Human Papillomavirus Reporting: Minimizing Patient and Laboratory Risk

The thesis of this article is that human papillomavirus (HPV) DNA testing, when used along with cytology in cervical cancer screening, represents a major opportunity to help standardize screening, increase screening test sensitivity, and increase negative predictive value such that screening effectiveness could for the first time approach levels reflecting high public expectations. Accordingly, supplemental HPV testing offers a major opportunity to decrease risk both for the patient and for the laboratory. A background perspective of this discussion is a local experience in Charleston, SC, in which more than 100 litigated cervical cancer cases from around the country have been reviewed since 1994 through the Center for Quality Improvement in Gynecologic Cytology at the School of Allied Health Professions of the Medical University of South Carolina. Case material has been continuously assessed for lessons helpful to ongoing risk management and patient safety enhancement efforts. This unique case material has consistently indicated that the clinically critical target lesions requiring cytologic identification are often difficult to identify or interpret cytologic manifestations of high-grade intraepithelial lesions (HSIL and adenocarcinoma in situ) and frequently equally difficult to recognize manifestations of invasive cervical carcinoma, especially adenocarcinoma.

Broadly speaking, risk management in cervical screening can be thought of as an attempt to lessen the difference between what the public expects from cervical cancer screening and what can actually be achieved with current methodologies. If that difference can be minimized or eliminated, then risks to both the patient and the laboratory will have been decreased. Public expectations of cervical screening effectiveness have long been unrealistically high. The reasons for this discrepancy are historical and give some insight into how rapidly a new technology like HPV DNA testing might be widely adopted. George Papanicolaou, MD, first presented his smear technique at a January 1928 conference in Battlecreek, Mich. The title of his paper was “New Cancer Diagnosis.” A New York newspaper article covering the meeting appeared on January 5, 1928, and stated:

As events played out, Papanicolaou was keenly disappointed that this preliminary paper failed to arouse interest among his clinician colleagues. He essentially abandoned investigation of the method until more than 10 years later, when a new chairman of the Cornell Medical Center Department of Anatomy and Physiology, Joseph Hinney, MD, urged him to resume the research with a gynecologist coinvestigator, Herbert F. Traut, MD. This collaboration led to their landmark publication in 1941.2 Following extensive early educational efforts, a strategic alliance with the newly named American Cancer Society was instrumental in widely introducing the method into clinical practice. But it was not until 1955, more than 25 years after Papanicolaou's initial report, that the first population studies by Erickson and colleagues3 in Tennessee and Christopherson et al4 in Kentucky established the validity of gynecologic cytology. The combination of these well-publicized early successes and sustained public education efforts by the American Cancer Society led both to continued widespread screening and to very high public expectations for test effectiveness.

The US cervical screening risk management landscape changed dramatically on November 2, 1987, with the publication of a Wall Street Journal article by Walt Bogdanich, entitled “Lax Laboratories: The Pap Test Misses Much Cervical Cancer Through Labs' Errors.” 5 This article brought to public attention for the first time the realization that Papanicolaou (Pap) tests may be falsely negative. Although the limitations of the valuable smear method had long been recognized in the scientific literature, the public up to that point had been largely unaware that there were any significant limitations to Pap test screening. The implication of the article that laboratory errors due to carelessness were the major problem led directly to the Clinical Laboratory Improvement Amendments of 1988 (CLIA '88), which legislated unprecedented regulatory intrusions into a specific area of medical practice, gynecologic cytopathology.6 Poorly understood rules in CLIA '88 were almost immediately reflected in dramatic increases in cervical screening litigation.7 Impossibly high public expectations for screening effectiveness were reflected in a 1997 trial lawyer's newsletter from Seattle, Wash, which stated, “If a woman develops cervical cancer and undergoes a hysterectomy or dies, there is almost certainly a claim for medical malpractice against some health care provider unless the woman utterly failed to get even periodic Pap smears.” 8 The unspoken implication of this outlook was that even a single periodic Pap smear should be the equivalent of a cervical cancer insurance policy that will pay a substantial monetary benefit if cervical cancer were ever to develop. Unappreciated, however, was that this view was incompatible with sustained availability of the Pap test as a historically low-priced, affordable mass screening test. Unrealistic expectations also extended beyond patients and trial lawyers to supposedly well-informed professional groups and individuals. For example, as recently as mid-2002, the Centers for Disease Control Breast and Cervical Cancer Screening Web site flatly stated, “Cervical cancer is nearly 100% preventable. Screening younger women using the Pap test is an important strategy that can prevent cervical cancer almost 100% of the time.” 9 The significant degree of both public and professional confusion about Pap test effectiveness and the lengthy period of time over which these misperceptions have hardened warn that expectations are not likely to relax significantly.

Issues surrounding test cost have also been closely linked to risk. Public health professionals and third-party payers have tended to favor limiting the cost of mass screening methods to afford wide screening coverage. Large independent laboratories discovered that the underpriced “loss leader” Pap test was an effective method to increase market share, and clinicians who marked up inexpensive Pap tests found a major source of revenue to help defray office overhead. Unnoticed for a good while was the reality that these practices undermined maintenance of quality laboratory practices and quality improvement efforts, even while shifting risk to the patient. A review article in the Journal of the National Cancer Institute reflected the idea that test cost had well-defined limits of reasonableness in concluding that about 30% of lost life expectancy caused by cervical cancer deaths is not preventable with “efficient screening policies.” 10 The public, in contrast, has been generally unaware of the fact that screening policies may actually regard a very significant portion of cervical cancer cases as unpreventable. In the current environment, patients and laboratories have become major holders of risk, based on assessments of “low” and “acceptable” risk arrived at by other unaccountable third parties.11 In the tort system, unlike in the public health system, the sole focus is on the rights of the individual, and the primary question is whether duties and obligations of health care providers to the individual patient have been faithfully fulfilled. As a result, cervical cancer screening–related liability now represents a disproportionately high share of all pathology issues and a major risk management challenge for cytopathology laboratories.12 If liability costs were born solely by the cytology laboratory and not shared by all areas of the laboratory, the increased burden could arguably destabilize the current US screening infrastructure. Currently, several daunting risk management issues face the gynecologic cytopathology laboratory, including (1) public expectations of 100% screening effectiveness, (2) good but less than 100% effectiveness, (3) economic barriers to more effective and more costly technology, and (4) rising incidence of difficult-to-prevent endocervical adenocarcinomas.

The National Institutes of Health Atypical Squamous Cells of Undetermined Significance (ASCUS)/Low-Grade Squamous Intraepithelial Lesion (LSIL) Triage Study (ALTS) is a major development in the field of cervical screening and the only currently available randomized, controlled, multicenter clinical trial (class I evidence) on which to base decisions about many aspects of cervical screening policies.13 Data from ALTS, including data in soon-to-be-published longitudinal follow-up studies, were substantially relied on to formulate “Consensus Guidelines for the Management of Women with Cervical Cytological Abnormalities” 14 and could impact screening policies in the future as profoundly as the Wall Street Journal article did almost 15 years earlier. The trials point to a number of important conclusions that are highly relevant to risk management and patient safety efforts. Furthermore, since ALTS used specific methods of liquid-based cytology (ThinPrep) and HPV DNA testing (Hybrid Capture 2) and data derived from use of these specific methods are integrally reflected in trial outcomes, Stoler15 asserted:

This standard is a significant hurdle for competitive test methods to address and for laboratories to consider in selecting testing methods.16 

A major conclusion of the American Society for Colposcopy and Cervical Pathology (ASCCP) Consensus Guidelines is that reflex HPV DNA testing is the preferred approach for the most prevalent form of cervical abnormality (ASCUS) when either liquid-based cytology is used or when co-collection for HPV DNA testing can be done. The substantial clinical advantages of reflex HPV testing include (1) immediate testing to determine need for colposcopic referral, (2) no 4- to 6-month follow-up delay for repeat testing to reassess the possible need for colposcopic referral, (3) higher negative predictive values for negative HPV DNA test results than for repeat cytology, (4) presumed lower loss to follow-up compliance, (5) early detection in patients with ASCUS Pap tests of underlying 5% to 17% rates of biopsy-confirmed HSIL and 0.1% to 0.2% rates of undetected invasive carcinoma without 100% colposcopic referral, and (6) cost-effective detection of clinically significant lesions compared to other methods. For the prospectively unknowable subset of patients with clinically significant high-grade and invasive lesions, the issue of time to follow-up becomes highly significant. Studies have shown that because of the prevalence of ASCUS as an interpretation, a very significant proportion of high-grade, biopsy-proven cases derive from patients presenting with ASCUS Pap tests (as high as 39% in one study from Kaiser Permanente).17 An important observation of ALTS data is that HPV triage is at least as sensitive if not more sensitive than 100% colposcopic referral for detection of HSIL or higher (HSIL+).13 

Observations from ALTS parallel other recent data from the important Shanxi Province Cervical Cancer Screening Study by Belinson's Cleveland Clinic group.18 This remarkable study is unique in that it is the only modern study that has assessed the prevalence of significant disease in a population by 100% biopsy of all patients, including all patients with negative Pap test results. Because of this study design, the true prevalence of significant disease (HSIL+) was as closely assessed as is scientifically possible. All women with negative colposcopic examinations underwent blind 4-quadrant cervical biopsies and endocervical sampling. The study results, which paralleled those in ALTS, showed that (1) HPV testing may be more sensitive than colposcopy in detecting HSIL+, and (2) high levels of cytology sensitivity comparable to HPV test sensitivity occurred only when the cytologic threshold for abnormality was lowered to ASCUS. Higher LSIL or HSIL thresholds for abnormal cytology resulted in lower sensitivity. The investigation also documented sensitive detection of biopsy-proven HSIL+ lesions using ThinPrep liquid-based cytology at rates (94% for CIN 2 and higher and 98% for CIN 3 and higher) significantly higher than previously reported using the conventional Pap test. The ALTS and Cleveland Clinic study findings have direct implications for recommendations to clinicians for optimal testing. In our laboratory, where we emphasize follow-up recommendations on our abnormal Pap reports as a routine policy, most ThinPrep ASCUS cases carry the following follow-up suggestion: “Consider preferred immediate HPV DNA testing, diagnostic colposcopic referral, or repeat Pap testing in 4–6 months.” 13 This educational note tries to alert the clinician to the preferred status of reflex HPV DNA testing for ASCUS cases and also acknowledges the ASCCP Consensus Guidelines. The preferred method is also the most cost-effective alternative to achieve the lowest possible risk.

Availability of adjunctive HPV testing is raising other significant risk management issues that are just beginning to be recognized. The 2001 Bethesda System publication states “pathologists are encouraged to judicially downgrade to negative a portion of the cases previously termed ASCUS favor reactive.” 19 Under the new Consensus Guidelines, any incorrectly downgraded case will not only likely have delayed follow-up, but also will not undergo HPV DNA testing. Such downgraded negative results could potentially cause patients to wait several years for another test under some testing formulations. Available data on the reproducibility of atypical squamous cells, both ASC-US and the newer ASC-H (atypical squamous cells, cannot exclude HSIL), have shown low or poor levels of interpretive reproducibility.20 In the ALTS trial itself, only 55% of the cytology specimens originally diagnosed as atypical squamous cells were subsequently given a diagnosis of atypical squamous cells by the pathology quality control group. In the ASCCP Consensus Guidelines, almost all abnormal categories of Pap test interpretation (cancer, atypical glandular cells, HSIL, ASC-H, and LSIL) are to undergo diagnostic colposcopic referral.14 The only category that is not necessarily referred to colposcopy is ASCUS. Furthermore, now there is an immediately available method of additional testing that can determine, with the same sensitivity as universal colposcopic referral, whether a patient is likely to have a significant cervical lesion. Accordingly, addition of a directive educational note on Pap test reports provides an important reminder to the clinical caregiver of preferred follow-up options.21 

The negative category remains the most problematic risk management challenge in cytology, as all studies have shown that some portion of Pap test–negative patients harbor significant undetected abnormalities. Patients with false-negative test results generally fail to have early follow-up and referral for diagnostic testing performed routinely, as is the case for patients correctly assessed as abnormal. Most false-negatives are thought to be due to sampling problems, which can be minimized by liquid-based cytology,22 but another significant proportion are due to screening and interpretation errors. What level of error is commonplace for screening and interpretation functions? The 1998 AutoPap clinical trial data probably provide the most objectively derived estimate, and this study showed in the current practice arm that the false-negative rate from 5 study sites averaged 7% for HSIL+ cases to 21% for ASCUS+ cases.23 These are not inconsequential inherent false-negative rates with current methods, and risk management strategies must take this factor into account. A mathematical model assessment of the AutoPap clinical trial data has concluded that the percentage of abnormal slides that can be reproducibly interpreted is likely quite small, probably less than 10% for any single target abnormal interpretation.24 Papanicolaou testing to prevent cervical cancer is a complex, fallible, multistep process that requires accurate sampling, screening, interpretation, and follow-up. Standardization of each step or steps in the process helps to minimize error. Interpretation appears to be the most difficult step to standardize.

The natural history of HPV infection and the biology of HPV in cervical precancer and cancer suggest intuitively that HPV DNA testing may be a helpful adjunct to screening.25 For example, a quality control study described in 1997 by Meiijer et al26 rescreened negative Pap tests blinded to HPV status and found that 7% of HPV-positive smears were found on rescreening to have cytologic abnormalities, compared to less than 1% for negative smears. The potential value of combining HPV DNA testing with cytology is the concept behind the recent March 2002 proposal for the DNA Pap test. This conditionally approved Food and Drug Administration (FDA) submission proposed expanded use of HPV Hybrid Capture testing to include its use “as a general population screening test for women age 30 or older in conjunction with the Pap test as an aid to determine the absence of high grade disease.” 27 Management and counseling of women who are HPV positive and cytology negative remains the major area to be addressed. Available studies using both sensitive PCR methodology28 and Hybrid Capture 2 testing29 indicate that patients with positive high-risk HPV DNA testing and negative cytology have high rates of development of HSIL. Most such individuals therefore do not represent false positives, but rather are true high-risk patients. In contrast, patients who test “double negative,” that is, negative for both HPV DNA and cytology, are in the lowest possible risk category. The findings in the Shanxi Province study are instructive in this regard:

The high negative predictive value of the double-negative result was also reviewed extensively and was supported with other unpublished data at the March 2002 FDA submission proposing the DNA Pap.27 The sensitivity of the double-testing method approaches the 100% expectations of the public, surpassing even the 93% to 94% sensitivity for HSIL+ lesions found in 2 overseas studies reviewed earlier using liquid-based ThinPrep18,30 and even more substantially exceeding the 70% sensitivity estimated for the conventional Pap smear for true precancer (HSIL) lesions.30–32 

Data reviewed in the ASCCP Consensus Guidelines also indicate that HPV testing can be used as a valid marker in a variety of follow-up settings, both as a follow-up of women not treated for atypical squamous cells or LSIL and for postcolposcopy management of women with LSIL, HPV-positive ASCUS, and ASC-H.14 In these settings, 2 Pap tests 6 months apart are regarded as an alternative to a single HPV test at 12 months, reflecting the lower sensitivity of repeat cytology in ALTS.33 Other studies indicate that HPV testing may be useful in following treated postconization HSIL patients for difficult-to-detect evidence of HSIL+ recurrence high within the endocervical canal.34,35 

Endocervical adenocarcinoma is a major risk management challenge because of the large discrepancy between the scientific capability of current screening methods and public expectations of perfect screening test performance. A variety of studies have shown that adenocarcinomas are increasing as a portion of cervical cancer,36 and current estimates are that adenocarcinomas comprise around a third of all cervical cancer cases.37 Furthermore, endocervical adenocarcinomas now in our experience represent more than half of all litigated cases of cervical cancer. Available international population studies, however, have concluded that that there is no convincing evidence that the conventional Pap test is an effective screening tool for decreasing the incidence of adenocarcinoma.38,39 In our experience, a common Pap test interpretation preceding some of these tragic cases has been “atypical endocervical cells favor reactive.” The “favor reactive” modifier has not been allowed in our laboratory for several years because of the misleading mixed message it conveys to clinicians. Accordingly, the “favor reactive” modifier for atypical cases was appropriately discontinued in the 2001 Bethesda System.19 The reproducibility of atypical glandular cells of undetermined significance/atypical glandular cells interpretations, like that for ASCUS, is low or poor.40 Human papillomavirus testing appears to offer a useful additional method to cytology for detection of adenocarcinomas. Ronnett et al41 in a Kaiser Permanente study detected all 5 adenocarcinoma in situ cases out of a much larger group of women referred with atypical glandular cells.41 The overall literature on adenocarcinoma by non–FDA-approved PCR methods of HPV testing also suggest that a very high percentage of these cases will test HPV positive.42 In our laboratory, the ASCCP Consensus Guidelines for atypical endocervical glandular cells are emphasized in the following report format: “Atypical endocervical glandular cells; cannot rule out a significant glandular intraepithelial lesion; recommend colposcopic evaluation with endocervical sampling; suggest endometrial sampling for women aged 35 and older.” 14 Our laboratory also routinely explicitly suggests HPV DNA testing, not as an alternative to recommended diagnostic evaluations, but rather as additional information to help assess subsequent follow-up options along with results of initial diagnostic studies. Follow-up recommendations on atypical glandular cell Pap reports may have special merit, as a recent study documented that even in an academic center, about half of gynecologists did not follow consensus follow-up guidelines.43 Human papillomavirus testing as an adjunct to screening has potential usefulness because of its high sensitivity. Human papillomavirus testing also has the advantage of helping to standardize the screening and interpretation process, while not constituting additional durable evidence, which is always subject to biased reinterpretation. Conventional Pap screening for glandular neoplasms, in contrast, has all of the performance limitations that have been discussed.

Whenever new technology methods have been proposed to enhance cervical screening and decrease risk, objections have inevitably been raised regarding cost-effectiveness and whether it is more practical to expend resources to recruit additional women for screening, rather than to use new and more sensitive screening methods. In a thoughtful evaluation of cervical cancer cost-effectiveness issues, Brown et al44 caution:

The authors go on to acknowledge, “There is no fixed threshold for what is cost-effective.” 44 Cost issues are certain to play a major role in adopting reflex HPV testing for ASCUS cases and in the possible adoption of the DNA Pap test. A compelling model study by Montz et al45 asked what could be accomplished if a screening test were available that could decrease the false-negative rate by 45%. The model evaluations described by Montz et al suggested that US Healthy People 2010 goals of decreasing cervical cancer incidence from around 12 per 100 000 per year to 8 per 100 000 were achievable with more sensitive technology, even at current rates of recruitment to screening. The model also suggested that if improved recruitment to screening were added to increased sensitivity, Healthy People 2010 goals could actually be exceeded.45 Available data on the DNA Pap test are promising when considered in the context of the model described by Montz and colleagues.

In conclusion, use of HPV testing as an adjunctive test with indeterminate atypical Pap interpretations or as a routine cotest with cytology offers the promise of at last approaching sensitive and effective detection of precancers and cancers at levels expected by the public. Longitudinal studies of various test combinations and screening intervals are needed for testing optimization. Resistance to payment for another improved cervical screening method will probably be the greatest challenge in realizing the potential benefit of the new test.