Aims—To determine by fluorescence in situ hybridisation (FISH) whether deletion of D17S34, a subtelomeric probe for 17p, occurs in invasive squamous carcinoma of the cervix, and to determine the extent of such loss by analysis of the p53 and HER2/NEU genes.
Methods—Fourteen invasive squamous cell carcinomas of the cervix were investigated by FISH for D17S34, p53, and HER2/NEU. Dual hybridisation of each probe with the chromosome 17 α satellite (D17Z1) probe was performed on paraffin wax embedded sections, and the fluorescence ratios of the paired signals were determined. Broad spectrum human papillomavirus (HPV) typing by ISH and GP5+/6+ polymerase chain reaction was also performed.
Results—Twelve tumours were HPV positive, nine with HPV-16, and one each with types 18, 31, and 39. Loss of D17S34 was identified in four tumours, one of which was HPV negative. In one tumour, D17S34 loss was accompanied by loss of p53 only, suggesting that deletion was limited to the p arm. A second tumour showed simultaneous losses of all probes, indicative of whole chromosome 17 loss during tumour growth. The two remaining specimens showed loss of D17S34 only, diffuse in one, and localised within the tumour in the other. Aberrations of p53 or HER2/NEU were not seen independently of D17S34 loss, and loss did not correlate with HPV presence or type.
Conclusions—These data show that D17S34 loss is prevalent, marking 28% of the invasive squamous carcinomas in this study. The observed intratumoral heterogeneity indicates that, at least in some cases, this loss occurs after invasion and is therefore a late event in the path of cervical carcinogenesis.
Statistics from Altmetric.com
Human papillomavirus (HPV) infection has been widely implicated in cervical carcinogenesis. However, current evidence suggests that HPV infection is an early event and that other abnormalities, whether induced directly by HPV or indirectly by other means, are required for biological transformation.1
Cytogenetic analysis of cervical tumours has shown that chromosomes 1, 3, 11, and 17 are commonly abnormal.2 Abnormalities of the short (p) arm of chromosome 17 were seen in over 40% of invasive cervical carcinomas in a metaphase cytogenetic study, with 17p+ marker chromosomes found to be the most common abnormality.3 Similarly, our previous interphase cytogenetic study demonstrated numerical abnormalities of chromosome 17 in 36% of a series of invasive squamous cell carcinomas,4 suggesting that numerical abnormalities are almost as common as structural abnormalities. Although both forms of loss suggest the presence of tumour suppressor genes on chromosome 17, p53 gene loss or mutation is uncommon in invasive cervical tumours.5 The presence of other suppressor loci on 17p has been hypothesised in breast cancer.6 Analyses of cervical carcinomas by microsatellite polymerase chain reaction (PCR) and restriction fragment length polymorphism (RFLP) analysis have produced conflicting results, with loss of markers on distal 17p reported in 15–55% of cases analysed.7, 8 Some possible explanations for this discrepancy include true biological variation between groups of tumours or the relative insensitivity of extraction based methods for the detection of regional genetic losses.
Therefore, the aims of our study were: (1) to determine by fluorescence in situ hybridisation (FISH) whether deletion of D17S34, a subtelomeric probe for 17p, occurs in invasive squamous carcinoma of the cervix; (2) to determine the extent of such loss by analysis of the p53 and HER2/NEU genes; and (3) to correlate these abnormalities with infecting HPV type.
Materials and methods
CHOICE OF SPECIMENS
Paraffin wax embedded biopsies from 14 invasive squamous cervical carcinomas were identified from the diagnostic files of the Royal Liverpool University Hospital, UK. The slides were reviewed and tumours classified according to the WHO classification of cervical tumours.9 Paraffin wax blocks from two normal cervixes removed as part of a routine hysterectomy from two patients with no evidence of intraepithelial or invasive cervical disease were also retrieved to serve as controls. Parallel 6 μm paraffin wax sections were cut for interphase cytogenetic analysis. Three 6 μm sections from each patient were also collected for HPV PCR typing.
Biotinylated chromosome specific pericentromeric probes for chromosomes 11 (D11Z1) and 17 (D17Z1) and digoxigenin labelled, locus specific probes for p53 (17p13.1), HER2/NEU (17q12.1), and D17S34 (17p13.3) (Oncor, Gaithersburg, Maryland, USA) were used for interphase karyotyping. D17S34 is immediately subtelomeric on the short arm of chromosome 17.10
FLUORESCENCE IN SITU HYBRIDISATION
Dewaxed sections were pretreated at 45°C for 15 minutes in 20% pretreatment powder (Oncor), digested with protein digestion enzyme (Oncor) for 30 minutes at 45°C, dehydrated through graded ethanols, and air dried. For dual hybridisation of locus specific and pericentromeric probes, 1 μl of pericentromeric probe was added to 10 μl of locus specific probe and the mixture applied directly to the section. For dual hybridisation of pericentromeric probes, 1 μl of each pericentromeric probe was added to 10 μl of Hybrisol VI (Oncor) and the mixture applied directly to the section. Glass coverslips were sealed with rubber cement, co-denaturation carried out at 80°C for eight minutes, and hybridisation at 37°C overnight. After hybridisation, sections were washed with 50% formamide, 2× saline sodium citrate (SSC); they were then pre-equilibrated to pH 7.0 at 42°C for 15 minutes, followed by 2× SSC at 37°C for eight minutes, and then 1× PBD (Oncor) for two minutes at room temperature. Detection was performed using the Oncor detection kit by incubation for 20 minutes at 37°C in a mixture of equal volumes of Texas red conjugated avidin and fluorescein conjugated antidigoxigenin in a humidified chamber. Sections were washed in three changes of 1× PBD, counterstained with 0.1 μg/ml DAPI/antifade (Oncor), and viewed microscopically. If amplification of the digoxigenin signal was required, the counsterstain was removed by incubation in 1× PBD for five minutes at room temperature and signal amplification was performed by sequential incubation in rabbit antisheep antibody and fluorescein conjugated antirabbit antibody at 37°C for 30 minutes each. Washes were carried out in 1× PBD.
The number of red and green signals in each nucleus was counted at a total magnification of ×1000 under oil immersion and the following rules were observed: (1) all signals were counted; (2) overlapping nuclei were not counted; and (3) split signals were counted as single signals. The signal number was recorded for each nucleus individually in the order of counting. One hundred tumour cell nuclei were analysed initially from each tumour using each probe. Further counts were performed if regional heterogeneity was identified on initial counting. Frequency distributions were generated and mean nuclear signal ratios calculated by expressing the mean nuclear signal number of red and green signals as a ratio. The same areas of each tumour were scored for each probe. Trisomic and tetrasomic populations were determined by the presence of three or four signals, respectively, in > 10% of nuclei, as previously validated for paraffin wax section analysis.11 The normal range was obtained by determining the mean ± 3 SD of the ratios obtained with each probe in the normal cervixes.
HPV typing by PCR
This was carried out as described previously.12 Briefly, DNA was extracted from 3 × 6 μm formalin fixed, paraffin wax embedded sections using proteinase K (Boehringer Mannheim, Mannheim, Germany) and the supernatant treated with Instagene (Bio-Rad, Hemel Hempstead, UK) to remove cellular debris. The quality of the DNA was assessed by the amplification of a 209 bp β globin fragment. GP5+/GP6+ primers, which amplify a 140–150 bp product (depending on HPV type) from the L1 gene, were used in a 40 cycle PCR reaction using the conditions described by Jacobs et al.13 Products were then run on an agarose gel and transferred to positively charged nylon membranes by diffusion blotting. Further PCR product was dotted on to nylon membranes for individual HPV analysis. A cocktail of 5′ digoxigenin labelled, type specific oligonucleotide probes for HPV types 6, 11, 16, 18, 31, 33, 39, 42, 43, 44, 45, 51, 52, 56, 58, and 6613, 14 was added to the Southern blot and individual probes were added to the individual dot blots. Hybridisation was carried out at 55°C overnight and was followed by signal development using sheep antidigoxigenin alkaline phosphatase conjugated antibody (Boehringer Mannheim) and nitroblue tetrazolium (NBT)/bromochloroindolyl phosphate (BCIP) for three to four hours.
Plasmid clones for HPV types 6, 11, 16, 18, 45, 51 (Dr E-M de Villiers, Germany), HPV-33 (Dr G Orth, France), HPV types 31 and 35 (Dr A T Lorincz, USA), HPV-58 (Dr Y Matsukura, Japan), and HPV types 43, 44, and 56 (purchased from the ATCC, USA); and sequenced PCR products for HPV types 39, 42, 52, and 66 were used as controls of probe specificity.
Twelve of the 14 tumours contained HPV sequences by PCR (table 1).
Each of the normal cervical tissues was analysed with all three of the locus specific probes and fluorescence ratios were determined. The mean ratio (3 SD) was 0.97 (0.08), giving a normal range of 0.89–1.05. Therefore, loss or gain was defined as a fluorescence ratio lying outside these limits. Loss of D17S34 was identified in four of the 14 samples (table 1). In one sample (patient 7), there was regional monosomy of the chromosome 17 centromere probe with normal fluorescence ratios for all three locus specific probes, suggesting loss of one copy of the whole chromosome. This was confirmed by dual hybridisation of this lesion with pericentromeric probes for chromosomes 11 and 17, which demonstrated loss of chromosome 17 relative to 11, which was disomic. The second sample (patient 8) showed loss of both D17S34 and p53, but with retention of the chromosome 17 centromere and HER2/NEU, indicating loss restricted to 17p. The final two samples (patients 2 and 3) showed loss of D17S34 only with retention of p53, the chromosome 17 centromere, and HER2/NEU. In one of these specimens (patient 3), loss of D17S34 was regional in distribution, with identification of three geographically separate cell populations, namely: (1) disomic for chromosome 17 and containing two copies of D17S34; (2) disomic for the chromosome 17 centromere, but with loss of one copy of D17S34; and (3) tetrasomic for the chromosome 17 centromere and disomic for D17S34 (figs 1–3). This suggests that D17S34 was lost before tetraploidisation; because it was seen only in part of the tumour the loss was probably acquired after the tumour became invasive. Abnormalities of p53 or HER2/NEU were not seen in the absence of D17S34 loss and no correlation with HPV type was observed.
These data demonstrate loss of D17S34 in a quarter to a third of the tumours. No consistent pattern of loss of probes was identified, indicating that the extent of the losses varies among tumours. Loss of part or all of the short arm of chromosome 17 was demonstrated in 47% of invasive cervical carcinomas by metaphase cytogenetics,3 but studies of loss of heterozygosity (LOH) from the short arm of chromosome 17 either by RFLP or microsatellite PCR have been inconsistent in their estimates of the prevalence of loss of this region. Two studies8, 9 that used the same marker (D17S34) used in our study, showed 15% and 55% loss in informative cases, respectively. This discrepancy is striking considering that both studies used similar methodologies and both analysed similar mixtures of squamous and glandular tumours. HPV analysis in one of these studies9 failed to demonstrate a clear relation between the loss of D17S34 and HPV type, although two of four informative specimens with loss of D17S34 were HPV negative. This is in keeping with our finding that one of the samples showing D17S34 loss was HPV negative. The use of other markers on 17p13.3 showed LOH in 32% (D17S1537) and 38% (D17S513) of 64 cervical tumours,15 figures that are broadly consistent with our findings. Moreover, in the same study, LOH of the p53 gene was found in 20% of tumours and LOH of markers on 17q in 4–7%. These findings are consistent with the frequency of loss found in our study. Specifically, D17S34 loss was present in twice as many tumours (four of 14) as p53 loss (two of 14), and loss of HER2/NEU was found focally in one tumour only.
The lack of a relation between genetic abnormalities and HPV type is consistent with previous evidence that HPV infection is an early event in the path of cervical carcinogenesis. The finding of p53 gene loss in an HPV negative tumour is interesting, although p53 loss was also identified in a tumour containing HPV-16, and was not identified in the second HPV negative sample, indicating that the relation between p53 loss and HPV infection is not simple. The findings of our study are consistent with the suggestion that p53 loss or mutation is a late event in cervical carcinogenesis, being identified more frequently in metastatic lesions and associated with poor prognosis.16
In two samples, there was regional heterogeneity of loss, one described above, and the other showing regional loss of the centromere and all three locus specific probes. The patterns of loss indicate that both were late events in tumour development, occurring after invasion. The presence of focal loss of D17S34 in one specimen indicates that progression to invasiveness did not require loss of this region. More specifically, the identification of three separate populations in this tumour showed that D17S34 loss occurred not only after invasion but before tetraploidisation. This finding is also consistent with the absence of two copies of D17S34 in a second tumour that was tetrasomic for the chromosome 17 centromere. Whether tetraploidisation occurred before or after invasion could not be determined for this latter tumour because no other populations were identified. Loss of D17S34 was not a universal accompaniment to tetrasomy 17 because normal ratios for all probes were identified in three tetrasomic specimens. These observations are compatible with a recent study examining synchronous intraepithelial, invasive, and metastatic cervical lesions, in which LOH of D17S513 (17p13.3) was absent from intraepithelial lesions, but present in some accompanying invasive tumours.17
HER2/NEU amplification has been identified in 14% of invasive squamous cell carcinomas of the cervix by molecular means.18 A consistent increase in chromosome 17 number was seen in five of the 14 tumours in this series. No abnormalities of HER2/NEU/centromere fluorescence ratios were identified, indicating that gain of this gene may often reflect gain of chromosome copy rather than linear amplification. These findings are in agreement with another study that used FISH methodology.19 HER2/NEU amplification was found in two of 23 cervical tumours but both of these tumours were adenocarcinomas: no amplification was found in the squamous carcinomas studied.
Thus, our study shows not only that loss of D17S34 occurs in a proportion of invasive squamous carcinomas of the cervix, but also that such loss is genetically heterogeneous and, at least in some cases, is a late event. We have shown that FISH analysis of paraffin wax embedded tissue sections can demonstrate regional loss of locus specific sequences.
CSH acknowledges the support of a fellowship from the Pathological Society of Great Britain and Ireland. We also thank Dr E-M de Villiers (Heidelberg, Germany), Dr G Orth (Paris, France), Dr AT Lorincz (Maryland, USA), and Dr Y Matsukura (Tokyo, Japan) for provision of HPV plasmid clones; and Dr AHN Hopman (Maastricht, the Netherlands) for provision of the pUC1.77 clone.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.