Human papillomavirus type 16 and 18 detection in the management of mild dyskaryosis

Transcription

1 British Journal of Obstetrics and Gynaecology September 1999, Vol106, pp Human papillomavirus type 16 and 18 detection in the management of mild dyskaryosis *Margaret E. Cruickshank Clinical Research Fellow,?Suzanne Buchan PhD Student,?William T. Melvin Senior Lecturer, *Henry C. Kitchener Senior Lecturer *WellBeing Centre for the Prevention of Cervical Cancer; Aberdeen;?Department of Molecular and Cell Biology, University of Aberdeen Objective To determine if semi-quantitative human papillomavirus (HPV) types 16 and 18 detection by polymerase chain reaction can increase the sensitivity and specificity of repeat cytology alone for underlying high grade cervical intraepithelial neoplasia (CIN). Design Prospective randomised study of immediate treatment and surveillance. Setting A dedicated colposcopy clinic serving a regional population. Sample Three hundred and four women with smears reported as mild dyskaryosis. Methods Repeat cytology, HPV 16 and 18 tests, and colposcopy were performed at study entry. Women were randomised to either immediate treatment or surveillance with repeated tests at 6 and 12 months. Unless all study smears were negative, women were treated at study exit by large loop excision of the transformation zone. Main outcome measures Sensitivity and specificity of HPV testing for types 16 and 18 in conjunction with cytology for high grade CIN. Results Combining repeat cytology with HPV 16 and 18 testing had a sensitivity of 94% and a specificity of 26%, a positive predictive value of 71%. and a negative predictive value of 71 %, for underlying high grade CIN. If used to secondary screen in conjunction with repeat cytology for mild dyskaryosis, 88% of women would have been referred for colposcopy on the basis of either test being positive. Conclusion Combining repeat cytology and HPV 16 and 18 detection would result in the majority of women being referred for immediate colposcopy. Taken with an overall default rate of 17%, immediate referral of all women with mild dyskaryosis for colposcopic assessment still appears to be a more effective clinical strategy. INTRODUCTION Currently there is considerable interest in the use of human papillomavirus (HPV) detection as a screening tool for cervical intra-epithelial neoplasia (CIN) of the uterine cervix, and, in particular, whether the strong association between so called high risk HPV types 16 and 18 with cervical disease can be exploited to improve the management of women with low grade cytological abnormalities. The problem of low grade smears lies in the large number of women with such abnormalities and the significant proportion of them who have underlying high grade CIN. Confusion between cytological terminologies and morphological criteria has resulted in both the United States' and United Kingdom2 introducing new ~~ Correspondence: Dr Margaret E. Cruickshank, WellBeing Centre for the Prevention of Cervical Cancer, Aberdeen Maternity Hospital, Aberdeen AB25 2ZD, UK. classification systems for reporting cervical smears. An earlier randomised trial of mild dyskaryosis from our centre3 showed that when the baseline histological diagnosis was adequately defined using excision of the cervical transformation zone, the underlying prevalence of high grade CIN in women with mild smear changes was 40%. The authors advocated immediate referral of all women with mild dyskaryosis for colposcopy because: 1. a significant proportion have underlying high grade disease which was incorrectly classified by cervical cytology; 2. most of these women will eventually be seen for colposcopy to investigate persistent abnormalities; and 3. it avoids default. The management of women with low grade smears could be improved if an additional screening test could increase the pick up of high grade CIN without reducing the specificity of cervical cytology alone. In this way, only those women with high grade CIN would require colposcopic assessment, and cytological surveillance could be reserved for those without 0 RCOG 1999 British Journal of Obstetrics and Gynaecology 969

2 970 M. E. CRUICKSHANK ET AL. significant pathological change in whom there is a reasonable prospect for regression. The method of semi-quantifying HPV 16 DNA detection by polymerase chain reaction has been proposed as an accurate, noninvasive method of determining the underlying histology in women with abnormal smears4 by maintaining the sensitivity of the smear test and increasing the specificity. Previous studies proposing HPV 16 detection as such a secondary screening tool for mild dy~karyosis~~ have been cross-sectional, and the diagnostic outcome inconsistent, using a combination of colposcopic assessment only, directed punch biopsy, endocervical curettage or large loop excision of the transformation zone. Before we consider adopting HPV testing in our routine practice, its performance needs to be tested prospectively in a clinical trial. Using the hypothesis that the addition of HPV 16 and 18 testing could improve the sensitivity and specificity of repeat cervical cytology alone to select women with mild dyskaryosis for a see and treat management strategy, we report the results of a randomised study of management of mild dyskaryosis incorporating undisclosed HPV 16 and 18 testing. METHODS All smears from women screened in Grampian Region are sent to a single cytology laboratory, allowing recruitment from a stable geographically defined population without referral bias. Our laboratory participates in the Scottish Quality Assurance Program'". Women whose smears showed mild dyskaryosis with no previous smear of a higher cytological grade were referred to a dedicated colposcopy clinic. Women with previous cervical surgery or CIN were excluded. Women were randomised at the clinic, after informed consent had been obtained, and before any examination. Randomisation was to two management strategies; a see and treat group and a surveillance group, using computer generated random numbers sealed in opaque envelopes. Each woman had a repeat cervical smear for cytological examination and, at each visit, a duplicate smear was immersed in phosphate buffered saline and stored at -70 C for HPV detection. All referral and study smears were reported by one of two consultant cytopathologists and classified according to the British Society for Clinical Cytology's terminology and guidelines for adequacy of cervical smears'. Ethical approval for this study was given by the local Joint Ethical Committee. DNA was extracted from the thawed samples using standard techniques". Polymerase chain reaction was performed on the DNA products with primers directed against a 120 bp fragment of the HPV 16 gene", a 100 bp fragment of the HPV 18 gene, and a 268 bp fragment of the P-globin gene using polymerase chain reac- tion programme conditions described by Ngan et al.''. The P-globin primer was used as an internal control for DNA integrity. Polymerase chain reaction products were separated electrophoretically through an 8% polyacrylamide gel. CaSki cell DNA was used as an HPV 16 positive control, HeLa cell DNA as an HPV 18 positive control, and distilled water as a negative control for each sample. Semi-quantitative levels were estimated by visual comparison with the intensity of the P-globin band by a single individual (S.B.), without knowledge of the cytology or histology results. No band was scored as negative, a band less intense than that of the P-globin band was scored as low, a band of equal intensity as intermediate and a stronger band as high. All first visit samples were tested for P-globin, HPV 16 and HPV 18. For subsequent visits P-globin and HPV 16 only were tested. Women in the see and treat group underwent large loop excision of the transformation zone as an outpatient procedure at the first visit, and tissue samples were sent for histological examination. This was to obtain reliable baseline histopathological data on the control group in order to interpret the sensitivity and specificity of the viral test and the surveillance group data. Histological reporting was based on the currently accepted UK criteriat3 with high grade CIN demonstrating undifferentiated, nonstratified cells with a high nucleocytoplasmic ratio involving two-thirds or more of the epithelial thickness. Women in the surveillance group were reviewed at 6 and 12 months, when colposcopic assessment and cervical smears were repeated as described above. In addition, at the 12 month visit, all women, except those for whom all study smears reported as negative, were treated by large loop excision of the transformation zone. Surveillance was discontinued if the cervical smear showed severe dyskaryosis, or if there were colposcopic features suggestive of invasive disease. These women were seen immediately for colposcopy and treatment and were included in the analysis. Those who failed to attend for review after two invitations, those who requested withdrawal from the study, and those who were unable to complete the study because of pregnancy were classified as defaulters. We aimed to recruit 300 women to the study. This gave an 80% chance of detecting an increase in the sensitivity of repeat testing for underlying high grade CIN from 50% to 70%, and an 80% chance of detecting an increase in the specificity from 65% to 80% by incorporating the HPV 16 and 18 test. Data were analysed by randomisation group using the Statistical Package for Social Science (SPSS). The x2 test was used to compare categorical variables and a weighted kappa statistic was used to assess the correlation between repeated measures. Defaulters were included

3 HPV 16 AND 18 DETECTION IN MILD DYSKARYOSIS 971 Table 1. Histological outcome for 279 women who completed the study protocol or were withdrawn. Values are given as n (%). CIN = cervical intraepithelial neoplasia. Histopatholo@cal See and treat group Surveillance group diagnosis (n = 142) (n = 137) All smears negative 0 (0) 5 (4) No CIN 9 (6) 14 (lo)* Viral changes 16 (11) 17 (12) CIN 1 25 (18) 29 (21) CIN 2 23 (16) 14 (10) CIN 3 67 (47) 57 (42) Micro-invasion 2 (1) 1 (1) *P = not significant. in the analysis according to their randomisation group. P = 0.05 was taken as significant. RESULTS Three hundred and four women with mild dyskaryosis were recruited. Fifty-three (17%) had previous borderline smears and 53 (17%) previous mildly dyskaryotic smears. Ages ranged from 17 to 61 years (median 28). One hundred and forty-two women were randomised to the see and treat group and 162 to cytological surveillance. One hundred and ten women (68%) in the surveillance group completed the study protocol. Two (1%) completed surveillance but refused treatment and had a histological sample taken by colposcopic directed punch biopsies. Twenty-two women (1 4%) were withdrawn because of a study smear reported as severe dyskaryosis. Twenty-eight women failed to attend for a repeat smear on surveillance, giving an overall default rate of 17%. One became pregnant during the study and was unable to complete the protocol. One woman underwent a hysterectomy for menorrhagia, and six requested withdrawal from surveillance as they wanted treatment. Of the 28, 20 women were regarded as true defaulters. The histopathological outcome was available for 142 women in the see and treat group and 137 women in the surveillance group (Table 1). Ninety women (63%) in the immediate treatment arm and 71 (52%) in the surveillance arm, a nonsignificant difference, were reported to have high grade CIN. There was a nonsignificant difference in the number of women with either no CIN on biopsy or three consecutive negative smears following 12 months surveillance, compared with those in the see and treat group (14% and 6%, respectively). Two women in the immediate treatment arm and one woman in the surveillance arm had micro-invasive disease of the cervix. Histology was reviewed and confirmed two FIGO 1 a1 and one FIGO 1 a2 tumours. The cytological results at 0 and 12 months for the surveillance group are shown in Table 2. These appear to demonstrate cytological regression and progression over time, but the weighted kappa value, based on no women with severe changes as they were withdrawn from surveillance, was 0.13, indicating a poor agreement between results over time. Despite an underlying rate of high grade CIN of over 50% at study entry and exit, only 14% of women were withdrawn from surveillance with severe dyskaryosis, and just five (3%) completed the surveillance protocol with all smears negative. Table 3 demonstrates this discrepancy between the cytological and histological diagnosis. Seventy-seven percent of women with high grade CIN had dyskaryotic smears of any grade at six months, compared with 37% of women with low grade CIN or less (P c 0.05). The six month repeat smear had a sensitivity for high grade CIN of 77% and specificity of 63%. All 304 women had cytology samples processed for HPV 16 and 18 from their first clinic visit. Ninety-four percent of women (133 of 141) attending for a second visit had an HPV 16 test processed, and 98% (110 of 112) at their third visit. Of all samples taken for polymerase chain reaction, 36 of 547 (6.5%) were discarded because they were negative for P-globin and therefore inadequate for HPV detection. This compares with 4% (95% CI -5.2 to 4.1) of smears reported as inadequate. At the first clinic visit, 142 of 279 (51%) of all the Table 2. Cytological results of repeat smears*. Values are given as n (%). Surveillance at 12 months Smear result Surveillance at 0 months Negative Borderline Mild Moderate Severe *8 women with severe dyskaryosis at 0 months were withdrawn from surveillance. 0 RCOG 1999 Br J Obstet Gynnecol 106,

4 ~ 972 M. E. CRUICKSHANK ET AL Table 3. Relationship between an immediate repeat smear, a repeat smear at 6 months and the histologic outcome. Values are given as n (%). CIN = cervical intraepithelial neoplasia. Immediate treatment at 0 months Surveillance at 0 months Surveillance at 6 months Nonow grade High grade Nonow grade High grade No/low grade High grade Smear CIN (n = 46) CIN (n = 90) CIN (n = 63) CIN (n = 70) CIN (n = 57) CIN (n = 64) Negative 13 (28) 9 (10) 12 (19) 3 (4) 21 (37) 5 (8) Borderline 12 (26) 6 (7) 12 (19) 8 (11) 15 (26) 10 (16) Mild 20 (44) 44 (48) 28 (44) 34 (49) 13 (23) 21 (33) Moderate 1 (2) 25 (28) 9 (14) 20 (29) 7 (12) 15 (23) Severe 0 (0) 6 (7) 2 (3) 5 (7) 1 (2) 13 (20) women had medium or high levels of HPV 16 present, and 50 of 279 (18%) were HPV 18 positive. Renty-six women (9%) were positive for HPV 16 and 18. Table 4 demonstrates fluctuations in HPV DNA positivity and viral levels over time. The number of high risk cases falls with the withdrawal of women with severe dyskaryosis and may contribute to the poor agreement between HPV levels over time (weighted kappa value 0.005). Table 5 shows the relationship between semi-quantitative levels of HPV 16 and 18 at 0 months and the histological outcome. An HPV level was considered to be positive if intermediate or high. When taken following a mildly abnormal smear, these viral tests had a sensitivity of 41% and a specificity of 88% with a positive predictive value of 79% and a negative predictive value of Table 4. Human papillomavirus (HPV) 16 semi-quantitative levels taken at 0 and 12 months. Values are given as n (%). HPV 16 level Surveillance at 12 months Surveillance at 0 months Negative Low Intermediate High Negative (n = 9) 1 (11) 5 (56) 3 (33) 0 (0) Low (n = 39) 3 (8) 19 (49) 14 (35) 3 (8) Intermediate(n=31) 5 (16) 15 (48) 10 (32) 1 (3) High (n = 21) 1 (5) 9 (43) 10 (47) 1 (5) 57% for underlying high grade CLN. The sensitivity of combining repeat cytology with HPV 16 and 18 level was 94% and the specificity 26%, with a positive predictive value of 71 % and a negative predictive value of 71%. An immediate HPV 16 and 18 test was used to test for the eventual histological outcome after 12 months surveillance to see if the small number of women who regressed to negative histology could be identified prospectively. There was no significant association between the initial HPV test and the histological diagnosis at 12 months. Thirty-one of 66 women (47%) who were negative for both HPV 16 and 18 at first visit had CIN on histology at their final visit. However, all three women with microinvasive disease would have been identified as high risk by the initial HPV test. Eightyeight percent of women with mild dyskaryosis would be referred for colposcopy on the basis of either the cytological or HPV 16 and 18 tests being positive. The outcome for women in the surveillance group in terms of default or withdrawal with severe dyskaryosis is shown in Fig. 1. Figure 2 shows how many cases of high grade CIN would have been detected during the study if the criteria for withdrawal had included an intermediatehigh HPV 16 result in addition to severe dyskaryosis. Fifty-eight percent (31 of 53) of cases of high grade CIN would have been detected following an immediate repeat smear and only 11 % (6 of 53) would have been left undetected at the third visit. Table 5. Relationship between semi-quantitative human papillomavirus (HFV) 16 and 18 level at first visit and histologic outcome. Values are given as n (%). CIN = cervical intraepithelial neoplasia. See and treat group Surveillance group HPV level Low grade CIN (n = 50) High grade CIN (n = 92) Low grade CIN (n = 65) High grade CIN (n = 72) HPV 16 and 18 Negativdow 24 (48) 38 (41) Mediumhigh 26 (52) 54 (59)* *P = not significant. 35 (54) 30 (46) 31 (43) 41 (57)* 0 RCOG 1999 Br J Obstet Gynaecol 106,

5 HPV 16 AND 18 DETECTION IN MILD DYSKARYOSIS 973 Fig. 1. Outcome for women randomised to surveillance and the number of cases of CIN 3 detected in women withdrawn from the study protocol and colposcoped. Figure 3 illustrates how efficiently each test or combination of tests functions. An ideal screening test, with high sensitivity and high specificity, would lie in the upper left hand comer of the graphi4. This shows that a repeat smear at six months is more useful as a secondary screening test than an immediate repeat smear or an HPV 16 test. DISCUSSION At the centre of the debate on the management of mild dyskaryosis is the problem of the subjective nature of interpreting cervical cytology with both intra- and interobserver variability, particularly in the classification of low grade abnormalities. Other important issues are the costs, both in direct economic terms and indirect costs to the colposcopy service, as well as psycho-social costs to the women themselves. It is now widely accepted, on both epidemiological and molecular evidence, that HPV infection is the most important factor in the early stages of cervical neoplasia. HPV types 16 and 18 have been formally classified

6 974 M. E. CRUICKSHANK ET AL. Fig. 2. Anticipated outcome for women randomised to surveillance if high risk HPV 16 level in addition to severe dyskaryosis has been used as the criteria for withdrawal from the study protocol and the number of cases of CIN 3 which could have been detected. as carcinogenic viruses by the International Agency for Research on Cancer. Although over 70 different HPV types have been identified to date, of those described as high risk for cervical disease (16, 18,31,33,35,45, 51, 52, 56, 58), HPV 16 has been detected in 47% of pre-invasive lesions and invasive cancers of the cer~ix'~, making it the most prevalent subtype. HPV 16 appears to be the single most important HPV type for predicting high grade CIN, and to be more specific than other subtypes both in primary and secondary screening5j6. The application of a secondary screening test using HPV detection is attractive as a more objective test, and as a possible means of reducing costs by rationalising the management for individual women. Any new screening technology needs to balance sensitivity with specificity. A woman with mild dyskaryosis has already been identified by conventional cytological screening as at increased risk of high grade CIN, which we have confirmed. A test with high specificity would be more useful in secondary screening to balance the high sensitivity of cytology. Although combined cytology and HPV 16 detection had

7 HPV 16 AND I8 DETECTION IN MILD DYSKARYOSIS 975 C loo 1 I 0 0 i* A *OI, I I, I I I,, I * W-specificity Fig. 3. Sensitivity plotted against 100-specificity. An ideal screening test would lie in the upper left hand comer. H smear and HPV 16 & 18; 0 HPV 16 & 18; A smear at 6 months; + smear at 0 months. great sensitivity for high grade CIN, this was at the expense of both specificity and positive predictive value. Bavin et al7 found that 66% of women with mild or moderate dyskaryosis but underlying high grade CIN had high/medium levels of HPV 16 DNA. That study suggested that if HPV 16 quantitation were used in conjunction with cytology, 89% of women with high grade disease would be identified and the number of women referred for colposcopy would be halved. Cuzick et al. showed similar levels of sensitivity and specificity with a positive predictive value of 100%. In contrast, our results suggest that secondary screening with combined cytology and HPV 16 and 18 detection would result in a high referral rate for colposcopy (88%), with those women remaining under surveillance at risk of defaulting. This clearly does not offer a major advantage over immediate referral of all women with mild dyskaryosis for colposcopy. A more recent study on women attending for colposcopy, including those with low grade smears17, found that the positive predictive value of a commercial test for a panel of high risk HPV types was only 22%, with a high frequency of HPV negative women having biopsy proven high grade CIN. Those authors concluded that HFV triage was not valuable, and recommended cytological follow up. This concurs with our finding that a repeat smear at six months appears to have a better balance of sensitivity against specificity than HPV testing. Nevertheless, the problems of cytological surveillance, default and persistent smear changes which would result in colposcopy referral were not addressed. Although semi-quantitative polymerase chain reaction by similar techniques was used in many of these studies, each laboratory was working independently and polymerase chain reaction remains a research tool with no formal inter-laboratory comparison. Standardisation and comparability of HPV testing are necessary if results from different centres are to be widely applicable. Differences may also arise from the variability in determining the histological outcome. Histology was taken from a large loop excision of the transformation zone specimen, unless the risk of unidentified CIN was negligible, to give complete information on histological changes in the transformation zone of the cervix, as punch biopsy alone will undercall the histological grade in 35% of women with CIN 3. This was confirmed in our previous study3, and may account in part for the high proportion of women with high grade CIN. The population which our clinic serves is not high risk and there is almost no cross boundary referral. Our laboratory participates in external quality control and our reporting rates for cytological abnormalities comply with national guidelines in the UKI9. Our results also confirm the finding of Bollen et al. that HPV status does not differ significantly between women with a single mild smear and those with a repeated abnormality. Three women in this study, including one in the surveillance arm, were found to have micro-invasive disease of the cervix on loop biopsy. This confirms the findings of Soutter and FletcheP that not only is this group of women at higher risk of developing cervical cancer, but that it may coexist with a low grade smear. Although HPV infection is acknowledged to be an important factor in the development of cervical neoplasia, HPV infection alone in cervical cells is much more prevalent than CIN. Detection of HPV DNA in cervical cells, including the viral type and viral load, which we confirmed, fluctuates over time2. It is therefore possible that HPV testing will increase the sensitivity for high grade CIN detection, but that the lack of specificity for high grade CIN will impair its use in widespread clinical practice. If secondary screening is the way forward in the management of low grade smears, then a more specific test is necessary. Because viral load appears to be an important factor in the development of CIN, this might be achieved by adjusting the threshold for a positive test, using a less sensitive test or requiring a higher viral load to register a positive result. In our study, had we called only a high level in semi-quantitative analysis a positive result, HPV 16 and 18 test alone would have given a sensitivity of only 24% but a specificity of 84%. Conversely, if the test is required to select out all cases of high grade CIN, the sensitivity could be increased by registering any HPV DNA as positive. In this situation, HPV 16 and 18 positivity alone would have a sensitivity of 80% and a specificity of 22%. Commercial HPV test kits have higher thresholds than polymerase chain reaction for HPV positivity, but have produced conflicting outcomes in different clinical settings1 -. HPV detection may be of more clinical use in a health care system

8 976 M. E. CRUICKSHANK ET AL where there is pressure to colposcope women with even lower grade smears, such as exists in the United States. The high sensitivity of combined testing may contribute more in the management of lower risk groups who have a low underlying prevalence of CIN, such as women with smears showing atypical squamous cells of uncertain significance or borderline changes**. Our results suggest that HPV 16 and 18 testing is unlikely to be of significant benefit in managing women with mild dyskaryosis, reinforcing our previous findings that referral for colposcopy is the most effective management for women with mild dyskaryosis. Acknowledgements This research was funded by an Acute Healthcare Committee project grant from the Scottish Office Department of Health, and by WellBeing. References 1 Lundberg GD. The 1988 Bethesda system for reporting cervicallvaginal cytologic diagnosis. JAMA 1989; 262: Evans DMD, Hudson EA, Brown CL et al. Terminology of gynaecological cytopathology. Report of working party of the British Society for Clinical Cytology. J Clin Patholl986; Flannelly G, Anderson D, Kitchener HC et al. The management of women with mild and moderate cervical dyskaryosis. BMJ 1994; Cuzick J, Terry G, Ho L, Hollingworth T, Anderson M. Human papillomavirus type 16 DNA in cervical smears as predictor of high-grade cervical intraepithelial neoplasia. Lancer 1992; Cuzick J, Terry G, Ho L, Hollingworth T, Anderson M. Type-specific human papillomavirus DNA in abnormal smears as a predictor of high grade cervical intraepithelial neoplasia. Br J Cancer 1994; 69: Burger MPM, Hollema H, Pieters WJLM, Quint WGV. Predictive value of human papillomavirus type for histological diagnosis of women with cervical cytological abnormalities. BMJ 1995; 310: I Bavin PJ, Giles JA, Deery A et al. Use of semi-quantitative human papillomavirus DNA type 16 to identify women with high grade cervical disease in a population presenting with a mildly dyskaryotic smear report. Br J Cuncer 1993; 67: Mansell ME, Ho L, Terry G. Singer A, Cuzick J. Semi-quantitative human papillomavirus DNA detection in the management of women with minor cytological abnormality. Br J Obsrer Gynaecol1994; 101: Bollen UM, Tjong-A-Hung SP, van der Velden J et al. Human papillomavirus deoxyribonucleic acid detection in mildly or moderately dysplastic smears: a possible method for selecting patients for colposcopy. Am J Obster Gynecoll997; 177: Working Partv in Internal Quality - - Control for Cervical CvtoPatholonv Laborat&es.-Report of Working Party on Internal Qualiiy control Kr Cervical Cytopathology Laboratories. Edinburgh: Scottish Office Home and Health Department, Young LS, Bevan IS, Johnson MA et al. The polymerase chain reaction: a new epidemiological tool for investigating cervical human papillomavirus infection. BMJ 1989; Ngan HYS, Stanley M, Liu SS, Ma HK. HPV and p53 in cervical cancer. Genitourin Med 1994; 70: Buckley CH. Butler GB, Fox H. Cervical intraepithelial neoplasia. J Clin Puthol1982; 35: Altman DG, Bland JM. Diagnostic tests 3: receiver operating characteristic plots. BMJ 1994; Lorincz AT, Reid R, Jenson AB, Greenberg MD, Lancaster W, Kurman RJ. Human papillomavirus infection of the cervix: relative risk associations of 15 common anogenital types. Obstet Gynecol 1992; Cuzick J, Szarewski A, Terry Get al. Human Papillomavirus testing in primary cervical screening. Lancer 1995; 345: Kaufman RH. Adam E, Icenogle J et al. Relevance of human papillomavirus screening in management of cervical intraepithelial neoplasia. Am J Obsrer Gynecol 1997; 176: Buxton El, Luesley SM, Shafi MI, Rollanson M. Colposcopically directed punch biopsies: a potentially misleading investigation. Br J Obstet Gynaecoll991; 98: Herbert A. Achievable standards, benchmarks for reporting and criteria for evaluating cervical cytopathology. Shefield: National Health Service Cervical Screening Programme, Soutter WP, Fletcher A. Invasive cancer of the cervix in women with mild dyskaryosis followed up cytologically. BMJ 1994, 308: Ho GYF, Burk RD, Klein S et al. Persistent genital human papillomavirus infection as a risk factor for persistent cervical dysplasia. J Natl Cancerlnst 1995; 87: Cox JT, Lorincz AT, Schiffman MH, Sherman ME, Cullen A, Kurman RJ. Human papillomavirus testing by hybrid capture appears to be useful in triaging women with a cytologic diagnosis of atypical squamous cells of undetermined significance. Am J Obstet Gynecol 1995; 172: Accepted I February 1999