers specific for the primary site tumor tissue. The marker genes must not be expressed in the normal secondary site tissue. New molecular markers may help to identify metastases at earlier stages and have a significant impact on surgical, radiation and systemic therapies for a given patient in the future. At present, the clinical utility of the RT-PCR assay as a tool for the prognosis and management of cancer patients remains an area of active investigation in the early stage of development.35, 36 Many studies are aimed at evaluating quantitative RT-PCR approaches and improvements of the technical parameters of the assay, but selection of a suitable marker(s) remains the most critical issue.
Our study was initiated to develop a sensitive and clinically applicable method to detect micrometastases by examining the lymph nodes for the presence of CA9 transcripts. Immunohistochemical detection of CA9 has been correlated with poor prognosis in several types of carcinoma.30 In addition, CA9 transcription, detected by either semiquantitative or quantitative RT-PCR, has been associated with malignant breast tissues and treatment outcome.18, 37 Therefore, CA9 was chosen as a cancer-related marker for our study. We confirmed its expression in primary vulvar carcinomas by both immunohistochemistry and nested RT-PCR. This stimulated us to examine the tissue samples of the secondary sites, which represented the clinically important pathways of dissemination of the vulvar cancer. We expected to detect embolized tumor cells by nested RT-PCR in the regional lymph nodes. In accordance with this expectation, a fraction of the lymph node specimens obtained from vulvar cancer patients showed positive RT-PCR results, which implies that these solid tumors were shedding cancerous cells expressing CA9 into the patients' lymphatic system.
The results of our preliminary survey, presented here, confirm that our RT-PCR is a sensitive tool for the detection of micrometastases and strongly support the value of the CA9 gene product as a useful marker in the diagnosis and molecular staging of vulvar carcinoma. Using this assay, we detected CA9 mRNA in all pathology-positive specimens tested. Histopathologic examination underestimates the number of patients with metastases. We have confirmed an expected higher sensitivity of the RT-PCR method in examining lymph node specimens; i.e., 28% of negative samples by pathologic examination produced a positive signal in RT-PCR. This number decreased to 23% when RT-PCR for CA9 was compared to immunohistochemistry with MNF 116 specific for cytokeratins. Moreover, there were no false-negatives with the RT-PCR method. Only one in 38 groins had a positive inguinal lymph node in the presence of a negative sentinel node (only by RT-PCR).
Identifying the sentinel node to decrease the morbidity associated with groin dissection has become a topic of interest in a gynecologic oncology. The sentinel node is defined as the first node in the lymphatic flow from the suspected lesion.38 It is used as a representative of the entire lymphatic chain and is expected to be the first site of a metastatic disease. If the sentinel node is negative for metastatic disease, the remaining nodes should also be free of tumor cells.39
Our results of RT-PCR examination of the sentinel and inguinal lymph nodes from 37 groins of vulvar cancer patients are consistent with those findings. Yet, the result for one groin was inconsistent with our expectations. The major part of each lymph node biopsy was studied by the pathologist since so far clinical diagnosis has been based on histopathologic examination. Therefore, different and unequal parts of each biopsy were analyzed by routine methods and by PCR; RT-PCR analysis could have been performed on a cancer cell-free part of the sentinel lymph node biopsy. This may provide an explanation of the inconsistency observed. Nevertheless, metastases bypassing the sentinel node and populating the inguinal node may occur naturally, as reported by de Hullu et al.;41 and our case might be one of these.
The concept of the sentinel node has been integrated in the routine clinical staging of melanoma and breast cancer patients.41, 42 Vulvar cancer has good access to regional lymph nodes; thus, sentinel node biopsy could be applicable. One of the potential benefits of sentinel node biopsy is omission of lymph node dissection. In some patients, modifications in surgical technique could be possible, allowing less radical excisions and thus decreasing morbidity without compromising survival. Finding a reliable method of diagnosing nodes is required before inguinal lymph node–sparing treatment of vulvar cancers could be accepted in the future. Until then, complete inguinofemoral lymphadenectomy will be performed.
Complications in wound healing may be the cause of delay of radiotherapy in patients with positive (metastatic) regional nodes who require adjuvant treatment, and that reduces 5-year survival rates.43 Therefore, in vulvar cancer patients, it is crucial to do the reliable regional lymph node assessment. On the one hand, it would limit disability due to the radicalism of therapy and, on the other hand would not, if indicated, prolong healing or prevent postoperative irradiation as there is an important role for an adjuvant therapy in cases of lymph node invasion. Present practice is mostly based on histopathologic evaluation of insufficient sensitivity to reliably detect micrometastases, whose presence has been demonstrated by cytokeratin immunohistochemistry as a significant prognostic factor in node-negative squamous cell carcinoma of the vulva.44 In the future, RT-PCR technology, which is even more sensitive as well as time- and cost-effective than immunohistochemistry, will likely serve as a useful supplement to clinical decision making in the accurate definition of regional lymph node status.
In conclusion, our data indicate that CA9 is expressed in primary vulvar tumors as well as in derived metastatic cells and suggest that detection of CA9-positive cells in the lymph nodes by nested RT-PCR represents a promising approach to improved management of vulvar cancer patients. We have shown a potential value and sensitivity of CA9 transcript detection in the molecular staging of vulvar carcinoma. Nevertheless, this CA9-specific RT-PCR approach needs further validation with a larger number of cases including clinical follow-up. The prognostic value of CA9-positive micrometastasis in the sentinel node in patients operated for vulvar cancer can then be evaluated. Moreover, examination of local recurrences in patients with PCR-positive CA9-expressing surgical margins might have an impact on clinical procedures. Our study has accomplished a first promising step toward these applications.
We are very grateful to Drs. R. Nowak and M. Chechlińska for helpful discussions and constructive comments on the manuscript.