Publication of the International Union Against Cancer Publication de l’Union Internationale Contre le Cancer Int. J. Cancer: 70, 502–507 (1997) r 1997 Wiley-Liss, Inc. INTERPHASE KARYOTYPIC ANALYSIS OF CHROMOSOMES 11, 17 AND X IN INVASIVE SQUAMOUS-CELL CARCINOMA OF THE CERVIX: MORPHOLOGICAL CORRELATION WITH HPV INFECTION Shirley A. SOUTHERN1 and C. Simon HERRINGTON1* 1Department of Pathology, University of Liverpool, Liverpool L69 3GA, UK Human papillomavirus (HPV) infection has been widely implicated in cervical carcinogenesis, but it appears to be an early event, with other genetic abnormalities being required for biological transformation. In this study, interphase cytogenetic analysis of numerical abnormalities of chromosomes 11, 17 and X was performed on paraffin-embedded tissue sections from 25 invasive squamous-cell carcinomas of the cervix and compared with both histopathological features and the morphological distribution of HPV sequences as determined by in situ hybridisation. Numerical differences between chromosomes were identified in 76% of cases, with underrepresentation of chromosomes 11 and/or 17 relative to X in 64% of the total; 22 of 25 cases were HPV-positive, containing either HPV 16, 18 or 31. There was no relationship between the distribution of viral sequences and chromosomal pattern, suggesting that HPV infection precedes karyotypic changes. Our findings suggest that relative reduction in number of chromosomes 11 and 17 is important in the development of invasive cervical neoplasia and are consistent with the putative presence of relevant tumour-suppressor genes on these chromosomes. Int. J. Cancer, 502–507, 1997. r 1997 Wiley-Liss, Inc. Human papillomavirus (HPV) infection has been widely implicated in cervical carcinogenesis (Herrington, 1994a). 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 (Herrington, 1995). As the transformed phenotype is associated in vitro with genetic abnormalities and chromosomal abnormalities have been reported in cervical tumours (Teyssier, 1989), it is logical to hypothesize that in vivo transformation may be associated with similar chromosomal changes. Cytogenetic analysis of cervical tumours has shown that chromosomes 1, 3, 11 and 17 are commonly abnormal (Teyssier, 1989). Induction of aneuploidy by HPV 16 sequences suggests that viral DNA may play a part in this process and chromosomal abnormalities can be induced by HPV DNA in both cell (Matlashewski et al., 1988) and raft (McCance et al., 1988) cultures. Based on the study of somatic cell hybrids, it has been postulated that cellular genes are capable of suppressing HPV gene expression (zur Hausen, 1986). Although not directly identified, it has been suggested that an element responsible for this suppression is located on chromosome 11 (Bosch et al., 1990; Jesudasan et al., 1995). Numerical abnormalities of chromosome 11 are demonstrable in cervical carcinoma-derived cell lines by interphase cytogenetics (Herrington et al., 1995), and breakpoints in cervical tumours tend to cluster at 11q23 (Sreekantaiah et al., 1991), a site at which a high frequency of allele loss has been observed using microdissection and minisatellite/PCR techniques (Bethwaite et al., 1995). Abnormalities of chromosome 17 have also been described in cytogenetic studies, and amplification of the c-erb-B2 gene has been demonstrated in a proportion of invasive cervical tumours (Mitra et al., 1994). These observations support the possibility that genes on chromosomes 11 and 17 are important in cervical neoplasia. One approach to the assessment of the genetic content of tissue samples is the estimation of tumour ploidy by either flow or static cytometry (Padberg et al., 1992; Takahashi et al., 1994). However, these techniques are insensitive to minor changes in DNA content and, in cervical neoplasia, are complicated by the presence of viral DNA, which may significantly alter the overall DNA content of infected cells, particularly in the presence of viral replication. An alternative approach involves direct demonstration of karyotypic abnormalities either by metaphase cytogenetics, in which chromosomes are dispersed in metaphase after limited culture, or interphase cytogenetics, in which intact cells are used (Herrington, 1994b). For the assessment of intact tissues, interphase cytogenetics is required, a further advantage being that this approach can be applied to archival material, allowing retrospective study and the correlation of molecular and morphological features. At the time of writing, no published studies evaluating karyotypic abnormalities directly within intact cervical tissues were identified. The aims of our study were, therefore, (i) to evaluate numerical abnormalities of chromosomes 11, 17 and X in paraffin sections from 25 invasive squamous cervical carcinomas, using the X chromosome as a ‘‘control’’ in view of it being infrequently abnormal in cytogenetic studies; (ii) to correlate the interphase karyotype with morphological features; (iii) to investigate intratumoural heterogeneity of chromosome number; and (iv) to analyse the relationship between karyotypic changes and the morphological distribution of HPV infection as assessed by in situ hybridisation. METHODS Choice of cases Twenty-five consecutive cases of invasive squamous cervical carcinoma were selected from the diagnostic files of the Royal Liverpool University Hospital. Slides were reviewed and tumours classified according to the WHO classification of cervical tumours (Scully et al., 1994). Grading was performed using the Broder system (Kurman et al., 1992). Parallel 6 µm paraffin sections were cut for interphase cytogenetic analysis and HPV typing by in situ hybridisation. Probes The chromosome-specific probes used were biotinylated D11Z1, D17Z1 and DXZ1 (Oncor, Gaithersburg, MD). These probes have been localised previously to the peri-centromeric region of the appropriate chromosome (Herrington et al., 1995). Individual HPV plasmid clones for HPV 6, 11, 16, 18, 31 and 33 were labelled with digoxigenin by nick translation as previously described (Herrington et al., 1989). Interphase cytogenetics The method used was based on that of Hopman et al. (1992). Briefly, paraffin sections were dewaxed and pre-treated for 10 min in 1 M NaSCN at 80°C. Sections were washed in water, then digested with 0.4% (w/v) pepsin (Sigma P7012, Poole, UK) in 0.2 M HCl for 20–30 min at 37°C. Following washes in water, sections were air-dried and the probe was applied at a concentration of 1 ng/µl in 2 3 SSC, 10% dextran sulphate, 60% formamide and 1 *Correspondence to: Department of Pathology, University of Liverpool, Duncan Building, Royal Liverpool University Hospital, Liverpool L69 3GA, UK. Fax: 44-151-706-5859. Received 21 August 1996; revised 5 November 1996. CHROMOSOMAL ABNORMALITIES IN CERVICAL CARCINOMA µg/ml sheared salmon sperm DNA. Denaturation was carried out at 80°C for 8 min and hybridisation at 37°C overnight. After hybridisation, sections were washed with 60% formamide, 2 3 SSC, pre-equilibrated to pH 7.0 at 42°C for 20 min and then PBS containing 3% (w/v) BSA and 0.05% Tween 20 (buffer A) for 20 min at room temperature. Detection was performed by sequential incubation at 37°C in monoclonal anti-biotin (1:100), peroxidaseconjugated rabbit anti-mouse (1:80) and peroxidase-conjugated swine anti-rabbit (1:100) for 30 min each. Washes were carried out in buffer A. Signals were developed using diaminobenzidine (DAB)/H2O2, and sections were counter-stained in haematoxylin and mounted in DPX. Three areas were marked on each slide, with care being taken that the same tumour foci were analysed for each chromosome. In each tumour, the foci analysed were chosen such that there was no identifiable morphological difference between the areas, to minimise the possible effects of variation due to nuclear truncation. The number of signals per nucleus was counted at a total magnification of 3630, and the following rules were observed: (i) all signals were counted; (ii) overlapping nuclei were not counted; (iii) split signals, as defined previously (Herrington et al., 1995), were counted as single signals. Signal number was recorded for each nucleus individually in the order of counting. Tumour cell nuclei (n 5 200) from each of the 3 separate areas were analysed from each tumour, using each probe and frequency distribution generated. We have shown previously that there is no significant intra- or inter-observer variation using this approach (Southern and Herrington, 1996). Signal distributions were generated from the same sections in the same way from 100 stromal cells, using each probe in each case to assess technical comparability. 503 FIGURE 2 – (a): Distributions obtained with chromosomes 11 and 17 from one area of case 18. ‘‘Trisomic’’ patterns are present with both probes, but there is a significant difference between the distributions. Plotting signal number as a function of counting order demonstrated a trisomic population only for chromosome 17 but morphologically discrete disomic and trisomic populations for chromosome 11 (b). This case showed intra-tumoural heterogeneity with all 3 probes (see Table I). Distributions were compared using the 2-tailed Mann-Whitney U test corrected for tied values (Altman, 1991). Heterogeneity was assessed by statistical comparison of distributions obtained from each area with the same probe and from the same area with each different probe. Statistical analyses were corrected for multiple comparisons using the Bonferroni method to avoid type I errors (Altman, 1991): each distribution was compared with 4 other data sets; therefore, the significance level used was p , 0.012. HPV in situ hybridisation Sections were dewaxed and digested using proteinase K at a concentration of 0.5 mg/ml in 0.1 M Tris-HCl (pH 7.6), 0.1 M NaCl (TBS) for 15 min at 37°C. Following washing in TBS, slides were air-dried, the probe was applied and sections were denatured at 95°C for 6 min followed by hybridisation at 37°C overnight. FIGURE 1 – Distributions generated from a single area of (a) case 15 showing monosomy 11 with no significant population of disomic cells compared with disomy X and trisomy 17 and (b) a population monosomic for chromosome 17 with disomy X and 11 in case 22. Note in (b) that a significant disomic population is also present with chromosome 17. FIGURE 3 – Heterogeneity of chromosome number in case 16. There was no significant difference between the areas analysed, but there is a difference between the chromosomal distributions. 504 SOUTHERN AND HERRINGTON FIGURE 4 – Interphase cytogenetic analysis of case 14 with probes for chromosomes 11 (a), 17 (b) and X (c). Frequency distributions generated from these sections showed trisomic, disomic and tetrasomic populations, respectively (see Table I). HPV 18 sequences were widely distributed throughout the tumour (d ). Note that the degree of nuclear positivity for HPV is not related to nuclear size. Scale bars 5 5 µm for a–c and 10 µm for d. CHROMOSOMAL ABNORMALITIES IN CERVICAL CARCINOMA 505 Cocktails of HPV 6, 11, 16, 18, 31 and 33 were used to screen for HPV, and those cases positive were typed using individual HPV probes at a concentration of 2 ng/µl in 50% formamide, 2 3 SSC, 10% dextran sulphate. Post-hybridisation washes were carried out in 4 3 SSC followed by the following detection sequence: monoclonal antidigoxin (1:5,000), biotinylated rabbit anti-mouse antibody (1:200) and strepatividin alkaline phosphate conjugate (1:100). The signal was developed with naphthol AS-MX phosphate/ Fast red. Control HPV-infected biopsies and cell lines (HeLa and SiHa) were analysed in each experiment. HPV positivity, type and signal distribution were determined independently of the interphase cytogenetic analysis. Positivity was classified as punctate or diffuse, as previously described (Cooper et al., 1991). These patterns have been shown to correlate with integrated and episomal viral sequences, respectively (Cooper et al., 1991; Berumen et al., 1995). RESULTS Morphological analysis Five of the 25 tumours were keratinising and 20 non-keratinising squamous-cell carcinomas. No adenocarcinomas or small cell carcinomas were included. Two tumours were grade 1, 18 grade 2 and 5 grade 3. Interphase cytogenetics Absolute chromosome number. The absolute chromosome number present in each tumour cannot be determined directly as the effects of truncation of nuclei of variable sizes leads to distributions which cannot be directly extrapolated to the karyotype of whole cells. However, the significant population with the highest signal number was determined previously using data derived from normal epithelium (Southern and Herrington, 1996) to give an estimate of chromosome number, in keeping with data reported by others (Hopman et al., 1991). The approach taken here was to ensure technical comparability of results obtained with each probe by analysis of internal control stromal cells. No significant difference was found between distributions obtained using each of the 3 probes analysed in any case, allowing direct comparison of distributions obtained from the same tumour area with different probes. Monosomic populations were identified when there was a significant ‘‘left shift’’ of distributions compared with other probes showing a disomic pattern in the same tumour area (Fig. 1); this approach controls for the effects of nuclear truncation and technical variation. Trisomic and tetrasomic populations were identified by the presence of 3 or 4 signals, respectively, in greater than 10% of nuclei; this cut-off is based on sensitivity calculations derived previously using normal epithelium (Southern and Herrington, 1996) and agrees with the cut-off used by others (Wolman et al., 1992; Micale et al., 1994). Although each tumour area is assigned a single chromosome number in this way, this does not exclude the presence of other populations and does not determine the proportion of the tumour that the population constitutes. Thus, areas with the same assigned chromosome number can have significantly different distributions due to the presence of more than one population within that area (Fig. 2). Nevertheless, this method gives a qualitative assessment of chromosome number. Relative chromosome number and intra-tumoural heterogeneity. As no significant difference was found between distributions obtained from internal control stromal cells, distributions generated from the same tumour area with different probes could be compared statistically. Significant differences were recorded as indicating relative under- or over-representation of the chromosomes (Figs. 3, 4). Intra-tumoural heterogeneity for individual chromosomes was defined by the finding of a statistically significant difference between distributions obtained from different tumour areas (Fig. 5). FIGURE 5 – Distributions generated from case 8 using probes for chromosomes 11 and X. Note the intra-tumoural heterogeneity, with a mixture of disomic and tetrasomic populations. There is no significant difference between the chromosomes. The interphase cytogenetic results obtained with the 25 tumours are shown in Table I. Briefly, 6 showed no difference between the chromosomes: 4 were karyotypically normal with the 3 probes used and 2 showed evidence of focal tetraploidisation. In 14 cases, there was under-representation of chromosomes 11 and 17 relative to the X chromosome (Figs. 3, 4): in 10 cases, there was significantly greater loss of chromosome 17 than 11, in 2 there was greater loss of chromosome 11 and in 2 there was equivalent loss of chromosomes 11 and 17. In 2 cases, there was under-representation of chromosome 17 relative to both chromosomes 11 and X. Three cases showed under-representation of the X chromosome relative to chromosomes 11 and 17. Heterogeneity between tumour areas with the same chromosome probe was present in 14 tumours. In only 4 cases was there demonstrable heterogeneity with all 3 probes. Two of these cases showed changes consistent with tetraploidisation, there being tetrasomic populations present with all 3 probes, with no significant difference between the distributions obtained with each probe (Fig. 5). In both of these cases, the tumour foci showing tetraploidisation were morphologically localised but showed no morphological differences. HPV analysis Of 25 tumours, 22 (88%) were HPV-positive by in situ hybridisation. All positive cases contained a punctate signal within tumour cell nuclei, with 8 cases containing diffusely labelled nuclei in addition (Fig. 3d); 19 of these cases were HPV 16-positive, 2 HPV SOUTHERN AND HERRINGTON 506 TABLE I – MORPHOLOGICAL, INTERPHASE CYTOGENETIC AND HPV DATA1 Case Morphology 11 17 X Relative number HPV type 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 NK G2 NK G2 K G1 NK G2 NK G2 NK G2 NK G3 NK G2 NK G2 NK G2 NK G2 K G2 NK G3 NK G2 NK G2 K G3 NK G3 NK G3 NK G2 NK G2 NK G2 K G1 K G2 NK G2 NK G2 2/3H 3 2 2 2 2 4 2/4H 4H 4 2 2 4H 3 1 3 3 3H 2 2/3H 2/3H 2 2 2/4H 2 2 2 2 2 2 1/2H 3 2/4H 2/4H 3H 2 2 4 2 3 2 2 3H 1/2 2/3H 2 1/2 2 2/4H 2 2/3H 3/4H 2 4 2 2/4H 4 2/4H 1/2/4H 4H 1/2H 2 4 4 2/4H 4 3 4H 3 3 2/3/4H 2 2 2/4H 1/2H 17 , 11 5 X 17 , 11 , X 11 5 17 5 X 11 5 17 , X 11 5 17 5 X 17 , 11 , X 17 , 11 , X 11 5 17 5 X X , 17 , 11 17 , 11 , X X , 11 5 17 11 5 17 5 X 17 , 11 , X 17 , 11 , X 11 , 17 , X 17 , 11 , X 17 , 11 , X 11 , 17 , X 17 , 11 , X 11 5 17 , X 17 , 11 , X 17 , 11 5 X 11 5 17 5 X 11 5 17 5 X X , 11 5 17 16 16 16 16 162 — 16 162 162 — 16 16 312 182 16 18 16 16 162 16 16 162 16 — 162 1NK 5 non-keratinising; K 5 keratinising; G 5 tumour grade. Numbers refer to the karyotype identified with probes for chromosomes 11, 17 and X. H denotes heterogeneity between separate areas. Relative number refers to relative differences between signal distributions: , represent a significantly lower signal number and 5 represents no significant difference between tumour distributions.–2Cases containing diffusely positive cells by HPV in situ hybridisation. 18-positive and 1 HPV 31-positive. HPV 6, 11, and 33 were not identified in any case. There was no relationship with chromosomal pattern except in case 15, where single punctate signals were present throughout the tumour but were markedly larger in the area showing tetrasomy X. It is of note that this tumour had a unique chromosomal pattern (Fig. 1a; Table I). DISCUSSION Our data demonstrate non-random numerical chromosome changes in a series of invasive squamous-cell carcinomas of the cervix, with relative under-representation of chromosome 11 and/or 17 compared with the X chromosome in 64% of the tumours investigated. This finding is consistent with the hypothesis that tumour-suppressor genes of importance in cervical carcinogenesis are located on these chromosomes. More frequent and extensive under-representation of chromosome 17 appears at first sight discordant with the finding of gene amplification (e.g., of the c-erb-B2 gene on chromosome 17) in some previous studies (Mitra et al., 1994). However, an absolute numerical increase in chromosomes 11 and 17 was identified in 14 (56%) and 9 (36%) tumours, respectively, in this series. This suggests that, in some cases, gene amplification may in part be due to an increase in chromosome number rather than linear amplification and that, although the absolute number of a particular chromosome may be increased, its relative number may be reduced. This analysis underscores the necessity of using more than one chromosome probe to allow assessment of relative chromosome number in interphase cytogenetic studies. Overall, heterogeneity between chromosomes was detectable in 76% of tumours. Moreover, tetraploidisation was detectable when morphologically localised. This process is thought to be associated with tumour progression (Hopman et al., 1992) and may underlie the induction of aneuploidy by HPV sequences. This is in keeping with the presence of a significant population tetrasomic for chromosome X in 10 of the 14 cases in which under-representation of chromosomes 11 and 17 was identified. The 4 other cases showed focal or widespread trisomy for one or more chromosomes, suggesting that gain of individual chromosomes also occurs. A further group of cases disomic for all 3 chromosomes or in which there was focal loss of either chromosome 17 or, more commonly, the X chromosome was identified. These findings suggest that pathways other than tetraploidisation are important in squamous carcinogenesis in some cases. Three of the 5 keratinising tumours analysed were disomic for all 3 chromosomes compared with only one of the remaining 20 non-keratinising tumours. The fourth keratinising tumour showed only focal monosomy 17, and the final case was HPV 18-infected and of high nuclear grade. The lack of marked abnormality in most of the keratinising tumours may partly reflect the low grade of these lesions but also implies differences in the interaction between HPV sequences and neoplastic squamous cells in the presence of keratinisation. Intra-tumoural heterogeneity with individual probes was identified in this study in 14 of 21 non-disomic tumours (67%). The identification of heterogeneity is unlikely simply to reflect variation in nuclear size between areas as only 2 of 5 grade 3 tumours, all of which showed marked nuclear pleomorphism, demonstrated heterogeneity using this approach. It is of note that both of the HPV 18-positive tumours showed the same chromosomal pattern, with no significant heterogeneity between tumour areas with any of the 3 probes analysed. This raises the possibility that HPV 18 infection is associated with specific chromosomal changes; investigation of this hypothesis requires analysis of a larger group of cases. No relationship was identified between either the presence of HPV sequences in these tumours, or the pattern of signal obtained, and chromosome number, except in one case where focal increase in signal size correlated with the presence of a tetrasomic X chromosome population. This tumour had a chromosomal pattern unique in this series. This observation is not likely simply to be due to viral integration into the X chromosome as the variation identified was in signal size, not number, suggesting that viral amplification accompanied X chromosome duplication. Analysis of the distribution of diffusely labelled cells in the 8 cases in which CHROMOSOMAL ABNORMALITIES IN CERVICAL CARCINOMA they were present showed no relationship with chromosomal pattern. This was most clearly demonstrated by case 8, in which separate islands of tumour showed tetraploidisation but no difference in either morphological appearance or HPV staining distribution. This suggests that tetraploidisation is not directly related to viral DNA replication in these tumours. However, it does not exclude the possibility that it is related to viral gene transcription as all of the tumours contained a punctate viral signal, suggesting that integrated viral sequences were also present (Cooper et al., 1991; Berumen et al., 1995); these could equally lead to enhanced gene transcription and, hence, induction of aneuploidy. This study demonstrates that intra-tumoural karyotypic heterogeneity is frequent in invasive squamous carcinoma of the cervix. The presence of HPV sequences, in a pattern which correlates with viral integration, in the majority of cases suggests that HPV infection 507 precedes karyotypic changes and is consistent with the in vitro demonstration that HPV infection can induce aneuploidy. The relative chromosome number found in these cases suggests that relative loss of chromosomes 11 and 17 is important in the development of invasive cervical neoplasia. The identification of a smaller group of tumours with disomy or isolated loss of the X chromosome suggests that alternative pathways occur. Analysis of intra-epithelial lesions should help to refine the point at which these changes occur. ACKNOWLEDGEMENTS We thank the University of Liverpool and Wellbeing/RCOG for funding. REFERENCES ALTMAN, D.G., Practical statistics for medical research, Chapman and Hall, London (1991). BERUMEN, J., UNGER, E.R., CASAS, L. and FIGUEROA, P., Amplification of human papillomavirus type-16 and type-18 in invasive cervical cancer. Hum. Pathol., 26, 676–681 (1995). BETHWAITE, P., KORETH, J., HERRINGTON, C.S. and MCGEE, J.O.D., Loss of heterozygosity occurs at the D11S29 locus on chromosome 11q23 in invasive cervical cancer. Brit. J. Cancer, 71, 814–818 (1995). BOSCH, F.X., SCHWARZ, E., BOUKAMP, P., FUSENIG, N.E., BARTSCH, D. and ZUR HAUSEN, H., Suppression in vivo of human papillomavirus type 18 E6-E7 gene expression in nontumorigenic HeLa 3 fibroblast hybrid cells. J. Virol., 64, 4743–4754 (1990). COOPER, K., HERRINGTON, C.S., STICKLAND, J.E., EVANS, M.F. and MCGEE, J.O.D., Episomal and integrated HPV in cervical neoplasia demonstrated by nonisotopic in situ hybridization. J. clin. Pathol. 44, 990–996 (1991). HERRINGTON, C.S., Human papillomaviruses and cervical neoplasia I: virology, classification, pathology and epidemiology. J. clin. Pathol., 47, 1066–1072 (1994a). HERRINGTON, C.S., Interphase cytogenetics: principles and applications. J. Histotechnol., 17, 219–234 (1994b). HERRINGTON, C.S., Human papillomaviruses and cervical neoplasia II: interaction with other factors. J. clin. Pathol., 48, 1–6 (1995). HERRINGTON, C.S., BURNS, J., GRAHAM, A.K., BHATT, B. and MCGEE, J.O.D., Interphase cytogenetics using biotin and digoxigenin labelled probes I: relative sensitivity of both reporter molecules for HPV16 detection in CaSki cells. J. clin. Pathol., 42, 592–600 (1989). HERRINGTON, C.S., COOPER, K. and MCGEE, J.O.D., Interphase cytogenetics: analysis of numerical chromosome aberrations in isolated cells. J. Pathol., 175, 283–295 (1995). HOPMAN, A.H., VAN HOOREN, E., VAN DE KAA, C., VOOIJS, P.G. and RAMAEKERS, F.C.S., Detection of numerical chromosome aberrations using in situ hybridization in paraffin sections of routinely processed bladder cancers. Mod. Pathol., 4, 503–513 (1991). HOPMAN, A.H.M., PODDIGHE, P., MOESKER, O. and RAMAEKERS, F.C.S., Interphase cytogenetics: an approach to the detection of genetic aberrations in tumours. In: C.S. Herrington and J.O’D. McGee (eds.), Diagnostic molecular pathology: a practical approach, vol. 1, pp. 141–167, Oxford University Press, Oxford (1992). JESUDASAN, R.A., RAHMAN, R.A., CHANDRASHEKHARAPPA, S., EVANS, G.A. and SRIVATSAN, E.S., Deletion and translocation of chromosome 11q23 sequences in cervical carcinoma cell lines. Amer. J. hum. Genet., 56, 705–715 (1995). KURMAN, R.J., NORRIS, H.J. and WILKINSON, E., Tumors of the cervix, vagina and vulva. Atlas of tumor pathology (3rd series, Fasc. 4), Armed Forces Institute of Pathology (1992). MATLASHEWSKI, G., OSBORN, K., BANKS, L., STANLEY, M. and CRAWFORD, L., Transformation of primary human fibroblast cells with human papillomavirus type 16 DNA and EJ-ras. Int. J. Cancer, 42, 232–238 (1988). MCCANCE, D.J., KOPAN, R., FUCHS, E. and LAIMINS, L.A., Human papillomavirus type 16 alters human epithelial cell differentiation in vitro. Proc. nat. Acad. Sci. (Wash.), 85, 7168–7173 (1988). MICALE, M.A., VISSCHER, W., GULINO, S.E. and WOLMAN, S.R., Chromosomal aneuploidy in proliferative breast disease. Hum. Pathol., 25, 29–35 (1994). MITRA, A.B., MURTY, V.V.V.S., PRATAP, M., SODHANI, P. and CHAGANTI, R.S.K., Erb-B2 (Her2/neu) oncogene is frequently amplified in squamous cell carcinoma of the uterine cervix. Cancer Res., 54, 637–639 (1994). PADBERG, B.C., ARPS, H., FRANKE, U., THIEDEMANN, C., REHPENNING, W., STEGNER, H.E., LIETZ, H., SCHRODER, S. and DIETEL, M., DNA cytophotometry and prognosis in ovarian tumours of borderline malignancy—a clinicomorphological study of 80 cases. Cancer, 69, 2510–2514 (1992). SCULLY, R.E., BONFIGLIO, T.A., KURMAN, R.J., SILVERBERG, S.G. and WILKINSON, E.J., Histological typing of female genital tract tumours. World Health Organisation International Histological Classification of Tumours, Springer-Verlag, Berlin (1994). SOUTHERN, S.A. and HERRINGTON, C.S., The assessment of intra-tumoural karyotypic heterogeneity by interphase cytogenetics in paraffin wax sections. J. clin. Pathol. mol. Pathol., 49, M283–M289 (1996). SREEKANTAIAH, C., DE BRAEKELEER, M. and HAAS, O., Cytogenetic findings in cervical carcinoma: a statistical approach. Cancer Genet. Cytogenet., 53, 75–81 (1991). TAKAHASHI, Y., TAKENAKA, A., ISHIGURO, T. and NODA, Y., Intratumoral DNA heterogeneity correlated with lymph node involvement and surgical staging in epithelial ovarian cancer by flow cytometry. Cancer, 73, 3011–3014 (1994). TEYSSIER, J.R., The chromosomal analysis of human solid tumours: a triple challenge. Cancer Genet. Cytogenet., 37, 103–125 (1989). WOLMAN, S.R., MACOSKA, J.A., MICALE, M.A. and SAKR, W.A., An approach to definition of genetic alterations in prostate cancer. Diagn. mol. Pathol., 1, 192–199 (1992). ZUR HAUSEN, H., Intracellular surveillance of persisting viral infections: human genital cancer results from deficient cellular control of papillomavirus gene expression. Lancet, ii, 489–491 (1986).