300 Sex Transm Inf 1998;74:300–301

HPV testing in cervical screening Sex Transm Infect: first published as 10.1136/sti.74.4.300 on 1 August 1998. Downloaded from

For debate Jack Cuzick

There is little doubt that well organised Improvements in cytology via thin layer cytology based screening programmes, which preparations and automated screening are achieve high compliance and good quality con- likely to improve performance, to some extent, trol are eVective in saving lives. This has been but this will be at considerable expense and I well documented in Scandinavia and Scotland, believe that a new approach, more closely and the recent downward mortality trend in related to the process of cervical carcinogen- England and Wales1 indicates that the changes esis, is likely to be the best way forward. Detec- to the programme which occurred around tion of the human papillomavirus (HPV) offers 1988 also produced positive results. such an approach. It is found in well over 90% However, a programme based on solely con- of all cancers8 and has well established ventional cytology has important limitations. oncogenic potential.9 Tests have been devel- In places where screening was properly imple- oped which can detect the virus in a cervical mented initially in the 1960s, mortality has scrape and which are automatable and provide dropped by 50–70% but is now stable, suggest- a quantitative output. ing that the limits of eVectiveness have been It is important to recognise that HPV can reached. A recent audit of the UK programme2 only be detected reliably by DNA based tests. found that 47% of the fully invasive cancers Morphological changes on cytology or histol- (that is, ignoring microinvasive disease) in ogy (koilocytosis) are not specific or sensitive women under the age of 70 years occurred in for oncogenic HPVs and more often detect individuals with an apparently adequate HPV 6 and other low risk types which produce screening history. A further 11% had abnor- benign lesions, and do not have oncogenic malities reported on cytology, but were not potential. Even among DNA tests, the per- diagnosed with cancer until at least 6 months formance diVers widely. Early tests based on (and often several years) later. filter in situ methods were neither suYciently An ideal screening test should be performed sensitive nor specific enough to be useful as infrequently and be capable of detecting screening tests. The use of polymerase chain precursor or early easily treatable lesions with reaction (PCR) based tests has improved great accuracy. A once in a lifetime sigmoidos- sensitivity, but unless carefully controlled, copy around age 60 to detect and remove these tests are not suYciently specific to be colorectal adenomas may be the best embodi- used in a screening context. http://sti.bmj.com/ ment of this principle.3 The sensitivity of cytol- Persistence is a key attribute of infections related to high grade disease. This can only be ogy is limited by sampling problems, in which directly verified by repeated testing, but fortu- the abnormal cells do not get placed on the nately there are correlates available which make smear, and interpretation problems, where the it possible to improve the predictive value of a few abnormal cells that do appear may be single test. The most important of these is age. missed when examining the 100 000 or so nor- Transient infections are much more common on September 30, 2021 by guest. Protected copyright. mal cells that also are sampled. in younger women, and restricting HPV testing In many studies where other tests have also to women over age 30 (at least for primary been employed to refer women with negative screening) substantially reduces the false posi- smears for , sensitivities for cytology tive rate. Viral load is also important. PCR of only 50–80% for high grade cervical based tests are able to detect very low levels of intraepithelial neoplasia (CIN) have been 4–6 virus, which are often transient and not of reported. Also cytological screening is inef- clinical significance, and quantitative assays fective for adenocarcinoma, which is rapidly with thresholds for positivity of about 105 HPV accounting for a larger fraction of cancers. The copies in a smear give much better specificity tediousness of the job of the cytoscreener must with little loss of sensitivity for high grade CIN. also be acknowledged, and regularity with HPV type is also important although less well which scandals appear in the popular press understood. Fortunately the commonest type highlight all of these weaknesses. (HPV 16) is most often associated with high Cytology not only has problems with sensi- grade disease. tivity but also with specificity. Screening is Types 18, 31, 33, and 58 also give good pre- drowning in the “dysplasia swamp” of border- dictive value, but other types are less specific line and mildly dyskaryotic smears, where the and the gain in sensitivity may not be worth the yield of high grade pathology is low and the increase in false positives, unless found repeat- Imperial Cancer cost of referral and follow up is enormous. The edly or other (cytological) abnormalities are Research Fund, UK programme is currently estimated to cost also present. London about £130m a year (J Patnick, personal com- Definitive studies have yet to be completed, Jack Cuzick munication) and annual estimates of $6 billion but a number have shown very promising 4–6 Accepted for publication have been made for cervical screening in the results. Preliminary communications from 22 June 1998 United States.7 studies using the commercially available hybrid HPV testing in cervical screening 301

capture microtitre assay are reporting even assessing treatment failures and could be used higher sensitivities and specificities. Overall, to safely return negative women to positive 12 13 these studies suggest that adding HPV testing screening after a single follow up. This Sex Transm Infect: first published as 10.1136/sti.74.4.300 on 1 August 1998. Downloaded from to primary screening could increase the yield of could be yet another way in which HPV testing high grade CIN by 50–100%, with a positive improves the management of women with cer- predictive value similar to that for moderate vical abnormalities. . This may both reduce the inci- In summary, available evidence indicates dence of cancer and allow the screening inter- great potential for HPV testing within the cer- val to be increased to 5 yearly or longer, espe- vical screening programme. It oVers the cially in women over age 50 who have never possibility of greater sensitivity, reduced follow had an abnormal smear. up of low grade cytological abnormalities and Performing HPV testing as part of primary treated lesions, increased screening intervals, screening also oVers the possibility of rapid and and overall cost reductions. Large scale evalua- more accurate evaluation of women with tion projects are urgently needed to verify and borderline or mildly abnormal smears. In Brit- refine these indications. ain approximately 6% of all smears show borderline or mildly dyskaryotic changes. Only 1 Sasieni P, Cuzick J, Farmery E. Accelerated decline in cervi- about 10–25% of these women will have high cal cancer mortality in England and Wales. Lancet 1995;346:1566–7. grade underlying CIN lesions, a slightly higher 2 Sasieni PD, Cuzick J, Lynch-Farmery E and the NCN proportion will have low grade lesions, and the Working Group. Estimating the eYcacy of screening by auditing smear histories of women with and without cervi- remainder will have no detectable CIN at cal cancer. Br J Cancer 1996;73:1001–5. colposcopy. Current British guidelines recom- 3 Atkin WS, Cuzick J, Northover JMA, et al. Prevention of colorectal cancer by once-only sigmoidoscopy. Lancet mend cytological follow up at 6–12 month 1993;341:736–40. intervals unless progression or persistence (two 4 Reid R, Greenberg MD, Lorincz A, et al. Should cervical cytologic testing be augmented by cervicography or human mild or three borderline smears) occurs. papillomavirus deoxyribonucleic acid detection? Am J Return to routine 3–5 yearly screening is Obstet Gynecol 1991;164:1461–71. 5 Cuzick J, Szarewski A, Terry G, et al. Human papillomavirus recommended after two consecutive normal testing in primary cervical screening. Lancet 1995;345: smears at least 6 months apart. This approach 1533–7. 6 Cox JT, Lorinez AT, SchiVman MH, et al. Human papillo- leads to a large number of extra smears at short mavirus testing by hybrid capture appears to be useful in intervals which are both costly and cause anxi- triaging women with a cytologic diagnosis of atypical squamous cells of undetermined significance. Am J Obstet ety. There are also an increasing number of Gynecol 1995;172:946–54. reports of invasive cancer occurring in women 7 Cox JT. Clinical role of HPV DNA testing. In: Lorincz AT, Reid R, eds. Obstetrics and gynecology clinics of North who had a minor abnormality many years pre- America: human papillomavirus 11. 2nd ed. Philadelphia: viously followed by a number of (presumably WB Saunders, 1996. 10 11 8 Bosch FX, Manos MM, Munoz N, et al. Prevalence of false) negative smears. The alternative human papillomavirus in : a worldwide per- approach of performing colposcopies on all spective. J Natl Cancer Inst 1995;87:796–802. 9 International Agency for Research on Cancer. IARC mono- these women is even more expensive and graphs on the evaluation of carcinogenic risks to humans. Vol results in overtreatment and the unnecessary 64. Human papillomaviruses. Lyons: IARC, 1995. 10 Stanbridge CM, Suleman BA, Persad RV, et al. A cervical anxiety of a hospital visit in the majority of smear review in women developing cervical carcinoma with http://sti.bmj.com/ cases. Testing for HPV DNA on material taken particular reference to age, false negative cytology and his- tologic type of the carcinoma. Int J Gynecol Cancer 1992;2: at the time of the index smear oVers the 92–100. possibility of better management for these 11 Soutter WP, Fletcher A. Invasive cancer of the in women with mild dyskaryosis followed up cytologically. women. BMJ 1994;308:1421–3. Additionally, HPV testing oVers scope for 12 Bollen LJM, Tjong-A-Hung SP, van der Velden J, et al. Human papillomavirus DNA after treatment of cervical better follow up of women who have been dysplasia: low prevalence in normal cytologic smears. Can- treated for CIN. Currently, these women cer 1996;77:2538–43. on September 30, 2021 by guest. Protected copyright. 13 Elfgren K, Bistoletti P, Dillner L, et al. Conization for cervi- receive annual smears for at least 5 years and cal intraepithelial neoplasia is followed by disappearance of often for the rest of their life. Several reports human papillomavirus deoxyribonucleic acid and a decline in serum and cervical mucus antibodies against human suggest that the persistence of HPV positivity papillomavirus antigens. Am J Obstet Gynecol 1996;174: after treatment is an accurate method of 937–42.