Molecular evidence of a common clonal origin and subsequent divergent clonal evolution in vulval intraepithelial neoplasia, vulval squamous cell carcinoma and lymph node metastases
Article first published online: 1 APR 2002
Copyright © 2002 Wiley-Liss, Inc.
International Journal of Cancer
Volume 99, Issue 4, pages 549–554, 1 June 2002
How to Cite
Rosenthal, A. N., Ryan, A., Hopster, D. and Jacobs, I. J. (2002), Molecular evidence of a common clonal origin and subsequent divergent clonal evolution in vulval intraepithelial neoplasia, vulval squamous cell carcinoma and lymph node metastases. Int. J. Cancer, 99: 549–554. doi: 10.1002/ijc.10362
- Issue published online: 30 APR 2002
- Article first published online: 1 APR 2002
- Manuscript Accepted: 25 JAN 2002
- Manuscript Revised: 24 JAN 2002
- Manuscript Received: 29 NOV 2001
- Gynaecology Cancer Research Fund
- Stroyberg Vagn Jensen Foundation
- vulval intraepithelial neoplasia;
- vulval cancer;
- loss of heterozygosity;
- X-chromosome inactivation;
- vulval squamous cell carcinoma
VIN is thought to be the precursor of some VSCCs because it is monoclonal, frequently occurs contiguously with VSCC and shares similar risk factors with a subgroup of VSCC. There has been no conclusive molecular evidence supporting this assumption. We performed X-chromosome inactivation analysis on 9 cases of lone VIN, 10 cases of VSCC and associated contiguous VIN and 11 cases of VSCC and associated noncontiguous VIN. Eight of the 9 cases of lone VIN appeared to be monoclonal. All 7 informative and monoclonal cases of VIN with contiguous VSCC and 6/9 informative cases of VIN with noncontiguous VSCC showed patterns of X-chromosome inactivation consistent with a common monoclonal origin for both VIN and VSCC. Two of the 9 cases of VIN with noncontiguous VSCC showed X-chromosome inactivation patterns consistent with a separate clonal origin. We performed LOH analysis at 6 chromosomal loci on these samples and 7 cases with lymph node metastases. Identical losses occurred 7 times in VIN and contiguous VSCC (random probability 1.2 × 10–9), twice in VIN and noncontiguous VSCC (random probability 1.5 × 10–3) and 3 times in VSCC and associated metastases (random probability 1.8 × 10–5). Some losses occurring in VSCC did not appear in the contiguous VIN or associated metastases and vice versa. These data provide molecular evidence that VIN is the precursor of VIN-associated VSCC, that multifocal disease may arise via either different clones or a single clone and that continued divergent clonal evolution may occur in vulval neoplasia. © 2002 Wiley-Liss, Inc.
VIN has long been thought to be a premalignant condition that may progress to invasive VSCC. This assumption is based on observations that VIN frequently occurs contiguously with VSCC,1 that VIN and a subgroup of VSCC are associated with similar risk factors (smoking,2 immunosuppression3 and HPV infection2, 4) and that VIN and VSCC are monoclonal neoplastic conditions.5, 6 Confirmation of this hypothesis is important for gaining insight into the process of vulval carcinogenesis. Furthermore, identifying molecular markers for risk of progression of VIN to VSCC is possible only if molecular evidence is compatible with a common monoclonal origin of the 2 conditions. We first wanted to confirm that VIN is a monoclonal condition. We then wanted to document the patterns of LOH events and X-chromosome inactivation in VSCC samples and contiguous and noncontiguous VIN, to examine whether VIN and VSCC could have arisen from the same clone. Because we had already studied the frequency of LOH at 6 chromosomal loci in lone VIN and VSCC,7 it was possible to calculate the odds of identical losses occurring by chance in both VSCC and associated VIN. Finally, as primary cancers and their metastases are thought to have the same clonal origin, we compared the number of identical losses occurring in primaries and their metastases with the number of identical losses occurring in VSCC and associated VIN.
MATERIAL AND METHODS
Samples from patients with VIN and VSCC diagnosed between 1989 and 1997 at St. Bartholomew's and the Royal London Hospitals were identified. Suitable samples containing both normal and neoplastic tissue were as follows: 9 cases of VIN alone, 7 cases of VSCC with lymph node metastases, 10 cases of VSCC with directly contiguous VIN (2 of which also had concurrent noncontiguous VSCC and 3 of which also had concurrent noncontiguous VIN) and 8 cases of VSCC with concurrent noncontiguous VIN. Of the 18 cases of VSCC with concurrent VIN, 13 were stage I, 2 were stage II and 2 were stage III; 1 had inadequate information available for accurate staging. The relevant paraffin-embedded tissue samples were serially sectioned. One 4 μ section was mounted, stained with HE, covered and used as a reference slide. One 10 μ section was also stained with HE but left uncovered for microdissection.
Microdissection of tissue samples
The uncovered 10 μ section was mounted on a dissecting microscope and compared with the reference slide. Tissue containing >70% VSCC, VIN or 100% normal cells was microdissected. The tissue was placed in 100 μl of 10% Chelex chelating resin (Sigma, St. Louis, MO) in distilled water.
Extraction of DNA
One ml of 20 mg/ml Proteinase K (Boehringer Mannheim, Lewes, UK) was added to the tube, incubated at 56°C for 30 min and boiled for 8 min. Samples were centrifuged at 10,000g for 10 min to pellet any remaining debris. One to 5 ml of the supernatant was used directly in the polymerase chain reaction (PCR) for LOH analysis.
Amplification of polymorphic microsatellite markers
PCR was performed in a volume of 20 μl, containing approximately 20–100 ng DNA, 0.2 mM deoxyribonucleotide triphosphates, 0.25 U Taq supreme DNA polymerase (Hellena Biosciences, Sunderland, UK), 1 × buffer (supplied with enzyme), 1.5 pmol forward primer, 2 pmol reverse primer and 0.5 pmol 32P-labelled forward primer. PCR was performed in a Touchdown Thermal Cycler (Hybaid, Ashford, UK). Chromosomal loci, primer pairs, magnesium concentrations and annealing temperatures are shown in Table I. All primers yielded products of <200 bp, making them suitable for use with archival material. PCR conditions were as follows: DNA denaturation for 5 min at 95°C, then hot start at 85°C, followed by 35 cycles of denaturation at 95°C for 40 sec (ramped at 1°C/sec) and annealing at the temperature given in Table I for 30 sec. This was followed by a final extension step at 72°C for 5 min.
|Chromosomal location||Primer||Primer sequence||Annealing temperature (°C)||Mg2+ (mM)|
|Xq11.2-12||AR-M||GTCGCGAGCGTATTTTTCGGC||Variable (see Material and Methods)||3.5|
X-chromosome inactivation assay
In adult females, 1 of the 2 X chromosomes is inactivated by DNA methylation. This inactivation is inherited through mitosis so that a clone of cells, all derived from an original progenitor cell, will have the same X-chromosome inactivated. Bisulphite chemical modification of DNA converts unmethylated cytosine to uracil. It is therefore possible to determine clonality by selectively amplifying a polymorphic region of the X-linked androgen receptor gene, using primers designed to bind to unmethylated sequences following chemical modification. Only samples that are heterozygous for the androgen receptor trinucleotide repeat polymorphism can be analysed for X-chromosome inactivation. To confirm heterozygosity at this locus, we performed LOH analysis at the androgen receptor locus using AR-W primers designed for amplification of the unmodified androgen receptor locus. PCR conditions for these were as above, with magnesium concentration, annealing temperature and sequence given in Table I. To assess X-chromosome inactivation, DNA was chemically modified8 to change unmethylated cytosine bases to uracil. Approximately 100 ng DNA were denatured in 0.3 M NaOH in a volume of 20 μl for 15 min at 37°C before addition of 120 μl 3.6 M sodium bisulphite (pH 5.0) and 0.6 mM hydroquinone (all freshly prepared). Samples were covered with 100 μl mineral oil and underwent 50 cycles of 95°C for 30 sec and 50°C for 15 min in the Touchdown Thermal Cycler. Samples were removed from below the mineral oil, desalted using the Wizard DNA clean-up system kit (Promega, Southampton, UK), desulphonated by addition of 5 M NaOH to 0.3 M and incubated at room temperature for 5 min. Samples were neutralised with glacial acetic acid and the DNA was precipitated with 2 volumes ethanol and 0.1 volume 3 M sodium acetate (pH 5.2) in the presence of yeast tRNA and 0.01 M MgCl2. DNA was resuspended in 20 μl ddH2O.
Modified DNA (4 μl) was used in a 20 μl PCR to amplify the androgen receptor locus using AR-M primers (Table I), designed to amplify the methylated androgen receptor sequence only. Touchdown PCR was performed as follows: 97°C for 5 min, 85°C hot start, 2 cycles of 95°C for 45 sec, 55°C for 45 sec, 68°C for 45 sec; the annealing temperature was subsequently decreased from 55°C to 43°C in 2°C increments for 2 cycles at each temperature, followed by 25 cycles with annealing temperature at 41°C and final extension at 68°C for 10 min.
Separation, visualisation and interpretation of PCR products
PCR products were electrophoresed on a 5% denaturing polyacrylamide gel with subsequent autoradiography. LOH was scored as loss of or >50% reduction in intensity of one of the alleles from the neoplastic tissue relative to the normal tissue of the same patient.9 All autoradiographs were read by 2 individuals (ANR, AR), who were blinded as to the histologic type of the samples. LOH autoradiographs from these samples were shown in our previous report.7 Similarly, samples exhibiting >50% reduction in the intensity ratio of the VIN- or VSCC-modified androgen receptor alleles compared to unmodified alleles were classed as having skewed X-chromosome inactivation, indicative of a monoclonal origin. Examples are shown in Figure 1. The presence of alleles in the VIN or VSCC of sizes not found in the normal tissue of that patient indicated MSI; therefore, these samples could not be interpreted for the presence of LOH.
Probabilities for VIN and concurrent VSCC samples losing the same allele at the same locus were calculated as follows: (frequency of LOH in VIN) × (frequency of LOH in VSCC) × 0.5. These LOH frequencies were known from our previous study.7 Where identical losses occurred at more than 1 locus in a sample pair, the odds of this occurring were calculated as the product of the probabilities for loss at each locus. The total probability of observing common losses in the different sample groups was calculated as the product of the probabilities of each sample in that group. This method of probability analysis is standard in investigations of this type.10 The frequency of identical losses in primary VSCC and metastases was compared to that in VIN and contiguous VSCC using Fisher's exact test.
DNA was successfully extracted and amplified from 100% of VIN and VSCC samples for LOH analysis. It was not possible to amplify modified DNA from 1 case of noncontiguous VIN for X-chromosome inactivation analysis. In 2 cases, the androgen receptor locus was noninformative, so X-chromosome inactivation analysis could not be performed.
Eight of the 9 cases of VIN not associated with VSCC had X-chromosome inactivation patterns consistent with the lesions being monoclonal. The patterns of LOH and X-chromosome inactivation in VSCC with contiguous VIN are shown in Table II. In 7/9 informative cases, X-chromosome inactivation patterns were consistent with the 2 conditions having arisen from the same clone. In the remaining 2 cases (cases 2 and 66), samples 66VIN, 66VSCC and 2VSCC appeared to be polyclonal. Loss of the same allele at the same locus occurred a total of 7 times in 5 cases. The probability of this happening by chance, based on the rates of LOH at these loci in lone VIN and VSCC,7 was 1.2 × 10–9.
|Sample number||X chromosome Inactivation||Chromosomal location|
|94V||MC||R||R||L upper||L upper||—||L|
|94C||MC||R||R||L upper||L upper||—||R|
|112V||MC||L upper||—||L lower||L||R||R|
|112C||MC||L upper||—||L lower||R||R||R|
In the 2 cases with both contiguous and noncontiguous VSCC (cases 63 and 104), X-chromosome inactivation patterns were consistent with noncontiguous VSCC having arisen from the same clone as the VIN and contiguous VSCC. However, in both cases, LOH events shared by VIN and contiguous VSCC had not occurred in the noncontiguous VSCC.
Patterns of LOH in primary VSCC and metastases are shown in Table III. Loss of the same allele at the same locus occurred a total of 3 times in 2 cases. The probability of this happening by chance was 1.8 × 10–5. When the frequency of shared losses in primary VSCC and metastases was compared to that in VIN and contiguous VSCC, no difference was found (Fisher's exact test, p > 0.05). Ten of the 32 informative pairs of events in primary VSCC and metastases showed losses in the VSCC which did not occur in the lymph node sample or vice versa.
|Sample number||Chromosomal location|
|67C||—||R||L||L||L upper||L lower|
|67LN||—||R||NA||R||L upper||L lower|
Patterns of LOH and X-chromosome inactivation in VSCC with noncontiguous VIN are shown in Table IV. X-chromosome inactivation patterns were consistent with the 2 conditions having arisen from the same clone in 6/9 informative cases and different clones in 2 cases. One case of VIN (sample 66V2) appeared to be polyclonal. Loss of the same allele at the same locus occurred a total of 2 times in 2 cases. The probability of this happening by chance was 1.5 × 10–3.
|Sample number||X-chromosome inactivation||Chromosomal location|
We studied patterns of LOH and X-chromosome inactivation in VSCC and contiguous and noncontiguous VIN to investigate their clonal origin. These techniques have previously been used to demonstrate a unifocal origin in metastatic epithelial ovarian cancer.11 We first wanted to confirm reports that VIN and VSCC are monoclonal conditions.5, 6 In 8/9 samples of VIN not associated with VSCC and in 16/18 informative samples of VSCC associated with VIN, we found this to be the case. Both ourselves7 and others9, 12, 13 have observed frequent LOH events in VIN and VSCC. This also implies that these lesions are clonal neoplasms as the likelihood of sufficient cells in a polyclonal epithelium independently losing the same allele at the same locus for LOH to be demonstrable is extremely low.
We analysed VSCC and contiguous and noncontiguous VIN for loss of loci on 6 chromosomes. The only previous similar report examined LOH in 3 different microdissected areas of VIN and VSCC from each of 2 cases.14 Three common losses were observed across 7 loci (2 in 1 case and 1 in the other case), implying that VIN and VSCC may have arisen from the same clone. However, in 1 case, 2 areas of VIN had lost a locus not lost in any of the areas of VSCC, suggesting that further losses can occur in VIN after the point at which a subclone has become malignant. This study did not include X-chromosome inactivation analysis, a technique that provides further evidence of the clonal relationship between 2 lesions. We therefore performed a larger study, which would also enable us to establish whether noncontiguous VIN was clonally related to VSCC. It was possible to calculate the probability of VSCC and associated VIN losing the same allele at the same locus by chance because we had previously defined the LOH frequencies at these loci in VIN and VSCC occurring independently.7
We found that 7/9 cases of VIN and contiguous VSCC had identical patterns of X-chromosome inactivation consistent with a monoclonal origin of the 2 lesions. In addition, identical chromosomal losses occurred in these contiguous lesions 7 times (Table II). The probability of this occurring by chance is extremely small. Indeed, such shared events occurred as often in VSCC with contiguous VIN as in VSCC with lymph node metastases. These data strongly support a monoclonal origin of VIN and contiguous VSCC, implying that VIN is the preinvasive lesion of VIN-associated VSCC.
Whilst some VSCC and noncontiguous VIN had clearly arisen from separate clones (Table IV, cases 30 and 47), 2 cases (Table IV, cases 25 and 112) shared both LOH and X-chromosome inactivation events. A report using multiple LOH markers found that multifocal CIN underwent identical losses, suggesting that apparently topographically distinct lesions had arisen from the same clone.10 However, another study using X-chromosome inactivation found a different clonal origin for 2 topographically distinct lesions.15 These apparently contradictory observations are similar to our own. There are 3 possible explanations for these findings.
Firstly, both our own study and the studies on CIN10, 15 were performed retrospectively on archival tissue. Consequently, it is possible that apparently separate lesions were actually an artifact of cross-cutting a single lesion with tortuous borders so that normal epithelium appeared between different areas of a single neoplastic lesion. This is unlikely as all of the noncontiguous VIN samples appearing to have a common clonal origin with VSCC were obtained from different histopathologic blocks, increasing the likelihood that they were biopsies of anatomically distinct lesions. It is, however, possible that both biopsies were taken from a single large lesion containing both VIN and VSCC. Unfortunately, this could not be clarified by case note review.
Secondly, as CIN can regress spontaneously,16 it is possible that what starts out as a single VIN lesion may become anatomically distinct lesions if areas within the lesion revert to normal, forming bridges of normal epithelium between areas of neoplasia. This is unlikely because only a specific type of VIN, typically present in pregnancy (so-called bowenoid papulosis) is thought to regress spontaneously.17, 18
Thirdly, it is possible that, as suggested in the study on CIN,10 a process of “epithelial spread” occurs. If this is correct, it implies that cells from the neoplastic clone acquire the ability to migrate within the epithelium. If this is the case, then it appears that multifocal VIN and VSCC arise either from different clones or from a process of epithelial spread from a single clone. This theory is consistent with studies in CIN10, 15 and with evidence in head-and-neck19, 20 and oesophageal21 cancers. If the epithelial spread theory is correct, one might expect women with multifocal VIN to be at high risk of progression to invasive disease as the ability of a neoplastic cell to migrate between normal cells is a feature of malignancy. Three studies describe multifocality in VIN. In the first of these, large confluent lesions were classified as multifocal if they involved different regions of the vulva.22 This is not truly multifocal disease (multiple, discrete, nonconfluent lesions separated by colposcopically normal vulval skin) and, therefore, not relevant to assessing the risk of invasion in women with anatomically distinct lesions. In the second, multifocality was not found to be a risk factor for invasive disease, but neither was it defined; however, multifocal disease was present in 82% of 69 patients.23 In a third study, 93% of 27 patients had true multifocal disease but no correlation with invasion was performed.24 If true multifocal VIN is as common as these last 2 studies suggest, then it is the norm rather than the exception. Given that our data suggest 2 possible aetiologies in multifocal VIN (separate clonal proliferations vs. epithelial spread of a single clonal proliferation), performing clonality assays may prove to be useful in predicting which women may develop invasive disease. If multifocal VIN from a patient appears to have a common clonal origin, then we speculate that the patient could be at higher risk than a patient with multifocal VIN of separate clonal origins.
The fact that identical losses in VIN and contiguous VSCC were not found in noncontiguous VSCC from the same patients (Table II, cases 63 and 104) suggests that the noncontiguous VSCC arose from a separate clone. This is because the chance of both the VIN and contiguous VSCC losing the same allele subsequent to the VIN giving rise to both contiguous and noncontiguous VSCC is extremely small. Identical X-chromosome inactivation patterns in contiguous and noncontiguous VSCC from these cases is not evidence against a separate clonal origin for these neoplasms as this observation would occur by chance in 50% of cases.
Two samples of VIN and 2 samples of VSCC appeared to be polyclonal (samples 2C, 66V, 66C and 80V), contrasting with the majority, which were monoclonal. It is interesting to note that 1/8 VIN samples in a previous study appeared to be polyclonal.5 One explanation could be that polyclonal samples were made up of 2 different patches that had become confluent but had originated from separate clones, each with a different X-chromosome inactivated. This explanation is unlikely as these samples would have to contain approximately equal proportions of DNA from each patch in order to give amplification bands of equal intensity. The authors of the previous study5 suggested that polyclonal status could result from numerical chromosomal abnormalities, MSI affecting methylation status or contamination with normal tissue. It is unlikely that the latter is true as some of the polyclonal lesions exhibited LOH (samples 66V and 80V), suggesting that they were actually monoclonal. Given this observation and the fact that none of our samples exhibited MSI at the androgen receptor locus, we favour a numerical chromosomal abnormality in the neoplastic tissue as the explanation (i.e., duplication of the X chromosome with inactivation of only 1 duplicate). Indeed, polyploidy has been reported in vulval cancer cell lines.25
We have previously reported that VIN with associated VSCC undergoes LOH more frequently than VSCC.7 This implies that some of the losses observed in VIN occurred after the point at which a subclone of that VIN has acquired the malignant phenotype. This was observed in 1 case in the previously described study.14 We observed the same phenomenon in 5/7 cases of clonally related VSCC and VIN and 3/7 cases of VSCC with lymph node metastasis. This implies that VIN, VSCC and metastases can follow clonally divergent paths after the point at which VIN progresses to invasion or VSCC metastasises. Such a model is speculative; but if correct, it has research implications. Firstly, it may be that this model is appropriate for describing the development of epithelial malignancies at other sites known to involve a recognised premalignant phase. Secondly, retrospective analysis of preinvasive lesions after the point at which invasion has occurred may identify spurious markers for progression, which occurred subsequent to, rather than prior to, invasion. The evidence suggests that this model of continual divergent clonal evolution occurs in the premalignant lesion Barrett's oesophagus and its invasive sequelae.21, 26
In conclusion, we have corroborated previous work suggesting that VIN and VSCC are monoclonal lesions. We have also shown that it is highly likely that VIN and contiguous VSCC arise from the same clone, supporting histopathologic and epidemiologic evidence that VIN is the preinvasive stage of VIN-associated VSCC. We also have preliminary evidence that multifocal VIN and VSCC in the same patient can arise from different clones; however, in some cases, apparently topographically distinct lesions appear to share the same clonal origin as invasive disease. This may indicate that some VIN cells can acquire the ability to migrate between normal epithelial cells and set up a separate but clonally related lesion. If this is the case, then clonality assays might prove useful in predicting which women with multifocal VIN will progress to invasive disease. Finally, it appears that VIN and VSCC may follow divergent paths of clonal evolution, after the point at which invasion or metastasis has occurred. These findings support those in intraepithelial neoplasia at other sites and imply that cross-sectional studies of coexisting premalignant and malignant lesions may produce spurious markers for disease progression.
ANR was supported by the Gynaecology Cancer Research Fund and AR was supported by the Stroyberg Vagn Jensen Foundation. We thank Mr A. Brown and his staff for cutting and mounting the tissue slides, Mr S. Jones for providing the list of patients with VIN and VSCC and Ms J. Thomas for assisting with the statistical analysis.
- 8Analysis of androgen receptor DNA reveals the independent clonal origins of uterine leiomyomata and the secondary nature of cytogenetic aberrations in the development of leiomyomata. Genes Chromosomes Cancer 1994;11: 1–6., , , , , , , .