October 2010
Feature Story

William Check, PhD

The world of cervical cancer screening has changed so quickly over the past two decades that it’s hard to keep up with the shifts—even for someone like Teresa Marie Darragh, MD, for whom it is the focus of her work. Liquid-based cytology, for one, has probably captured 90 percent of market share in the U.S., estimates Dr. Darragh, professor of clinical pathology and of obstetrics, gynecology, and reproductive science at the University of California, San Francisco, Mount Zion Medical Center.

Other changes: CLIA regulations for cytology labs, Bethesda System for reporting gynecologic cytology, computer-assisted screening, HPV testing, ALTS trial data, ASCCP consensus conferences on managing abnormal Paps and cervical biopsies, new screening guidelines, and, more recently, FDA approval of two vaccines for HPV. “Most recently we have seen FDA approval of an HPV genotyping test,” Dr. Darragh says. “As a result of these advances, cervical cancer screening will be completely different for young women in the next generation.”

With regard to human papillomavirus testing, Dr. Darragh’s main message: Use the tests appropriately. She calls HPV testing a “powerful and important tool for screening and managing cervical cancer and its precursors in women,” but says, “It carries a real potential for abuse.” A few simple rules can prevent most misuse. First, HPV testing should not be done in women under age 21, who have a very high rate of HPV infection but a very low rate of cancer. Second, it should not be done more frequently than once per year. “Ninety percent of HPV infections in young women are cleared in two to three years regardless of Pap interpretation,” Dr. Darragh says. Finally: “There is no place for testing for low-risk HPV genotypes.”

In the United States now, HPV primary screening is approved only in the context of co-testing—in conjunction with a Pap test—and only in women age 30 and older, says David C. Wilbur, MD, director of cytopathology at Massachusetts General Hospital and professor of pathology, Harvard Medical School. There is a lot of interest in using HPV as a primary screen to get all women who might have disease into surveillance, he says. “HPV testing is far more sensitive than Pap testing. Close to 100 percent of cervical cancer and most high-grade abnormalities are associated with high-risk HPV ­[hrHPV] genotypes.” On the other hand, Pap testing is much more specific. Trials are underway in Europe using HPV as a primary screen and doing a Pap test only when HPV is positive. “This approach probably makes sense in women over age 30,” in whom the background prevalence of HPV is markedly decreased, Dr. Wilbur says.

Though HPV testing has become much more mainstream in recent years, HPV plus Pap co-testing is “tremendously underutilized” in the United States, says Mark H. Stoler, MD, professor of pathology and of clinical gynecology and associate director of surgical pathology and cytopathology, University of Virginia Health System. Only about 30 percent of women over age 30 are getting the dual screen, he says. “It puzzles me why HPV testing hasn’t caught on,” he adds. “It is so much better in correctly categorizing a woman’s cancer risk than the Pap test, which has been shown in six major clinical trials to have a significant false-negative rate. If we designed a screening system today, women would be screened with HPV.”

Joel S. Bentz, MD, medical director of the molecular pathology and cytopathology labs at Aurora Diagnostics-LMC, Las Vegas, agrees that co-testing has been slow to enter clinical practice. “I don’t know why. I think a big part of it is education,” he speculates. What Dr. Bentz does know is that the literature is “very compelling” that HPV is the best test now a woman could have to get the highest assurance she does not have cervical cancer. Because of its very high sensitivity, HPV testing has a very high negative predictive value.

One possible explanation for the slow adoption of HPV testing is the apparent success of the Pap test. Says Dr. Wilbur: “Cervical cancer rates in the U.S. are much lower than before 1950. So the Pap test is clearly effective. Some people might ask, ‘Why mess with a good thing?’ On the other hand, you have to do it so often, which is not true of HPV testing.”

Another factor is the success of public health messages about the Pap test. “We come from this history,” Dr. Stoler says, “where we have brainwashed both patients and clinicians that a woman should get an annual Pap test. With the HPV test, we can make patients safer with significantly less frequent testing and less cost.”

Dr. Bentz does see large-scale adoption of HPV testing in another important clinical context—as a reflex test after an atypical Pap, most often ASC-US.

In addition to appropriate use, Dr. Stoler cites two other important HPV testing issues: clinical validation and predictive value. “To me, the No. 1 issue is to use a clinically validated test,” he says. “Lab-developed tests are still a major issue. This is not virology. We are trying to identify women who have precancer or cancer while reassuring the vast majority that they don’t have disease. In my opinion, that’s pretty difficult to do with a homebrew test.”

As for predictive value, he says, “It is not widely appreciated how much risk is associated with a positive HPV test.” For example, in data from the ATHENA trial, reported at the 2010 International Papillomavirus Conference, women with a normal Pap test who were positive for HPV genotype 16 on the Roche Cobas 4800 assay had a 10 percent chance of having prevalent CIN2 or CIN3 lesions, a 20-fold magnification over the rate in the general population.

Clinical validation is a critical issue, agrees Elizabeth R. Unger, MD, PhD, acting chief of the Chronic Viral Diseases Branch, Centers for Disease Control and Prevention. “It is hard to say what the gold standard for HPV testing is,” she says. “The Hybrid Capture assay is the de facto standard because Digene made it available for research early on and got a huge amount of data, which allows good clinical correlation. It’s now difficult for any other test to achieve that.” Yet, in Dr. Unger’s view, each new test should be clinically validated, which requires more than just showing comparability to an approved test. “Companies need to show not only that each new test finds HPV,” she says, “but that it predicts disease, which is our goal in cervical cancer screening.”

Testing for HPV has two main clinical applications, says Dr. Darragh, whose interest in optimal clinical application of HPV testing is reflected in her position as an attending in the UCSF Dysplasia Clinic. First is co-testing in conjunction with a Pap test as a screen for disease in asymptomatic women age 30 and over. If both tests are negative, which is the case about 90 percent of the time, a woman has less than a two percent risk of having a high-grade cervical lesion over the next several years. Guidelines recommend re-screening in three years, which is probably overly cautious, but still presents a major change from current attitudes. “Switching from the model of the annual Pap test, which is so ingrained in the minds of clinicians and patients, can present a problem,” Dr. Darragh says. Annual co-testing would waste the health care dollars that HPV testing saves. Annual visits can be useful, however, for blood pressure checks and pelvic and breast exams.

Managing a woman with positive cytology but a negative HPV result depends on the Pap interpretation. Patients with low-grade squamous intraepithelial lesion or higher are triaged to colposcopy; patients with ASC-US (atypical squamous cells-undetermined significance) should have repeat cytology in 12 months.

When cytology is negative but HPV is positive, Dr. Darragh says, “the situation gets a bit unclear.” One way of triaging these patients is to reflex test with the highest-risk HPV genotypes, 16 and 18. Patients positive for 16/18 proceed to colposcopy. Another approach is to repeat both Pap and hr­HPV in one year; if the patient is still positive for hrHPV, she is sent for colposcopy. “This second approach takes longer,” Dr. Darragh says, “but it gets the persisters.”

Consensus guidelines for managing women with abnormal cervical cancer screening tests have been published by the American Society for Colposcopy and Cervical Pathology on its website (asccp.org) and in two journals (Wright Jr. TC, et al. J Low Genit Tract Dis. 2007;11:201–222; Wright Jr. TC, et al. Am J Obstet Gynecol. 2007;197:346–355).

The second clinical application of HPV testing is as a reflex test for triage of ASC-US cytology discovered outside the context of co-testing. “No one knows what to do about this reading,” Dr. Darragh says. “These cells are worrisome, but we can’t interpret them as true squamous intraepithelial lesion [SIL].” Patients who are positive on reflex HPV testing in this situation are referred immediately for colposcopy rather than waiting and having a repeat Pap test.

Our understanding of how to manage ASC-US cytology comes largely from the NIH-sponsored multicenter ASCUS/LSIL Triage Study, or ALTS (Solomon D, et al. J Natl Cancer Inst. 2001;93:293–299). “ALTS showed that reflexing all ASC-US Paps to HPV testing and sending only HPV-positive women for colposcopy triaged about half of the women to colposcopy, rather than all, and enriched the pool of women who had disease,” Dr. Darragh says. Testing ASC-US cases for HPV resulted in 53 percent of women being sent for colposcopy while finding 92 percent of CIN3 or higher.

An alternative algorithm, repeating the Pap test six months later and sending women who still had an abnormal Pap for colposcopy, found a bit lower rate of precancerous lesions, indicating that some had regressed. “However, this result occurred later in time, so some patients were lost to followup,” Dr. Darragh says. “In the real world, many people feel that having an answer now is preferable.”

“What ALTS showed,” Dr. Wilbur says, “is that doing HPV testing on all ASC-US cytology allows you to find almost all disease but spend only half the resources.” He underlines the remarkable finding that ASC-US followed by a positive HPV test is equivalent in terms of detecting cervical pathology to LSIL cytology followed by biopsy.

Achieving these results requires an accurate HPV assay. “Many people are still using homebrew tests,” Dr. Stoler says, “which in my opinion will never do it. They don’t know the performance characteristics of their tests. Most are probably insensitive. Others, particularly some PCR-based tests, can be way too sensitive, which leads to overreferral and overtreatment.”

Dr. Wilbur stresses the need for both analytical and clinical validation of all HPV tests used in patient care, as well as the need for internal quality control and external peer comparison. He refers to the 2007 article by Dr. Stoler and colleagues for details of validation (Am J Clin Pathol. 2007;127:335–337). To satisfy the CLIA ’88 requirement for annual proficiency testing, labs can participate in the CAP’s Human Papillomavirus (High-Risk) Survey for Cytopathology and Other Laboratories. Writing in the May 2009 issue of CAP TODAY, Dr. Bentz called this Survey “the only hrHPV proficiency test available in the United States that is specific to transport/preservative media type, and [that] includes all of the commercial liquid-based cytology media used for gynecologic cytopathology.” Among 3,296 labs enrolled in the 2008 Survey, there was 98.3 percent concordance with the reference results. “We analyzed the effect of the various combinations of media and assay types and found the highest concordance rate (100 percent) for user-developed assays with all three media types,” Dr. Bentz reported in CAP TODAY. Four percent of those participating were using user-developed tests. Other intriguing findings emerged, such as that cytotechnologists perform HPV testing in 16.8 percent of labs and that performance of low-risk HPV testing decreased to 32.4 percent from the 2006 figure of 44 percent.

The FDA-approved tests are the Digene Hybrid Capture II assay (approved in 1999 and now marketed by Qiagen), which detects 13 high-risk HPV genotypes and which 82 percent of participants in the 2008 CAP Survey used; and the Cervista HPV HR and HPV 16/18 tests (approved in March 2009 and marketed by Hologic, formerly Third Wave Technologies), which detect 14 high-risk types and types 16/18, respectively, and which nine percent of participants used.

In a recently published commentary, Dr. Stoler and colleagues wrote that the Cervista assay “demonstrated excessive test positivity,” 18 percent, in its premarketing approval trial (Kinney W, et al. Am J Clin Pathol. 2010;134:193–199). Conversely, in other comparisons the Cervista assay has shown greater specificity relative to the Digene assay (Ginocchio C, et al. J Clin Microbiol. 2008;46:1641–1646; Schutzbank TE, et al. J Clin Microbiol. 2007;45:4067–4069). “Which specificity are you talking about?” Dr. Stoler says of the two studies’ findings. “Cervista may be more specific for HPV, but not for disease.”

Dr. Wilbur suspects that the higher positive rate with Cervista in the approval trial might have been a prevalence effect. “The Hologic study may have been performed in a high-prevalence population,” he says. “In some studies with the Digene assay, the background prevalence was as high and even higher.”

“HPV prevalence is very population dependent,” Dr. Unger weighs in. “People’s behavior, population locality, and sexual network can change the observed prevalence of HPV,” she says. “That makes it difficult to say with confidence that there is a problem if different tests give different frequencies in different populations. The best indicator would be head-to-head comparison.”

In fact, results of the first head-to-head comparison of the clinical performance of HC2 and Cervista HPV are becoming available now. Jerome Belinson, MD, of the Cleveland Clinic, presented the final data analysis in July at the 26th International Papillomavirus Conference in Montreal. Among more than 8,000 subjects in Shenzhen, China, HC2 and Cervista had equivalent clinical performance for detecting CIN2 and above, while Cervista HPV HR had significantly better specificity. Among women with normal cytology, the positivity rate was significantly lower for Cervista HPV, 6.0 percent, than for HC2, 7.9 percent.

Roche’s Cobas 4800 HPV test detects 12 high-risk HPV genotypes and genotypes 16 and 18 individually. It was launched in Europe in 2009 and is now under review and awaiting the FDA’s premarket approval. Final data from the ATHENA trial, reported by Thomas C. Wright Jr., MD, of Columbia University, at the 26th International Papillomavirus Conference, showed that 10 percent of women with a normal Pap test who had a positive result for HPV 16/18 probes in the Cobas 4800 HPV test had a precancerous lesion.

“PCR is a pretty standard method for detecting HPV,” Dr. Wilbur says. “It’s what they use in Europe and it’s the reference method in many clinical trials.” However, he adds, “The big problem with PCR is that all management guidelines have been developed in conjunction with the FDA-approved HC2 assay. If you increase the sensitivity, which PCR should do, you may be detecting clinically insignificant infections that will never cause any disease. So you could end up overtreating” (Hesselink AT, et al. J Clin Microbiol. 2008;46:3215–3221; Unger ER, et al. Diagn Mol Pathol. In press).

Overtreatment is more than an abstract concern. “We are now aware that our treatments can have consequences that we were not previously aware of,” Dr. Darragh says. She mentions LEEP (loop electrosurgical excision procedure), used to treat CIN3, in which a donut of tissue is deleted from the cervical os. “LEEP became a treatment of choice, largely replacing cryotherapy, because it is easy to do under local anesthesia,” Dr. Darragh says. However, patients in their childbearing years treated with LEEP can have complications in future pregnancies, including premature rupture of membranes and premature delivery. “We are now trying to balance finding all patients with cervical precancer through screening programs with trying to figure out which patients actually need treatment,” she says. “Many lesions will go away. Even some CIN3 on biopsy will go away. We don’t want to overtreat those women who will not have problems.” Unfortunately, knowledge gained from the existing tools—Pap, colposcopy, biopsy—cannot predict which lesions will progress to cancer.

“We need something more specific than presence of virus, something downstream,” Dr. Darragh says. “We are looking for that magic marker.”

Dr. Wilbur says a fair amount of new data are becoming available on biomarkers. “There is real promise here. Some of these markers can increase the specificity of the findings because they detect events more associated with neoplastic transformation of cells, rather than just HPV infection,” he says. Dr. Wilbur and his colleagues are experimenting with combinations of p16 and ki67. Initial data using this combination of markers, he says, suggest it may be a good tool for stratifying patients with ASC-US who are HPV-positive (and potentially even LSIL) with regard to who has the potential for high-grade disease. “Use of such markers might decrease the pool of patients requiring intervention further than with HPV testing alone” (Denton KJ, et al. Am J Clin Pathol. 2010;134:12–21).

Dr. Unger likes the triage potential of 3q amplifications. This abnormality is measured now by FISH, which is not good for high-throughput screening. “However,” Dr. Unger says, “if it works on another platform, it could become clinically feasible.” In addition, she says, “It’s a lot to ask a biomarker to increase specificity by itself.” Dr. Unger and her colleagues are looking at a combination of biological and clinical parameters.

One of the goals of Dr. Unger’s group is to evaluate the efficacy of the HPV vaccines. As more women become vaccinated, she predicts, “we will have to have more efficient ways of screening.” With the current algorithm, less than one-third of women sent for colposcopy are found to have disease and be in need of treatment. “This is very inefficient,” Dr. Unger says. “Unless we change our screening algorithm, we will end up with even more inefficient referral when HPV vaccination becomes more widespread.”

Scientists and physicians are exploring alternative algorithms. In studies taking place in Europe, some of which were presented in Montreal, patients are screened with the most sensitive test, hrHPV DNA, and those positive for HPV are triaged with a more specific test, such as the Pap test. “The specificity of HSIL on Pap is really good; if there is high-grade SIL on the Pap, 90 percent of the time you find CIN3 on the biopsy,” Dr. Darragh says. “So this is one paradigm shift we could see in the next decade.” Dr. Wilbur can envision such a shift, “but not for a while yet,” he says. “We will need more data and it will take a lot of time to change attitudes.”

It may appear to be premature to talk about HPV replacing the Pap test as the primary screen when only a minority of women in the United States are now co-tested. But for Drs. Darragh and Wilbur and the others who are dedicated to improving cervical cancer screening, it’s never too soon to think about the next step.

Related Links

• Are physicians following the guidelines?