Does HPV cause cervical cancer? Will the new data transform our approach to primary screening?
Cervical cancer is now a disease that canat least theoreticallybe prevented by screening the population at risk, detecting cervical cancer precursors, and appropriately treating those precursors and preventing the development of invasive cancer. In earlier symposia and articles in Contemporary OB/GYN ["Cervical neoplasia: past, present, and future," May 1998, Special 25th Anniversary Issue; "HPV DNA testing comes of age," July 1995, among others], we've highlighted the technology of HPV DNA and how the new commercially available tests allow us to detect high-grade lesions and invasive cancers with a high degree of probability and low-grade lesions with a reasonable degree of probability.
The continued advance of technology makes this an especially exciting time to update the topic, which we have done by bringing together a distinguished panel at the 18th annual Human Papillomavirus (HPV) Meeting in Barcelona, Spain. At this juncture, all the basic work that has been applied to the understanding of HPV and its role in human disease is finally culminating in some really practical potential applications for screening. Our experts grapple with such topics as: How do we apply this in primary screening? Given the high false-negative rate of cytology, will the Pap smear give way to HPV DNA? What will be its role in evaluating the gray area of ASCUS Pap smears? Can we use high-risk HPV testing for triage?
Ralph M. Richart, MDModerator
Richart: Dr. Kjaer, please briefly outline the evidence that HPV is the cause of human cervical cancer.
Kjaer: Evidence comes from both the epidemiologic and laboratory fields. HPV has been found in more than 95% of cervical cancers worldwide.1,2 Furthermore, when women who are infected with the so-called high-risk HPV types are compared with HPV-negative women in casecontrol studies, they have been found to have a substantially increased risk for high-grade cervical lesions.3-5 Most recently, data from prospective cohort studies have also shown that HPV both precedes and predicts the development of high-grade squamous intraepithelial lesions (HSIL), and that especially persistent HPV infection is a key factor.6,7
Richart: And the evidence doesn't stop there, does it?
Meijer: Indeed not. A third argument is the good histologic and cytologic evidence that high-risk HPV types can be demonstrated only in the tumor cells or the neoplastic cells but not in the reactive cells.
Cuzick: And finally there is good evidence that the virus has a direct role in the carcinogenetic processes leading up to cancer. The virus makes oncogenic proteinscalled E6 and E7that bind with the cell-regulatory proteins p53 and pRB respectively, thereby unblocking the cell cycle, which essentially leads to genetic instability. This later gives rise to additional genetic changes that lead to cervical cancer by preventing apoptosis of cells with DNA mutations. This process has been shown to be specific to the high-risk HPV types and does not occur, for example, with HPV 6 and 11. These two types are associated with benign warts, but have no role in cancer.
Prospective studies have clearly shown that HPV is necessary for the development, maintenance, and progression of the precursor lesionsthe CIN (cervical intraepithelial neoplasia) lesionand especially CIN 3.8 And without high-risk HPV a woman will not develop cervical cancer. HPV 16 is by far the most common and important type worldwide, but at least 12 other high-risk types are known.
Richart: The etiology of cervical cancer has been debated for years, and we've outlined the reasons to believe it is caused by HPV. Is there any room for doubt about this anymoreor has this issue finally been settled?
Cuzick: In every respect, this is the clearest cause of any human cancer that we know. The evidence for HPV and cervical cancer is far stronger than even the evidence for smoking and lung cancer, in that the relative risks are hundreds rather than tens. The mechanisms are much more carefully worked out, as well.
Richart: Dr. Kjaer, you've been heavily involved in studying the natural history of HPV infections. Please outline the basics.
Kjaer: Basically, there is a high prevalence of HPV infection in sexually active young women, and most of these infections seem to be of a transient nature. The prevalence goes down significantly with age, so it is reasonably low among women over 30. On the other hand, we also know that the proportion of persistent infections increases with age. Some studies have shown that we probably will see a second peak in prevalence among women over 60 or maybe even a little younger.9,10 But whether this is occurring in both so-called high-risk and low-risk populations still has to be shown.
Richart: Why does the prevalence go down with age?
Kjaer: It is hypothesized to be due to an age-related change in the sexual habits of women, with fewer new sex partners and consequently less exposure to HPV. But another explanation is an acquired immunity. It's interesting that in a population of Danish prostitutesa group of women who tend not to change their sexual habits with agewe still see this very significant decrease in HPV prevalence.11 That means that even in a very sexually active population, cervical cancer screening using HPV is still a possible option. I think this is important. Recognizing that HPV persistence seems to be a key step in cervical carcinogenesis, we've looked at the risk factors for persistence of HPV, and it's clear from the few studies done so far that having oncogenic types (compared with nononcogenic types) is one of the main factors.6,12 Also a factor is that the risk of having a persistent HPV infection increases with age.
Richart: Do multiple infections play a role?
Kjaer: Two studies found that being infected with multiple HPV types increases the risk of persistence but the association is still uncertain (S.K. Kjaer, MD, unpublished data, 2000).6
The results regarding the role of smoking in relation to the risk of cervical cancer are not consistent, but if smoking plays a role, it could be because smoking affects the risk of persistence of the HPV infection. Here the results are also a little equivocal: Although we've found that smoking increases the risk of persistence, two other studies found the opposite.6,12
Richart: Dr. Cuzick, describe the role immunology plays in these infections.
Cuzick: Clearly, immunology is crucial, based on the fact that 80% to 90% of all infections are actually cleared spontaneously. Only the 5% to 10% that persist are of any concern regarding cancer. Certainly an immunologic factor is involved in sorting out its cause, although the current evidence is really quite weak.
Richart: Undoubtedly, there is a lot of interest in trying to pin this down.
Cuzick: Yes. The thing we're most sure aboutand even this is somewhat debatableis the role of human leukocyte antigen (HLA). Increasingly, studies are beginning to show rather minor effects of different HLA types. Although not strong enough to be of any clinical use, these studies do indicate a variability of response to the virus. The role of smoking is the other area where some indirect evidence is accumulating. It's likely that smoking may depress the immune response and allow the virus to have a greater chance of being persistentif it does have an effect at all.
Richart: And for the same reason, anyone who has any degree of immunosuppressionHIV-positive patients, transplant patients, pregnant women, for examplewill have trouble clearing the virus.
Cuzick: Yes. They certainly are the subgroups of immunosuppressed patients much more likely to have a persistent virus and who are more certain to have CIN disease anyway.
Richart: And the data suggest that help from the immune system is required to successfully treat a woman with an established lesion. So when we biopsy or do cryotherapy or perform a Loop electrosurgical excision procedure (LEEP), we're not only mechanically changing what happens, but we're also counting on the immune system to help us clear it. Is that true?
Cuzick: Yes. There's quite a lot of evidence, particularly with the lower-grade lesions, that simply biopsying the lesion will dramatically improve the chances of regression.
Richart: To me, the most interesting thing that's happening now is the increasing discussion about using HPV DNA testing clinically. A number of different applications have been suggested. One would be a quality assurance function; the second would be what's been called secondary triage of patients with ASCUS (atypical squamous cells of undetermined significance) and AGUS (atypical glandular cells of undetermined significance); and the third, of course, is primary screening with HPV DNA.
Dr. Meijer, what are the advantages and disadvantages of the tests now available? Which ones are clinically useful and which are not?
Meijer: The first test is the Hybrid Capture II test (Digene Diagnostics, Silver Spring, Md.), which uses specific DNA probes to detect HPV DNA. Using antibodies labelled to specifically detect the RNA/DNA hybrid, specific signal amplification is obtained. The test is the only FDA-approved commercially available test for clinical use and has a good intra- and inter-laboratory reproducibility. The Hybrid Capture II test is easy to use for routine purposes and because the signal is largely linearly related to the amount of virus, the HC II test can be used for semiquantitative detection of the virus.
Richart: And the second?
Meijer: The second is the polymerase chain reaction (PCR)-based test. One begins this test by amplifying a part of the L1 region of the viruswhich is present in all mucosal types of HPVusing two consensus primers and Taq polymerase. The amplified DNA (about 150 bp) shows enough heterogenity to type for different HPV types by specific oligomers. Using a combination of oligomers of high-risk HPV types, all high-risk HPV types can be detected in one test. The detection can be based on different formats, that is, radioactivity, ELISA format, or with line strip assay. This test shows good inter- and intra-laboratory reproducibility, but requires basic knowledge and practical experience with PCR testing.13 The two PCR systems used are the Manos system, using the My 09/11 primer system, and the Walboomers system, using the GP5+, 6+ consensus primer.14 The latter is the best clinically validated PCR test available at the moment.
Richart: People always used to talk about the Southern blot. Dr. Kjaer, is that no longer clinically useful?
Kjaer: I don't see a big use any longer for the Southern blot, which is a laborious test. As I understand it, it has been used in research studies to find those infections with a high viral load. But I think if you want to restrict your testing to high viral-load lesions in a clinical setting, you can use, for example, the HC II test insteadand just change your cut-off point. As you said, Dr. Meijer, the newer test (HC II) has been shown to be sensitive enough.
Richart: Dr. Cuzick, how do you use HPV DNA testing in a quality assurance program, both in the laboratory and clinically, if that is applicable?
Cuzick: One role of HPV testing is in conjunction with cytology on a subset of patients or on all patients. And I think the quality control issue indicates that a HPV-positive woman certainly has a greater chance of actually having a cytological abnormality. Or one of the simplest roles for primary screening would be doing HPV testing on a subset of all patients. The test would highlight a subgroup of patients for which the slides should be looked at very carefully.
Richart: Dr. Meijer, your group has had experience using HPV as a quality assurance procedure in the cytology laboratory. What are your data?
Meijer: Of about 500 Pap smears obtained from the screening pool we did, 5% of them were HPV positive. Technicians then rechecked the HPV-positive normal (negative) Pap smears as found in a routine setting for the presence of certain cervical abnormalities. Results clearly showed that 7% had definitealthough fairly minorcervical abnormalities. And if lab clinicians recheck normal (HPV-negative) Pap smears, this percentage is 1% (or a bit more), so definitely not as high as for HPV-positive normal Pap smears. So in our opinion, it's a good quality-control test in a cytological laboratory.
Richart: We see a lot of overdiagnosissquamous metaplasia called a low-grade CIN, highly immature squamous metaplasia called a high-grade CIN. What percentage should a laboratory shoot for? If you do HPV DNA with HC II on women diagnosed histologically as having low-grade CIN, what percentage of those patients should be HPV DNA positive? What percentage for patients diagnosed with high-grade CIN?
Meijer: For CIN 1, about 70%; CIN 2, 85%; CIN 3, 98%. Then the real question isas we asked in a study published in The Lancet on the role of testing for HPV as part of screeningWhat is the significance of having high-risk HPV testing in such a lesion?8
Richart: And what did you learn?
Meijer: In that study, we followed women with abnormal cytology who met certain criteria, by cytology, colposcopy, and testing for high-risk HPV.8 Our primary end point was clinical progressionwhich we defined as CIN 3, covering three or more cervical quadrants on colposcopy, or a cytologic invasive result of suspected cervical cancer. We saw clear progression only in the lesions that were high-risk HPV positive. Moreover, at the end of the study, CIN 3 was only found in women who were high-risk HPV positive at the baseline and staged high-risk HPV positive. So, when cytology is abnormal, HPV positivity is, in fact, important for detecting the presence of CIN 3 cervical cancer.
Richart: How is high-risk HPV testing best used in a clinical setting?
Meijer: First be aware that testing for all high-risk HPV types in one test is sufficient for clinical purposes. This follows from all the casecontrol studies performed until now: No indication is obtained that there is a significantly different risk for individual high-risk HPV types for cervical cancer.
Clinically, high-risk HPV testing can be applied in the following areas: (1) triage of ASCUS or AGUS; (2) detection of residual or recurrent CIN 3 after treatment for CIN 3; (3) as an adjunct to cytology for primary cervical cancer screening; and (4) primary screening in underdeveloped countries with or without vaginal self-sampling. Be aware of the age dependency of high-risk HPV testing when you wish to use it for primary cervical cancer screening.
In low-prevalence populations such as most Western countries, you find about a 10% to 15% prevalence of high-risk HPV in women between ages 20 and 25. As women pass the age of 30, this falls to about 5%. In The Netherlands, it's 3.6%. For screening to be economically feasible, prevalence should not be over 5%. HPV screening is less useful and cost effective for women under 30 years, but it is important for women over 30.
Richart: How do you follow up on women with abnormal smears?
Meijer: In the Netherlands, the classical follow up of women with mild dyskaryotic smears is a repeat smear within 6 months and when still abnormal, referral for colposcopically directed biopsy. When moderate dyskaryosis or more severe lesions are found, the women are directly referred for colposcopy. For mild-to-moderate dyskaryosis and in the area of AGUS and ASCUS, I think there is a place for high-risk HPV testing in conjunction with cytology, especially in women over age 30, since at that age high-risk HPV is an important risk factor for CIN 3 and cervical carcinoma. Two ways of follow up are possible: The first is to directly refer all high-risk HPV positive women with mild-to-moderate dysplasia or with ASCUS or AGUS for colposcopy. Women with high-risk HPV negative smears do not have to be referred because they do not progress to CIN 3 or cervical cancer.8
Although this strategy is useful and reduces the number of referrals by about 20% to 25%, I personally prefer to follow up these women by high-risk HPV testing and cytology for 6 months because it prevents further overtreatment. In a prospective follow-up study of women with cytology, we showed that only women with high-risk HPV showed clinical progression to CIN 3 or had CIN 3 at the end of follow up.8 We showed also that 50% of the women with mild-to-moderate dyskaryotic smears clear the virus within 14 months. Thus, a second high-risk HPV test in these women can be useful in selecting those women at increased risk for clinical progression and development of cervical cancer. The interval when a second biopsy should be taken6, 9, or 12 monthsshould be balanced with the risk of cervical cancer development in that period. Sensitivity (97%) and negative predictive value (99%) of a second high-risk HPV test 6 months later are much higher than those of a second Pap test (70% and 87% respectively).
Based on our results, we recommend referral for colposcopy only for those women with mild-to-moderate dysplasia who have a second positive high-risk HPV test result after 6 months. For these women, the risk of developing cervical cancer within 6 months is outweighed by the number of women who self-cure the HPV infection and the current lesion. In this way, the number of women referred to the gynecologist can be reduced considerably and overtreatment can be prevented, whereas the risk of missing a CIN 3 or higher lesion is minimal.
Richart: How do women acquire CIN lesions?
Meijer: As Dr. Kjaer stated earlier, we now know more about the natural history of an HPV infection. We know that when women become sexually active, they acquire HPV infections, but 80% of these will be transient. The women will clear the virus within a median time of about 8 months without getting any lesion. The other 20% will get CIN lesions, but only when the high-risk HPV infection persists and is not cleared will these eventually lead to cervical carcinoma in about 13 years. When the virus persists, as Dr. Cuzick said, its viral oncogenes, E6 and E7, bind with the gene-products of tumor suppressor genes, leading to irregularity in cell-cycle control and shortcomings in DNA repair. Genetic instability results and an increased risk of mutations, which are essential for progression to cervical cancer. When CIN lesions occur, they still can regress when the high-risk HPV infection is cured. This depends on a woman's immune status and the degree of atypia of the CIN lesion. So in HIV-positive women who have a poor immune response, CIN lesions do not regress. Moreover, the regression rate of CIN 3 is very low whereas it is much higher in CIN 2 and CIN 1 lesions. In fact, one can say that cervical cancer is a rare complication of a persistent high-risk HPV infection. It is comparable with the flu: Most people recover, but only a few get pneumonia.
Cuzick: Yes, and the belief in Britain is that doing HPV testing on women with high-grade cytological abnormalities is pointless because virtually all such patients will have a lesion and all will be high-risk HPV positive. Rather, those patients should all immediately have a colposcopy.
Richart: What about lower-grade lesions?
Cuzick: The issue is more complicated for both low-grade dyskaryosis (LSIL) and particularly for borderline or ASCUS smears. For ASCUS, there is good evidence that HPV will be clinically useful but very likely its use will depend on age. An older woman who's HPV positive and has an ASCUS lesion probably should be referred immediately, although that hasn't been confirmed. On the other hand, in a younger woman, we know that many HPV-positive ASCUS lesions will still be transient. The appropriate policy for these women is probably to call them back in 6 to 12 months.
Richart: What is the most useful aspect of HPV testing in low-grade or borderline cytology?
Cuzick: Its negative predictive value makes it very useful in minimizing the amount of referral. Clinicians need not refer a woman with an ASCUS smear that's actually HPV negative. And how soon she needs to be retestedshort-term versus routine follow upis really the only open question. That is still an unresolved issue (Figure 1).
Richart: Dr. Kjaer, is age important here? The message being transmitted very strongly now, as Dr. Cuzick just enunciated, is that HPV DNA testing is useful as a secondary triage procedure for patients with ASCUS, and certainly for AGUS. Is age important?
Kjaer: I believe it is. We should consider the background prevalence of HPV. And we know many of the infections in the younger age group are really transient. So in younger women, I am not convinced that triage would be as useful because I would worry about specificity. I fear that we would examine too many and would not gain a lot from using HPV.
Cuzick: But I think it may have a different use in younger women because an ASCUS smear that's HPV negative may just as sensibly be treated as a normal smear. Although I don't think an HPV-positive smear is grounds for immediate referral in younger women, it may be in older women. It is certainly grounds for more careful surveillance of younger women.
Richart: Dr. Cuzick, I understand that in England you're beginning trials to see whether you'll use this as your routine for triage of these atypical smears. Is that true?
Cuzick: Yes. Recently a systematic review of the role of HPV testing in all of its possible aspects was submitted to the government.15 The government has accepted the primary recommendation to begin using HPV testing in ASCUS and AGUS smearsand to introduce it and monitor it carefully. In the second half of 2001, three centers will begin routine HPV testing of ASCUS smears to evaluate its usefulness.
Richart: There's a Web site on which these data appear: http://www.hta.nhsweb.nhs.uk .
Kjaer: May I come back to the age once more?
Kjaer: Although I agree in principle, we know that if low grade is a morphological manifestation of HPV, you would still find a lot of HPV low-grade cells, especially in the younger women, which means nothingas they will self-cure. So should we really use the money to triage in younger women?
Cuzick: Yes. Only because I think in ASCUS, you're probably going to get about 50% HPV positive, 50% negative, depending on how your cytologist reads those Paps. And I think the value here primarily will be in the 50% that are negative who no longer need careful surveillance. I'd agree with you that the HPV positivity of the other 50% is not a cause for additional concern, but they are the patients who do need some surveillance, much as they would get now.
Meijer: I basically agree, Dr. Kjaer, that screening of women under 30 does not seem useful, because under that age the prevalence of high-risk HPV exceeds the critical percentage of 5% considerably, with decreased specificity for CIN lesions and cervical carcinoma. So it will be cost ineffective in countries like the Netherlands where the peak prevalence of high-risk HPV is at 20 to 24 years. In countries where the peak prevalence of high-risk HPV is 5 years earlier, such as Greenland, however, screening should be done 5 years earlier. Furthermore, good quality data from the Cancer Registry in Connecticut show that the prevalence of cervical cancerdespite annual screening for young womenhas not decreased in nearly 70 years. At the same time, the quality of cytologists' ability to read slides has increased enormously. And you see that the number of affected women over age 30 is decreasing.
Cuzick: Trying to monitor the trends in young women is very, very difficult. In Britain, for example, during a constant screening program, the incidence of invasive cancer in women under age 30 increased threefold probably for reasons associated with changes in sexual behaviorrather than because of screening. A current reversal of this trend is just as difficult to ascribe to screening. Multiple factors affecting trends in rare cancers make it dangerous to over-interpret those data.
Meijer: I'd suggest doing a prevalence studyand if your peak prevalence is earlieras it is, for instance, for the Eskimos in Greenlandthen screen those women 5 years earlier. First, get the basic prevalence data. But we definitely should not be screening at as frequent intervals as we are nowevery year from age 18 on.
Cuzick: I agree completely. The only justification for any screening below age 25 is that the sensitivity of the screening test currently in use (the Pap smear) is sufficiently poor to necessitate building up two or three negative screens before your patient is really negative. If you had a screening test like combination cytology plus HPV where a negative really meant negative, I would see no justification for beginning screening under the age of 25. But I think 25 is about the right starting point myself.
Meijer: I agree.
Kjaer: As do I.
Richart: Let's address the secondary triage of AGUS, atypical glandular cells. These patients are generally believed to have a higher risk than ASCUS patients, both for high-grade CINbecause the two are easily confusedand for adenocarcinoma or even invasive squamous cell cancer. Can you still use HPV DNA for secondary triage of AGUS patients? Or is this a different category?
Meijer: The diagnosis of AGUS is a big problem and reproducibility of the diagnosis is limited. It's not easily done in less than state-of-the-art labs. And as casecontrol prevalence studies done by IARC clearly show, more than 90% of these adenocarcinomas contain high-risk HPV.16 Given cytology's doubtful ability to diagnose precursor adenocarcinoma lesions, I think that high-risk HPV testing is a good way to monitor for them.
Richart: And if they are HPV negative, what would you do?
Meijer: I don't have a real solution for that because I am not aware of a test that's as good as high-risk HPV for diagnosing precursor adenocarcinoma lesions.
Richart: How about posttreatment patients? Is HPV DNA testing useful, for example, to monitor a woman with an HPV-related lesion who is treated with cryotherapy or perhaps LEEP?
Cuzick: Evidence is certainly accumulating that it ismostly still in the form of fairly small studies or anecdotal evidence like our own. But it's quite clear that if you take an HPV test at the time of biopsy and then monitor that 3 to 6 months later, the persistence of the HPV infection is a very good indicator of incomplete excision. Certainly in our experience, we've found additional disease on second look. Does absence of HPV mean patients no longer need surveillance? That's a very important possibility but unproven at this stage.
Meijer: We recently completed a study on the significance of high-risk HPV testing in women treated for CIN 3. Our standard rule in the Netherlands is that a woman with three consecutive negative Pap smears at 6, 12, and 24 months after treatment returns to the screening pool. At 3 months, a positive high-risk HPV test has twice the sensitivity for CIN 3 of a positive Pap smear and also a higher negative predictive value. At 6 months the results for high-risk HPV testing and cytology are more or less similar. Since both high-risk HPV testing and the Pap smear miss a few cases of CIN 3, we advocate follow up by both high-risk HPV testing and cytology.
Richart: The rule of thumb (in the United States at least) is after treatment, if you have three negative Pap smears, the patient can go back in the screening pool. Likewise given a negative colposcopy and two negative Paps. Do you believe, Dr. Kjaer, that a negative Pap and negative HPV DNA is a sufficient endpoint? Or is cytology even necessary? Is HPV DNA screening enough?
Kjaer: I don't think one HPV test alone is enough. We haven't done any of those studies yet. But it would also have something to do with the age of the patient. In my view you have to have repeated measurements anyway.
Richart: Dr. Meijer, HPV alone or HPV plus Paps?
Meijer: Our results clearly show that from 6 months on, both high-risk HPV testing and cytology yield similar resultsboth miss a few CIN 3 lesions. Doing both together yields very high negative predictive values. Therefore, we argued that the standard rule of three consecutive Pap smears at 6, 12, and 24 months can be changed given two consecutive negative results for high-risk HPV testing and Pap smears at 6 and 24 months after treatment.
Kjaer: But if you use HPV alone, don't you need to take into account the time interval after treatment? For example, how do you interpret results for a young woman who was HPV negative just after treatment and HPV positive, let's say, 6 months later? Is this a sign of an insufficiently treated lesion or is it a new and probably transient infection?
Cuzick: In the British program, anyone who's being followed up after treatment actually requires an annual smear for at least 5 yearsin some centers, that's for lifeas opposed to the usual screening interval of every 3 to 5 years. So anything would be useful that would enable you to decide whether additional disease is present and/or whether you can put the woman back into routine screening more quickly.
Richart: In the 1980s, we published a multicenter study of postcryotherapy patients using three negative Pap smears as an end point.17 We had almost 1,500 patients, 40,000 woman-years of follow up and 450 patients in year 14. The number of years that we continued the study enabled our results to be meaningful.
Our studywhich is the only large study I know of in the literaturefound that if a woman had three negative Pap smears, she was at no greater risk for a new abnormal smear than the general population. So I think there's really good evidence to suggest that if you have three negativesor now using modern technology if you're HPV DNA negativethere's no reason to be followed any more closely than the general population.
Meijer: Standard rule in the Netherlands: three negative Pap smears at 6, 12, and 24 months.
Richart: The conventional Pap smear, which has been the standard for 50 years, is usually quoted as having a false-negative rate of about 20%. More recently, however, fluid-based samples have been studied and at least 50 papers have now shown that Thin Prepone of several commercially available fluid-based sampleswill substantially increase the detection rate of low- and high-grade lesions and will decrease the frequency of both ASCUS and unsatisfactory samples.18
Recently, however, many publications have suggested that the conventional Pap is, in fact, a much poorer performer than we thought and the data appear to be that it has a false-negative rate of at least 50%, and in some studies as high as 60% to 70%. From these new databased on large-scale studies that are extremely well controlledthe consensus is that this is real, that the Pap does not have a 20% false-negative rate, it has at least a 50% false-negative rate, and that the fluid-based sample probably has a false-negative rate on the order of 10% to 15%.19,20
Even the best Pap sampling will have a false-negative rate of 15% in patients who have HPV-related lesions. The opinion is growing that the days of the conventional Pap smear may be numbered, that this is a technology whose time may have come and gone. Despite the Pap having been extraordinarily useful, we're now starting to look at other technologies to screen for cervical neoplasia. The principal focus is to use HPV DNA for primary screening. Why should that work?
Cuzick: The shortcomings of cytology and our knowledge that HPV is the primary cause of cervical cancer all play a part.
When we published our first HPV screening study in 1995, there was a great reluctance to accept the high false-negative results for cytology that we reported. This has now been confirmed in a number of studies. Our initial studies used a limited spectrum of HPV types as we wanted to improve sensitivity as an adjunct to cytology without dramatically reducing the specificity or positive predictive value.
Despite quite an initial reluctance in the UK to consider HPV in primary screening, the litigation issues have been an important motivator. In Britain the total cost of dealing with cytologyadding in the litigation costmakes it substantially more expensive. One of the values of HPV in this climate is that's it's not only more sensitive but results are absolutely black and white. Its more objective results avoid this interpretive issue, which is making it difficult to get British clinicians to do cytology screening in the wake of a number of scandals that have hit the press.
So its high sensitivity is among many indications for the primary screening approach, but the issue really is how best to do that without increasing the false-positive rate.
Richart: Dr. Meijer, how do you apply HPV DNA testing for primary screening? Who should be screenedand when? Can we use HPV DNA alone or combined with cytology? What does it do to the false-negative rate of cytology, which is a problem for us and for the women we serve?
Meijer: In 1992, I originally suggested at a WHO meeting in Brussels that we use high-risk HPV testing alone, augmenting it with a Pap smear afterwards for those who are high-risk HPV positive. But the argument that is always raised is that invariably a few high-grade HPV-negative CIN 3 lesions will provoke litigation if undetected.
We countered that argument by recommending combining high-risk HPV testing with cytology. That raised the cost of screening and prompted discussion on reducing that cost. We've already discussed the practice in the Netherlands of screening women over 30 as one of two possible ways to reduce costs.
Richart: And the other possibility?
Meijer: Postponing screening intervals. Our data suggest that theoretically we can increase to 10 years the screening interval of a woman who is high-risk HPV negative and has a normal Pap smear. The cost of an extra high-risk HPV test is brought down through the cost reduction achieved by increasing the screening interval. Moreover it takes about 13 years to acquire cervical carcinoma from a persistent high-risk HPV infection.21
Richart: What degree of sensitivity are we talking about?
Meijer: Using cytology and high-risk HPV testing, the sensitivity for picking up a high-grade lesion and cervical cancer is over 95%. So if a woman were tested only once every 10 years, employing such a regimen would be very useful.
Richart: What form do you think the combination testing will take? Will it be a Pap smear plus high-risk HPV test? Or will it be liquid cytology with a high-risk Pap test?
Meijer: In a recent review in the Journal of Clinical Pathology, I argued that at the moment liquid cytology is quite expensive in The Netherlands.22 Its cost is about one quarter of the maximum amount paid by insurance companies. At present, if I had to choose between liquid cytology or a high-risk HPV test, I'd choose the latter. But if the pharmaceutical companies and the doctors could negotiate a less costly screening program that combined the tests, I think that would be best for screening.
Richart: The so-called unit pricing approach consists of packaging the elements and charging a single fee for the packagealmost like an insurance policyso that it's spread across the population and used population-wide. This would be useful, would it not?
Meijer: By doing high-risk HPV testingplus cytologywe detect in our screening population about 4% of women who are high-risk HPV positive and have normal smears. And as we've already published and recently updated, if you follow these 4% for 4 years, then 8% of them in fact show CIN 3.23 Moreover by combining both tests, our false-negative rate is decreasing and early on we're picking up at-risk women who were formerly missed. (In The Netherlands we assume this represents about one quarter of the high-grade CIN lesions that were missed.)
This increase in detection rate of high-grade precursor lesions is another argument in support of high-risk HPV testing combined with cytology.
Richart: The bane of the cytologist's life and that of the gynecologist is the missed cervical cancer. And the use of HPV DNA testing either alone or in combination with a Pap smear provides an opportunity, does it not, to virtually eliminate the false negatives? Do they just go away?
Meijer: Never is not always never. Several studies show a negative predictive value of over 98% (compared to about 70% for cytology). Such high negative predictive value percentages will make it acceptable to use HPV DNA testing alone in the future followed by ctyology on the high-risk HPV positive ones.
Richart: Dr. Cuzick, if we still have some misses even with the HPV DNA, would it be worthwhile doing two consecutive tests? Would that help?
Cuzick: That's possible. Revisiting the issue of the role of cytology. I would still be more concerned about following up women who are HPV positive but cytology negative. In the UK we've recently done an extensive audit of cervical cancers in certain regions of the UK. Because the media coverage of cytology is so high, 85% of women have had a screen within the last 3 to 5 years. We're finding that half of the cancers are in women who've had a completely normal screening history and have been screened within the last 3 to 5 years.
Richart: So you are saying that one real advantage of HPV as a primary test will be its ability to pick up a substantial number of cancers that are missed by cytology?
Cuzick: Yes. Our rule from the beginning was to consider HPV as an adjunct to cytology and to do both tests. Certainly, with the very high sensitivity we're seeing, I think we're now going to pick up an awful lot more cancers by routinely adding HPV testing to cytology, possibly at longer intervals.
A key benefit of increasing the screening interval is that it can very easily pay for the cost of the test. We calculated that in Britain, increasing the interval from 3 to 5 yearseven paying for the HPV test at a total cost of £10still ends up saving £30 million a year on the screening program. So the general principle is: If you're going to screen, screen with the best available tests as infrequently as possible. A very attractive strategy is to consider using both tests while stretching the screening interval out to at least 5 years and possibly longer.
Richart: Hasn't Schiffman suggested that adding cytology to the HPV DNA test merely increases your detection by 1% to 2%? That's a lot of money for a very small gain, is it not?
Cuzick: That's another important issue. The test has evolved even as we've been doing these new studies. If, in fact, the sensitivity is as high as we think90% to 95%you could argue that the role of cytology is really quite limited and maybe we should be talking about cytological triage of HPV-positive lesions rather than the other way about. It's too early to say until both procedures are extensively evaluated as primary screening tests.
Richart: Of course, the Schiffman results are based just on women with ASCUS (LSIL) smears.
Cuzick: That's right. We can expect different results once we begin studying older women. It's too early to recommend HPV testing alone as the primary screening; We still need cytology. But we are moving in that direction. And if the results that we're seeing thus far about the very, very high sensitivity of HPV hold up in a screening context in larger studies, then we'll begin to think about a whole new approach to screening.
Richart: Which would be...?
Cuzick: One in which you might start screening at age 25 only with HPV testing and then use cytology only to decide how to manage the HPV-positive results. You'll still need a secondary test, particularly for younger women, most of whose HPV-positive lesions are transient. The clinician will turn to cytology to decide between immediate referral and surveillance.
Kjaer: I fully agree that the negative predictive value of HPV testing is extremely valuablethe value of knowing that a negative test is really a negative test cannot be overestimated. But the other side of the coin is the specificity. Even if we restrict screening to women, let's say, older than 30, where persistent infections are more prevalent, we will still see women with HPV infection who do not progress. Thus one thing we can hope for is that additional markers for persistent infection will emerge in time. Further, additional markers for progression among HPV-positive, cytologically negative women will possibly solve the "false-positivity" problem.
Richart: Dr. Kjaer, are the false positives you referred to really false positives or are these women who in fact have HPV in their genital tract and simply don't have a detectable lesion? False positive means that they don't have disease, they're detected as having disease when they don't. Is that a fair term to use for that?
Meijer: That raises the important issue of educating women that a false-positive test result is defined, in fact, as a woman who has high-risk HPV but no cellular abnormalities. We educate both the general practitioners and these women by explaining that they have acquired a common genital virus that in the large majority is cleared by the women, but rarely may lead to cervical cancer. Simply by testing these women after 1 year again by high-risk HPV testing and cytology, we can find out which of them remain at risk and need simple LLTZ and which have cleared the virus.
Richart: And you further educate a woman about her immune system's ability usually to clear the virus by itself, of course.
Meijer: Yes, explaining whether we intervene and treat her only when the virus is not cleared. So I don't know that we should label this false positive. I'd tell a woman that we can follow her up to assess self-healing of the virus but I wouldn't call it a cancer virus.
Gynecologists and general practitioners should all be educated in like manner to describe HPV to patients not as a cancer virus, but just as a virus that the woman's immune system can usually cure.
Richart: Generally, people perceive a test that's termed false positive as one that spews out so many false positives that it's not a very useful test. The facts are that it is the population of women who are Pap-smear negative and HPV DNA-positive from which virtually all the new abnormal smears come. So another value of HPV DNA testing and primary screening is that it can predict the future. It allows us to predict which women in the population need to be followed more closely. And in that sense it's not at all a false positive. It is, in fact, an important diagnostic test that has predictive value. Would you agree?
Meijer: I agree with that, yes.
Kjaer: Yes, but I'd like to see some dataand I'm not aware of anyamong older women who are HPV positive and cytologically negative. I perfectly agree that a large proportion of them will have persistent infection and the test result will put them on notice that they are at risk of developing HSIL. On the other hand, how big is that proportion? Because I would consider a transient HPV infection as a true false positive because it doesn't predict HSIL to the same degree.
Meijer: Well let's go into the figures. In The Netherlands when we screen normal cells, 3.6% of the women are high-risk HPV positive, 8% of these women we know are at risk for getting HSIL. So we are talking about 8% of 4% (0.32%) of the screening population of women between 30 and 60 years who are really at risk and the rest of this 3.6% isaccording to your criteriaa false-positive test.
Kjaer: I agree, but I still think that it's very important that before we roll out the program that we explore all the potential pitfalls. And if the proportion of transient infections is too high among those older women, then we have created a new "ASCUS" category, and although it probably will only concern relatively few women, we still don't know how to manage those women who are HPV positive but cytologically negative.
Cuzick: I agree, the perfect screening test would be positive only when there's a high-grade lesion, and for anything less than that, it would be called negative. Because if we can detect all the high-grade lesions with great certainty, then we don't need to know about anything else. So any positive result that is not immediately associated with a high-grade lesion is a problemand we mustn't forget that cytology has an awful lot of false positives as well, using that definition.
Richart: What is the "inadequate" rate in the UK?
Cuzick: It's running at about 8% to 9% in the program. Those are effectively false positives because the women must go through the whole process of being rescreened. Plus the ASCUS rate is anywhere from 3% to 5%. So we're talking about a false-positive rate in the sense that in many centers nearly 15% of women have to return. Now in older women, the positive rate for HPV is about 8% in the U.K. Because there are so few inadequate HPV tests, the false-positive rate is lower than that of cytology to begin with. The test doesn't require as good a sample. So with HPV testing, we're already talking about fewer women subjected to an additional set of formal tests than with cytology.
Richart: Semantics really do influence our thinking a lot. And calling this a false-positive test implies that something in the test has led to a failure. And again, because I think this is not a failure of the test, I say we should not use the term false positive. Instead, we should use the term predictive value, because that allows us to predict which portion of the population is, in fact, at risk. Again, what's super important about this new development is we can say negative is negative. We can say we can virtually eliminate the false negatives among the high-grade lesions and invasive cancers, and that we're able to identify that proportion of the population that is prospectively at high risk for developing an abnormal smear. That's just dynamite!
Kjaer: It comes down to the point of whether this should be considered as a screening test, and not a diagnostic test. Then you could focus on screening purposeson identifying women with an HSIL now and those who are at high risk of getting HSILthen it's easier to explain the HPV-positive/cytologically negative women. Then you don't see them as false positive.
Richart: Dr. Kjaer, Dr. Meijer mentioned the need for education. You've been heavily involved in educational efforts. How do you think this will change whom we have to educate and how we do it? What do we tell them?
Kjaer: First of all, it won't matter how good a test we have if patients and clinicians are both not knowledgeable about how its performance compares with the currently available tests.
A first step might be to make women aware of the Pap's high false-negative rate. Without knowing that, it must be hard for them to understand why we are taking a new tack. A related issue is encouraging tact and diplomacy in informing patients that HPV is a sexually transmitted disease. Presented in the wrong way, that information could be very stigmatizing, especially in older women who probably have been monogamous for years.
Richart: Do you have an educational program in the UK, Dr. Cuzick, as you roll out HPV testing?
Cuzick: We've had to face that issue in our screening studies thus far, and find that we really have to direct our attention to the GP, who is the person actually conveying test results. Quite often GPs know very little about HPV or have erroneous information. One GP I know of very insensitively told a woman she was HPV positive, causing her a lot of problems. The focus has to be on finding a way to tell a woman that she's got the virus without actually creating the stigma. There's still a lot to learn about how best to do that.
Richart: The data are certainly compelling. Are we there yet, Dr. Meijer? Are you ready to recommend introducing HPV DNA for primary screening?
Meijer: At present I am running a trial in The Netherlands with 44,000 women. One arm is HPV with cytology, the other, cytology. The HPV with cytology arm has intervention both on abnormal Pap smear and/or the presence of high-risk HPV. On the other cytology arm, HPV is detected but blinded, and our primary end point is the yield of CIN 3 and/or cervical cancer after 5 years (the usual screening interval). By referring only women with mild-to-moderate dyskaryosis, who twice have a high-risk HPV positive test after 6 months, you can also reduce the number of women referred to the gynecologist, as I've said.
Once the results of that trial are in, the government will decide on a policy. We have now taken in 20,000 women and the preliminary results suggest that high-risk HPV testing will be coming to The Netherlands after 5 years.
Richart: Dr. Cuzick, last word. Are we there yet?
Cuzick: Probably not. The study we're doing in the UK is focusing on the management of HPV-positive/cytology-negative or ASCUS smears. (Access study Web site from http://www.icnet.uk .) There is still a lot of uncertainty as to how best to manage those. Our current study of 12,000 women is actually doing HPV testing and cytology on all women over 30. And the primary issue that we're looking at is how best to manage women who are HPV positive and cytology negative. There are issues of how you convey the information, how quickly you should refer these women, how often you need to retest them, when you can actually let them go. Until we've actually learned how to manage HPV-positive/cytology negative women, I think it would be premature to introduce this as a national program. Despite a few unresolved issues, overall it's very positive.
Kjaer: We might even say that with all the evidence that we've heard now and the awareness of Pap test caveats and the more good data we can provide, and the more problems we can solve with the addition of this new screening test, it may eventually be unethical not to use HPV in this context, because again, a negative is negative. And we will detect a lot more HSILs than we did with cytology.
Richart: Sounds like we're getting close to making major changes in how we screen for cervical lesions and that the brave new world is almost at hand.
Meijer: Women are the ones who will really fare well here.
Richart: Thank you all for your time.
1. Bosch FX, Manos MM, Munoz N, et al. Prevalence of human papillomavirus in cervical cancers: a worldwide perspective. International biological study on cervical cancer (IBSCC) Study Group. J Natl Cancer Inst. 1995;87:796-802.
2. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol. 1999;189:12-19.
3. Bosch FX, Munoz N, de Sanjose S, et al. Human papillomavirus and cervical intraepithelial neoplasia grade III. Carcinoma in situ: a casecontrol study in Spain and Columbia. Cancer Epidemiol Biomarkers Prev. 1993;2:415-422.
4. Schiffman MH, Bauer HM, Hoover RN, et al. Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasia. J Natl Cancer Inst. 1993;85:958:964.
5. Kjaer SK, van den Brule AJ, Bock JE, et al. Human papillomavirusthe most significant risk determinant of cervical intraepthelial neoplasia. Int J Cancer. 1996;65:601-606.
6. Ho GY, Biermanm R, Beardsly L, et al. Natural history of cervicovaginal papillomavirus infection in young women. N Engl J Med. 1998;338: 423-428.
7. Liaw KL, Glass AG, Manos MM, et al. Detection of human papillomavirus DNA in cytologically normal women and subsequent cervical squamous intraepithelial lesions. J Natl Cancer Inst. 1999;91:954-960.
8. Nobbenhuis MA, Walboomers JM, Helmerhorst TJ, et al. Relation of human papillomavirus status to cervical lesions and consequences for cervical cancer screening: a prospective study. Lancet. 1999;354:20-25.
9. Svare EI, Kjaer SK, Smits HL, et al. RIsk factors for HPV detection in archival Pap smears. A population-based study from Greenland and Denmark. Eur J Cancer. 1998;34:1230-1234.
10. Herrero R, Hildesheim A, Bratti C, et al. Population-based study of human papillomavirus infection and cervical neoplasia in rural Costa RIca. J Natl Cancer Inst. 2000;92:464-474.
11. Kjaer SK, Svare EI, Worm AM, et al. Human papillomavirus infection in Danish female sex workers. Decreasing prevalence with age despite continuously high sexual activity. Sex Transm Dis. 2000;27:438-445.
12. Hildesheim A, Schiffman MH, Gravit PE, et al. Persistence of type-specific human papillomavirus infection among cytologically normal women. J Infect Dis. 1994;169:235-240.
13. Nindl I, Jacobs M, Walboomers JM, et al. Interlaboratory agreement of different human papillomavirus DNA detection and typing assays in cervical scrapes. Int J Cancer. 1999;71:666-668.
14. Jacobs MV, Snijders PJ, van den Brule AJ, et al. A general primer GP5+/GP6(+)-mediated PCR-enzyme immunoassay method for rapid detection of 14 high-risk and 6 low-risk human papillomavirus genotypes in cervical scrapings. J Clin Microbiol. 1997;35:791-795.
15. Cuzick J, Sasieni P, Davies P, et al. A systemic review of the role of human papillomavirus testing within a cervical screeing programme. Health Technol Assess. 1999;3:1-204.
16. Ngelangel C, Munoz N, Bosch FX, et al. Causes of cervical cancer in the Philippines: a casecontrol study. J Natl Cancer Inst. 1998;90:43-49.
17. Richart RM, Townsend DE, Crisp W, et al. An analysis of "long-term" follow-up results in patients with cervical intraepithelial neoplasia treated by cryotherapy. Am J Obstet Gynecol. 1980;137:823-826.
18. Hutchinson ML, Zahniser DJ, Serhman ME, et al. Utility of liquid-based cytology for cervical carcinoma screening: results of a population-based study conducted in a region of Costa Rica with a high incidence of cervical carcinoma. Cancer. 1999;87:48-55.
19. Fahey MT, Irwig L, Macaskill P, et al. Meta-analysis of Pap test accuracy. Am J Epidemiol. 1995;141:680-689.
20. McCory DC, Mather DB, Bastain L, et al. AHCPR Pub #99 E010. Evaluation of Cervical Cytology Evidence Report and Technical Assistance # 5.
21. Meijer CJ, Snijders PJ, van den Brule AJ. Screening for cervical cancer: should we test for infection with high-risk HPV? CMAJ. 2000;5:535-538.
22. Meijer CJ, Walboomers JM. Cervical cytology after 2000: where to go? J Clin Pathol. 2000;53:41-43.
23. Rozendaal L, Westerga J, van der Linden JC, et al. PCR-based high-risk HPV testing is superior to neural network based screening for predicting incident CIN III in women with normal cytology and borderline changes. J Clin Pathol. 2000;53:606-611.
Ralph Richart, Chris Meijer, Jack Cuzick, Susanne Kjaer. Symposium: Human papillomavirus and cervical cancer screening. Contemporary Ob/Gyn 2000;11:71-101.