Our objective was to review current large studies of human papillomavirus (HPV) DNA testing as an adjunct to the Papanicolaou test for cervical cancer screening programs. We analyzed 10 large screening studies that used the Hybrid Capture 2 test and 3 studies that used the polymerase chain reaction test in a manner that enabled reliable estimates of accuracy for detecting or predicting high-grade cervical intraepithelial neoplasia (CIN). Most studies allowed comparison of HPV DNA and Papanicolaou testing and estimates of the performance of Papanicolaou and HPV DNA as combined tests. The studies were selected on the basis of a sufficient number of cases of high-grade CIN and cancer to provide meaningful statistical values. Investigators had to demonstrate the ability to generate reasonably reliable Hybrid Capture 2 or polymerase chain reaction data that were either minimally biased by nature of study design or that permitted analytical techniques for addressing issues of study bias to be applied. Studies had to provide data for the calculation of test sensitivity, specificity, predictive values, odds ratios, relative risks, confidence intervals, and other relevant measures. Final data were abstracted directly from published articles or estimated from descriptive statistics presented in the articles. In some studies, new analyses were performed from raw data supplied by the principal investigators. We concluded that HPV DNA testing was a more sensitive indicator for prevalent high-grade CIN than either conventional or liquid cytology. A combination of HPV DNA and Papanicolaou testing had almost 100% sensitivity and negative predictive value. The specificity of the combined tests was slightly lower than the specificity of the Papanicolaou test alone, but this decrease could potentially be offset by greater protection from neoplastic progression and cost savings available from extended screening intervals. One “double-negative” HPV DNA and Papanicolaou test indicated better prognostic assurance against risk of future CIN 3 than 3 subsequent negative conventional Papanicolaou tests and may safely allow 3-year screening intervals for such low-risk women.

Human papillomaviruses (HPVs) are genetically diverse with more than 100 types that infect different body sites, resulting in a variety of disease manifestations. A specific group of HPVs, referred to as the high oncogenic risk types, are now recognized either individually or in combinations as the necessary but insufficient cause of cervical cancer.1 Ninety-five percent to 100% of cervical cancers have detectable HPV DNA,2–6 and only a very small minority of cervical cancers are now believed to arise by non–HPV-related mechanisms. Although virtually 100% of the most common cervical cancers appear to have an HPV causality, it is extremely difficult to prove that point formally because of the possibility of some level of HPV contamination of cancers from other genital tract lesions and because in some cases HPV DNA may be lost from cancerous cells after the initiation and promotion events. The HPV-attributable proportion of cervical cancers is currently estimated at more than 90%.1 

The Papanicolaou test has been the mainstay of cervical cancer prevention for the last 50 years and has resulted in a dramatic lowering of cervical cancer incidence and deaths in many countries. Despite the success of cytology-based screening programs, the estimated true sensitivity of the conventional Papanicolaou test is on the order of 50% to 60% in the routine screening setting.7–9 The impressive reduction in cervical cancer incidence is the result of carefully orchestrated Papanicolaou test programs involving repetition of the test annually or every few years. Recent improvements in the Papanicolaou test, such as the introduction of liquid-based cytology, have increased the test's sensitivity and may allow screening to be undertaken less frequently than with a conventional Papanicolaou test. However, even such improved cytology tests may miss 15% to 35% of cervical intraepithelial neoplasia (CIN) 3 or cancer (collectively termed CIN 3 in this review) in a routine screening setting.10,11 

Extensive data show that HPV DNA is a good marker for women at risk of neoplasia1–3,12,13 and that HPV DNA testing may be considered as an adjunct to cervical cytology in cancer prevention programs, either as a management or screening tool. The potential uses for HPV DNA testing include triage of patients with atypical squamous cells of undetermined significance (ASCUS, ASC-US) Papanicolaou tests; resolution of discordant cytology, colposcopy, or histology findings; follow-up after treatment for “test of cure”; follow-up after normal colposcopy; or population screening as a primary test or as an adjunct to the Papanicolaou test.

The value of HPV DNA triage for ASC-US is now clear and convincing, as described in the recent Atypical Squamous Cells of Undetermined Significance/Low-Grade Squamous Intraepithelial Lesion Triage Study (ALTS)11,14 and the new American Society for Colposcopy and Cervical Pathology guidelines for ASC-US triage.15 Human papillomavirus DNA testing is the preferred test for quickly resolving this vexing category of Papanicolaou tests, which number more than 2 million in the United States each year.

The rationale for the use of adjunctive HPV DNA testing in screening applications is based on the increasingly accepted concept of necessary causality and on the basis of the very high negative predictive value (NPV) of the combined HPV DNA plus Papanicolaou tests, typically 99.9% to 100%. Data in support of combined HPV DNA and Papanicolaou testing are presented in this review. The main reason to screen with both HPV DNA testing and cytology is that most of the small number of cases of CIN 3 missed by the HPV DNA test can be detected by the Papanicolaou test, providing a more reassuring screening result. In the future, if HPV DNA tests or tests using other markers with a virtually 100% NPV can be developed or if cost-effectiveness considerations allow for some missed cancers, then the Papanicolaou test may be performed only on HPV DNA or marker-positive women. This would permit efficient selection of those women who need immediate colposcopy from those who are at lower risk and who could be followed up instead by more frequent screening for future disease.

The HPV DNA tests considered in this review were either the Hybrid Capture 2 (HC2) test or a polymerase chain reaction (PCR) test performed in an expert laboratory. The HC2 test is a standardized test approved by the Food and Drug Administration (FDA) that has been used extensively in research studies and has been in routine clinical use for more than 3 years. There are currently no FDA-approved PCR tests for HPV DNA, and it was necessary for this review to rely on tests that were established in certain laboratories with demonstrable extensive previous experience and installed quality assurance procedures.

Hybrid Capture 216 can detect 1 or more of 13 oncogenic-risk HPV types (16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, and 68) at the level of 1 pg/mL each, which corresponds to 5000 HPV genomes per test well. All HC2 technology operates on the principle of signal amplification and thus requires minimal specimen preparation. Interference from materials present in clinical specimens that may cause inhibition or negative results in other tests, such as PCR target-amplification-based procedures, is not a concern with HC2. The HC2 test also does not suffer from problems of amplicon contamination that can lead to false-positive results with PCR in poorly controlled laboratory settings.

Polymerase chain reaction17,18 relies on the enzymatic amplification of HPV DNA to allow the detection of very low levels of HPV infection. The test has excellent performance in appropriately equipped and experienced laboratories.19 The most common PCR systems in use are those based on the MY09/MY11 primers18 and the GP5+/6+ primers.17 This review included only studies using these methods.

All studies were approved by local institutional review boards, and almost all were conducted with written subject consent. We focused mainly on studies that collected exfoliated cervical cells according to validated procedures, but included 2 important longitudinal studies that used alternative collection strategies.12,20 Such specimens were collected from the transformation zone by several different techniques. Most commonly, a small conical brush was used (Hybrid Capture Cervical Sampler, Digene Corporation, Gaithersburg, Md); the brush was rotated 3 times in the cervical os and then placed into a 1-mL tube of transport medium. Some studies, however, used a standard plastic spatula and cytobrush combination21 (Cooper Instruments, Hollywood, Fla) or a broom device22 (Cervex-Brush, Unimar, Wilton, Conn) to collect specimens for deposition into 20 mL of PreservCyt (Cytyc Corporation, Boxborough, Mass) liquid cytology medium. None of these collection methods required refrigeration of the specimen for at least several weeks and thus were reliable and convenient for the studies. Hybrid Capture 2 and PCR tests were performed either from the Digene transport medium specimens or from the residua of specimens used for liquid-based cytology examinations. In the study by Kjaer et al,20 PCR testing was performed on specimens collected into phosphate-buffered saline. It is notable that HPV DNA testing in the Portland, Ore, study was conducted on cervicovaginal lavage specimens, a technique that is not as good for sampling the endocervical canal as direct brushing of the os. This specimen limitation is likely accentuated in older women who have a higher frequency of receding transformation zone and stenotic os23,24 and thus may have compromised the HPV DNA test data.

For our analyses, we initially focused on large screening studies of 1000 women or more that employed HC2 or PCR in a manner that allowed reliable estimates of accuracy for detecting high-grade CIN or cancer. We excluded studies that investigated poorly characterized populations of women or that had more than a minor proportion of special groups, such as sexually transmitted disease or hospital clinics, unless convincing evidence was supplied that these groups were representative of the larger screening population in the general locale of the study. We were interested only in studies that had a well-described, high-grade cervical disease reference standard rendered by expert pathologists or panels of reviewers. For simplicity, we indicate cervical disease reference standards as either CIN 2/3 or CIN 3. Cancers were included in these groups because they constituted less than 10% of the total number of high-grade intraepithelial lesions in all studies, and the results were essentially unaffected by the presence or absence of the cancers. Most studies presented performance data on concurrent cytology that allowed for comparison of the 2 tests, as well as estimates of the performance of cytology and HPV DNA as combined tests.

Final selection was on the basis of a sufficient number of cases of CIN 2/3 for meaningful statistical values. Twenty cases of CIN 2/3 were chosen as the minimum number to qualify for this review. Investigators had to demonstrate the ability to generate reasonably reliable HC2 or PCR data that were either less biased by nature of study design or that permitted the use of analytical techniques for addressing issues of study bias. The studies had to provide data for the calculation of test sensitivity, specificity, predictive values, confidence intervals, and other relevant statistical measures. Selected studies were those that used HC2 testing by formally trained laboratory staff or PCR data from recognized expert laboratories that specialized in such tests, as demonstrated by extensive publications. Salient details of the selected studies are shown in Table 1.

Table 1.

Human Papillomavirus (HPV) DNA Screening Studies Fulfilling the Selection Criteria

Human Papillomavirus (HPV) DNA Screening Studies Fulfilling the Selection Criteria
Human Papillomavirus (HPV) DNA Screening Studies Fulfilling the Selection Criteria

Final data were abstracted directly from the publications or estimated from data tabulations in the articles. For some studies, raw data furnished by the principal investigators were analyzed specifically for this review. Occasionally, the numbers of women or the number of cases of disease presented here differ from those highlighted in the articles. In such instances, we chose subsets of patients from the studies that better fit our presentation criteria.

We use the commonly understood meanings of clinical sensitivity (proportion of test-positive women among women with disease) and clinical specificity (proportion of test-negative women among women without disease) and refer to them simply as sensitivity and specificity. The terms relative sensitivity and relative specificity have been suggested for estimates not corrected or free of verification bias. For simplicity, we chose to not adopt that terminology, because we feel that verification bias is not an important consideration in the majority of our selected studies. Some studies corrected for verification bias and are indicated in Tables 2 and 3, but data for the corrections were not well described. Corrections should be based on very large groups of extensively characterized women with double-negative test results, and there is a danger of erroneous correction with small control groups; hence, we feel that uncorrected results are also valuable.

Table 2.

Cross-Sectional Papanicolaou Test and Human Papillomavirus (HPV) DNA Test (Hybrid Capture 2) Screening Studies. Data Are for Women of All Ages (16 to Older Than 80 Years of Age) Unless Otherwise Indicated*

Cross-Sectional Papanicolaou Test and Human Papillomavirus (HPV) DNA Test (Hybrid Capture 2) Screening Studies. Data Are for Women of All Ages (16 to Older Than 80 Years of Age) Unless Otherwise Indicated*
Cross-Sectional Papanicolaou Test and Human Papillomavirus (HPV) DNA Test (Hybrid Capture 2) Screening Studies. Data Are for Women of All Ages (16 to Older Than 80 Years of Age) Unless Otherwise Indicated*
Table 3.

Papanicolaou Test and Human Papillomavirus (HPV) DNA Screening Studies by Polymerase Chain Reaction for Carcinogenic HPV Types*

Papanicolaou Test and Human Papillomavirus (HPV) DNA Screening Studies by Polymerase Chain Reaction for Carcinogenic HPV Types*
Papanicolaou Test and Human Papillomavirus (HPV) DNA Screening Studies by Polymerase Chain Reaction for Carcinogenic HPV Types*

Our analytical techniques included estimates of sensitivity, specificity, positive predictive value (PPV; proportion of women with disease among women with positive test results), and NPV (proportion of women without disease among women with negative test results) from contingency tables. Confidence intervals were determined by exact methods. Odds ratios in longitudinal studies were as reported (Danish study). Cumulative incidence was estimated by Kaplan-Meier methods, and relative risk values were obtained by maximum likelihood estimation techniques (Portland study). Statistical software used for the analyses were SPSS 11.0 (SPSS, Chicago, Ill), StatExact (Cytel, Cambridge, Mass), Instat (GraphPad, San Diego, Calif), custom software as needed, and a number of other programs that are described in the source publications.

Table 1 lists general characteristics of the studies. With respect to HPV DNA testing, 10 of the 12 studies used HC210,21,22,24–28,30,31 and 3 used PCR,10,20,29 of which the Seattle study provided both HC2 and PCR data.10 Ten studies provided cross-sectional data. Figure 1 shows a simplified protocol for 9 of the studies, which is intended as a rough approximation to assist visualization (the reader is advised to consult the original articles for full details of each protocol). A 10th study in Shanxi Province, China,21 referred all women to colposcopy. Each woman in this study had multiple biopsies regardless of the colposcopic impression; hence, it is the only study that by design can be assumed to be mostly free of verification bias. Three studies provided longitudinal data (Copenhagen, Denmark20; Portland, Ore24; and Reims, France28). Figure 2 shows a simplified protocol for the Portland study. Table 2 provides salient data from HC2 studies, Table 3 shows the PCR studies, and Figure 3 summarizes selected data from the HC2 cross-sectional studies. Figure 4 shows the cumulative incidence of CIN 3 in the Portland study for 3 testing modalities.

Figure 1.

Generalized protocols for cross-sectional components of human papillomavirus (HPV) DNA screening studies, with the exception of the Shanxi study, in which all women received 4-quadrant colposcopic biopsies

Figure 1.

Generalized protocols for cross-sectional components of human papillomavirus (HPV) DNA screening studies, with the exception of the Shanxi study, in which all women received 4-quadrant colposcopic biopsies

Close modal
Figure 2.

Simplified overview of the follow-up protocol for the Portland study. HPV indicates human papillomavirus

Figure 2.

Simplified overview of the follow-up protocol for the Portland study. HPV indicates human papillomavirus

Close modal
Figure 3.

Selected data from the Hybrid Capture 2 cross-sectional human papillomavirus (HPV) studies. Disease reference standards were cervical intraepithelial neoplasia (CIN) 2/3 (London, Reims, Newfoundland, Cape Town [individual tests only; the combined Papanicolaou and HPV DNA tests in this series refer to CIN 3], and Shanxi [individual tests only; the combined Pap and HPV DNA tests in this series refer to CIN 3]) and CIN 3 (Seattle, Guanacaste, Cape Town [both Pap and HPV DNA tests; the individual tests in this series refer to CIN 2/3], Shanxi [both Pap and HPV DNA tests; the individual tests in this series refer to CIN 2/3], Hannover, and Morelos). Data are for women aged 30 years and older, except for the study from Newfoundland. A, Sensitivity of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Papanicolaou and HPV DNA test data were unavailable for the London and Seattle studies. B, Specificity of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Papanicolaou and HPV DNA test data were unavailable for the London, Seattle, and Reims studies. C, Positive predictive value (PPV) of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Papanicolaou and HPV DNA test data were unavailable for the London, Seattle, and Reims studies. D, Negative predictive value (NPV) of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Pap and HPV DNA test data were unavailable for the London and Seattle studies

Figure 3.

Selected data from the Hybrid Capture 2 cross-sectional human papillomavirus (HPV) studies. Disease reference standards were cervical intraepithelial neoplasia (CIN) 2/3 (London, Reims, Newfoundland, Cape Town [individual tests only; the combined Papanicolaou and HPV DNA tests in this series refer to CIN 3], and Shanxi [individual tests only; the combined Pap and HPV DNA tests in this series refer to CIN 3]) and CIN 3 (Seattle, Guanacaste, Cape Town [both Pap and HPV DNA tests; the individual tests in this series refer to CIN 2/3], Shanxi [both Pap and HPV DNA tests; the individual tests in this series refer to CIN 2/3], Hannover, and Morelos). Data are for women aged 30 years and older, except for the study from Newfoundland. A, Sensitivity of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Papanicolaou and HPV DNA test data were unavailable for the London and Seattle studies. B, Specificity of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Papanicolaou and HPV DNA test data were unavailable for the London, Seattle, and Reims studies. C, Positive predictive value (PPV) of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Papanicolaou and HPV DNA test data were unavailable for the London, Seattle, and Reims studies. D, Negative predictive value (NPV) of Papanicolaou, HPV DNA, and Papanicolaou and HPV DNA tests for CIN 2/3 or CIN 3. Combined Pap and HPV DNA test data were unavailable for the London and Seattle studies

Close modal
Figure 4.

Baseline test detection of 171 cases of cervical intraepithelial neoplasia (CIN) 3 diagnosed during a 10-year span as a consequence of annual cytologic screening in the Portland study. Pap indicates Papanicolaou; HPV, human papillomavirus

Figure 4.

Baseline test detection of 171 cases of cervical intraepithelial neoplasia (CIN) 3 diagnosed during a 10-year span as a consequence of annual cytologic screening in the Portland study. Pap indicates Papanicolaou; HPV, human papillomavirus

Close modal

The studies spanned a broad range of geographic, ethnic, and socioeconomic groupings, representing many of the major populations worldwide. The studies varied widely in population size, from 1365 women in Cape Town, South Africa,27 to 20 810 women in Portland.24 Overall, the studies included more than 77 000 women and more than 1000 cases of CIN 2/3, spanning 4 continents and 11 countries. Our preferred disease reference standard was CIN 2/3, because we felt that it allowed assessment of test performance on the basis of the ability to detect all reasonable suspicion of potentially malignant disease. Arguments can be made for a more specific end point, such as CIN 3; indeed, for some studies only CIN 3 data were available. For studies in which both CIN 2/3 and CIN 3 data and age categorization were provided, we present data on the detection of CIN 3 in women older than age 30 years as a group of particular interest. We noticed little difference in the performance of the tests whether CIN 2/3 or CIN 3 was the reference, except in the Seattle study10 (Table 2), suggesting that generally there was low misclassification between CIN 1 and CIN 2.

The prevalence of CIN 2/3 varied widely (as would be expected given the study settings) and spanned the gamut from unscreened high-risk populations, such as Shanxi Province, China21 (CIN 2/3 prevalence of 4.3%), to much lower risk populations, such as the United Kingdom25 (CIN 2/3 prevalence of 1.2%). As would be expected, the prevalence of CIN 3 was lower but exhibited a similar trend as the CIN 2/3 data with respect to the different populations.

It is evident from Table 2 and Figure 3, A, that HPV DNA testing by HC2 had a higher sensitivity (in some cases much higher) than cytology. For example, in the study from Reims28 HC2 HPV DNA testing detected 100% of CIN 2/3, as compared to 58% for the conventional Papanicolaou test (a 72% increase for HPV DNA) and 84% for the ThinPrep test (a 19% increase for HPV DNA). Similar or greater differences between HPV DNA testing and cytology were seen in the studies from Newfoundland, Canada26; Seattle, Wash10; Morelos, Mexico30; and Hannover-Tubingen, Germany31; however, in the other studies the improvement in HPV DNA sensitivity relative to the Papanicolaou test was somewhat less. There was not a single study in which the sensitivity of the Papanicolaou test equaled or exceeded the sensitivity of the HPV DNA test. The same observations can be made for the NPVs (Figure 3, D); the HPV DNA NPVs exceeded the Papanicolaou test NPVs in all studies. In most of the studies (all with >2000 women), these differences in sensitivity and NPV were significant (data not shown).

The specificity values for HC2 HPV DNA testing were generally lower than the specificity values of the Papanicolaou test, except in the Shanxi study,21 in which the specificity of HPV DNA was higher than the specificity of the ThinPrep test (Figure 3, B). The PPVs of the Papanicolaou test were overall a little higher than the PPVs for HPV DNA testing (Figure 3, C). The PPV of the Papanicolaou test was higher in the studies from London,25 Reims,28 Cape Town,27 and Hannover,31 and was unusually high in Morelos,30 whereas the HPV PPV was higher in the studies from Seattle,10 Newfoundland,26 Guanacaste,22 and Shanxi.21 

For most studies, the sensitivities of the Papanicolaou test and the HC2 HPV DNA test combined were higher than that of either test alone, although in a few studies the Papanicolaou test added no sensitivity to the HPV DNA test (Figure 3, A). It can be concluded that the tests can complement each other effectively. In contrast, the specificities (Figure 3, B) of the combined tests were less than the specificities for either test alone, but in most of these combinations, the specificity decreases were small, on the order of a few percent. The PPVs of the individual or the combined tests were quite similar in some studies and decreased for the combination in others (Figure 3, C), demonstrating that the HPV DNA test detects additional CIN 2/3 at a comparable or slightly lower rate per positive test result as compared with the Papanicolaou test. As expected, the NPVs of HPV DNA combined with the Papanicolaou test were higher and in a few studies were 100% (Figure 3, D). The NPV data are shown on a scale from 98% to 100%, because in screening settings in which high-grade neoplastic disease is generally relatively rare the NPVs are always high, but very small absolute differences in NPV can have a major impact on safety and cost-effectiveness considerations.

Table 3 shows the data for the 3 PCR studies. The conclusions from these studies are broadly the same as for the HC2 studies, that is, the PCR HPV DNA test was substantially more sensitive than the Papanicolaou test. Of note, the sensitivity of the Papanicolaou test in Jena, Germany,29 was only 20%, as compared to the sensitivity of the HPV DNA test, which was 89%.29 However, in this study the Papanicolaou test had a specificity of 99%, compared to 94% for the HPV DNA PCR test. In the study from Seattle,10 the PCR test had a sensitivity of 88% for CIN 3, as compared to 61% for the ThinPrep Papanicolaou test. When restricted to women older than 30 years, the PCR test detected 80% of CIN 3 as compared to 50% for the ThinPrep Papanicolaou test. In contrast to the study from Jena,29 the specificity values of the HPV DNA PCR test and the ThinPrep Papanicolaou test in the Seattle study10 were similar.

The third PCR population screening study, which was conducted in Denmark,20 was longitudinal and presented the sensitivity of the PCR test for CIN 2/3 as 93%. A key contribution of this study was the estimated risk of CIN 2/3 in women who were cytologically normal but HPV DNA positive at baseline and who were then monitored for 2 years. The odds ratios for CIN 2/3 were 692 for persistence of any high-risk HPV type and 813 for persistence of the same high-risk HPV type compared to the reference group of women who were negative for HPV DNA at both time points.

Taken as a whole, the studies indicate that the sensitivities of HPV DNA testing for CIN 2/3 or CIN 3 were comparable for HC2 and PCR and ranged from 63% to 100% with a median of approximately 92%. In comparison, the sensitivity of the conventional Papanicolaou test ranged from 20% to 86% with a median value of approximately 58%, whereas the sensitivity of ThinPrep liquid cytology fell between 38% and 94% with insufficient data to approximate a median value. It is clear that the sensitivity of HPV DNA testing combined with the Papanicolaou test was much higher (76%–100% with a median value of approximately 95%) than the sensitivity of the Papanicolaou test alone, regardless of whether it was conventional or liquid-based.

The Portland study was a 10-year longitudinal study of HPV natural history that provided data on the risk of future CIN 3 in cytologically normal, HPV-infected women. The study protocol and data have been described extensively in several publications.12,24,32–34  Figure 2 is a simplified representation of key aspects of the study. In brief, between April 1, 1989, and November 2, 1990, 23 702 women were enrolled at Kaiser Permanente clinics in Portland. Participants provided a baseline conventional Papanicolaou test and a cervicovaginal lavage specimen for HPV DNA testing. A small percentage of women were excluded for reasons including refusal to participate, prior hysterectomy, inadequate specimen for Papanicolaou or HPV DNA testing, or premature colposcopy. The remaining 20 810 women were divided into several groups for further analysis. Human papillomavirus DNA testing for oncogenic-risk HPV types on archived baseline specimens was conducted in spring 2001 by HC2.

Several questions related to test performance were studied, including (a) the ability of baseline HPV DNA and Papanicolaou tests to identify women diagnosed with CIN 3 during the 10-year duration of the study, (b) the relative risk for future CIN 3 in the subset of women who were cytologically normal at baseline as a function of HPV DNA positivity, and (c) the relative risk of CIN 3 in women with negative results on HPV DNA and Papanicolaou tests at baseline as compared to women who had 3 normal follow-up Papanicolaou tests regardless of HPV status. Human papillomavirus DNA test results were not used for any aspect of the clinical management or follow-up of the women.

The cohort was monitored for up to 122 months by standard cytologic screening, which at the time involved annual conventional Papanicolaou tests for most women. There were no important differences in follow-up characteristics of HPV DNA–positive versus HPV DNA–negative women, indicating minimal if any bias in assessing the effects of HPV DNA testing.24 One hundred seventy-one cases of CIN 3 were detected cumulatively during the follow-up period.

Tables 4 and 5 show the relative risks of CIN 3 according to Papanicolaou test, HPV DNA test, or combined test status. Figure 4 shows the secular trend for CIN 3 diagnosis during follow-up according to initial positivity by the Papanicolaou test at a cutoff of ASCUS or higher, by HPV DNA alone, or by a combination of the 2 tests. Conceptually, the groups can be viewed as risk stratifications based on the initial Papanicolaou test, initial HPV DNA, or either test positive as a combination, with the percentage of the CIN 3 cases that emanated from each group plotted versus time. Thus, 33% of the cumulative CIN 3 cases during the 10 years had a positive Papanicolaou test at baseline. In comparison, 64% of CIN 3 patients were HPV DNA positive at baseline, and 69% of CIN 3 patients were positive for Papanicolaou, HPV DNA, or both tests at baseline. Eighty-six percent (95% confidence interval = 80.3%–92.6%) of CIN 3 patients diagnosed during the first 45 months were positive by Papanicolaou, HPV DNA, or both tests at baseline.24 

Table 4.

Risk of Future Cervical Intraepithelial Neoplasia (CIN) 3 in Women During Follow-up in the Portland Study According to Baseline Test Result Status*

Risk of Future Cervical Intraepithelial Neoplasia (CIN) 3 in Women During Follow-up in the Portland Study According to Baseline Test Result Status*
Risk of Future Cervical Intraepithelial Neoplasia (CIN) 3 in Women During Follow-up in the Portland Study According to Baseline Test Result Status*
Table 5.

Risk of Future Cervical Intraepithelial Neoplasia (CIN) 3 in Women in the Portland Study According to Baseline or Follow-up Test Status*

Risk of Future Cervical Intraepithelial Neoplasia (CIN) 3 in Women in the Portland Study According to Baseline or Follow-up Test Status*
Risk of Future Cervical Intraepithelial Neoplasia (CIN) 3 in Women in the Portland Study According to Baseline or Follow-up Test Status*

Although not part of our larger data set, a recent longitudinal study by Bory et al35 is consistent with the observations of Kjaer et al,20 Sherman et al,24 and our own data in the Portland study. Bory et al35 observed a cohort of 3091 cytologically normal women for more than 4 years. These women were a subset of the population of women described by Clavel et al28 (referred to herein as the Reims study). The Reims investigators found that 7.7% (51/659) of the women initially positive for HPV DNA at baseline and 21.2% (51/241) of women who were persistently positive for oncogenic HPV DNA types by HC2 were diagnosed with CIN 2/3 on histology within 36 months, compared to only 0.08% (2/2432) of women who were initially HPV negative. Of note, both these latter cases of CIN 2/3 were found to be HPV DNA positive at diagnosis. The Reims investigators estimated the relative risk of CIN 2/3 given persistent HPV positivity (second HPV DNA test positive within 4–10 months of the baseline positive) as 237.

The results of these large HPV DNA screening studies provide compelling evidence for the clinical utility of HPV DNA testing as an adjunct to cervical cytology for routine screening in women older than 30 years and perhaps in certain settings also at younger ages. The wide diversity of study locations and risk groupings is a particular strength of the combined results and indicates that the data may be generalized to many screening settings worldwide. The cross-sectional and longitudinal data complement each other and show that women infected with oncogenic HPV constitute a higher risk group requiring more vigilant follow-up, as compared to women with no evidence of HPV infection who can be regarded as being at low risk for cervical cancer.

There are weaknesses in this data set and in certain of the studies. One criticism relates to the lack of true assessment of all potential CIN 2/3 on the cervix, given that the tests and even the combinations of the tests may still miss a large proportion of true high-grade neoplasia on the cervix (the verification bias effect). The argument states that such biases will make the HPV test appear better than it really is. This argument is diminished by the study from Shanxi Province in China, in which there was a rigorous assessment of all women by colposcopy and biopsy that indicated that the HPV DNA and Papanicolaou tests combined detected essentially all CIN 2/3 and cancers. In several other studies, the authors noted that there was virtually no CIN 2/3 detected in colposcopy control groups of double-negative women.10,22,35 We feel that there is little justification to be overly concerned about verification bias in these studies. Even if such a bias were present at high levels in some or most studies, it would not change the relative performance relationships of the tests. In other words, HPV DNA would still be the more sensitive test, with or without verification bias. The real impact of the bias is on the accuracy of our assessment of the absolute values of sensitivity and specificity.

It has been argued that additional CIN 3 detected by HPV DNA testing may be a different and less important category of disease compared to the CIN 3 detected by the Papanicolaou test and that perhaps this extra HPV-related disease may not be progressive. This argument is not supported by data from longitudinal studies that relied on follow-up Papanicolaou tests to detect disease among cohorts of women who were either HPV DNA positive or HPV DNA negative at baseline. As can be seen in Tables 3 through 5, and in Figure 4, there was a much larger risk of CIN 3 in the HPV DNA–positive women, as compared to the HPV DNA–negative women. However, for all cases of disease detected, it was the Papanicolaou test that detected this disease, indicating that there is no discernible difference among hypothetical alternative classes of CIN 3. Virtually all CIN 3 is HPV DNA positive, and there is no experimental or epidemiological basis for suggesting only the subset of CIN 3 detected by initial or baseline Papanicolaou tests is potentially progressive. Even if this unlikely argument is given the benefit of the doubt, one has to ask which Papanicolaou test has these special attributes, the version that can detect 50%, 85%, or some other percentage of CIN 3?

Human papillomavirus DNA testing has been commercially available for decades; however, the sensitivity and specificity of earlier test versions were inadequate to be clinically useful. Although some methods were more sensitive than others, they were technically daunting and available only in specialized research laboratories.

Because older, relatively insensitive HPV DNA tests were commercially marketed and their evaluation for clinical utility was consistently found to be wanting, both pathologists and clinicians concluded that the promise of HPV DNA testing far exceeded its ability to deliver useful results, and successive entries were greeted with skepticism. It was only after decades of development that user-friendly HPV DNA tests became available for general clinical use and proved to be up to the job. It was the development of the HC2 test and the use of the Manos and Walboomers primers in PCR applications that finally resulted in relevant data on the clinical usefulness of HPV DNA testing being widely recognized and accepted. The first important clinical applications for HPV DNA testing were for the triage and management of women with abnormal Papanicolaou tests (particularly in the management of patients with the new category of ASC-US or “borderline” Papanicolaou tests), for triage of postmenopausal patients with low-grade cytological abnormalities, and as a test of cure in patients treated for CIN.36–38 The recognition that HPV DNA testing should play an important role in the management of patients with abnormal Papanicolaou tests was memorialized in the American Society for Colposcopy and Cervical Pathology publication, “2001 Consensus Guidelines for the Management of Women With Cervical Cytological Abnormalities.” 15 

Screening for cervical cancer and its precursors and treatment of women with detected cervical disease in population-based programs constitute one of the most effective cancer prevention tools ever devised, in some countries decreasing the incidence of, and deaths from, cervical cancer by 80% or more. Such programs have almost universally used the 50-year-old conventional Papanicolaou test for screening. In recent studies, however, it has been reported that conventional Papanicolaou tests are only about 50% to 60% sensitive in detecting high-grade CIN and cervical cancer and are even less sensitive for low-grade lesions.7–9 

It is because of this relatively low sensitivity that the Papanicolaou test had to be repeated on a regular basis to be effective. This strategy was successful only because the transit time from HPV infection to invasive cancer generally exceeds 10 years or more, enabling most patients at risk to have their disease detected by repeated testing, even with a relatively insensitive tool.39,40 However, despite the favorable relationship between the natural history of cervical neoplasia and Papanicolaou testing programs, it is almost universally agreed that even the best sampling procedures are likely to result in missed cases, that even the best laboratories fail occasionally to detect abnormal cells on the slides, and that Papanicolaou test interpretation is part art and part science and not very reproducible. It was widely reported that in retrospective reviews of Papanicolaou tests reported as negative in patients with high-grade CIN or cancer, 30% to 80% of those “negative” Papanicolaou tests contained abnormal cells missed by the screeners.40,41 Furthermore, almost all such retrospectively reviewed samples, whether containing abnormal cells or not, had detectable HPV DNA when the cells were stripped from the slide and tested.39,40,42–44 

In an effort to improve cervical sampling and increase the sensitivity of Papanicolaou tests, new collection and preparation procedures were devised, leading to the development of the fluid-based Papanicolaou technique.45 With fluid-based testing the sensitivity for the detection of CIN 2/3 climbed to about 85%. In the United States, more than 60% of Papanicolaou tests are now fluid based. Even with this improvement, however, at least 15% of clinically important lesions will be undetected on a single screen.

Papanicolaou test litigation is a problem for both pathologists and clinicians in the United States and is becoming a problem in some other countries as well. Few patients understand that the Papanicolaou test is a screening test—not a diagnostic test—and few understand that there is an unavoidable false-negative fraction, no matter how careful the clinician is in taking the sample nor how skilled and compulsively careful the laboratory is.

There is a pressing need for a test that can discriminate between patients who have cervical neoplasia, patients who do not, and patients who are at proximate risk for developing disease. It is clear from the present review that HPV DNA testing using HC2 or PCR can identify almost all patients with CIN 3 or higher. Adding a fluid-based cytology test to the HPV DNA test increases sensitivity by approximately 5%. As is apparent from Figure 3, A, the sensitivity value for CIN 3 or higher using a combination of HPV DNA testing and cytology was greater than 90% in 6 of the 7 studies for which the data were available and was 100% in 3 of the 7 studies. More importantly, the NPVs for the combinations were above 99% for all 7 studies and were 100% in 4 of the 7. If a patient is negative for HPV DNA and has a negative Papanicolaou test, the clinician can state with reasonable certainty that “negative means negative.” Although the specificity and PPVs were slightly reduced in the combined test groups, this trade-off against detecting disease seems acceptable.

Women who are HPV DNA positive but who do not have an abnormal Papanicolaou test or clinical evidence of HPV-related disease should not be viewed as having “false-positive” tests. These are the women at greatest risk of developing an abnormal Papanicolaou test and cervical neoplasia prospectively, as shown by the Portland, Copenhagen, and Reims investigators. Such cases can be managed by close follow-up and repeat testing.

It seems likely, if combined HPV DNA and Papanicolaou testing is widely adopted, that the results would be salutary. False negatives would be expected to be dramatically reduced, double-negative patients could safely be screened at longer intervals (offsetting increased testing costs),46–48 and patients identified as being high risk but not having identifiable disease could be monitored closely. These outcomes would benefit patients, doctors, and the health care system. Consistent with these ideas, the American Cancer Society has recently released new guidelines for the early detection of cervical neoplasia and cancer in the United States that incorporate adjunctive HPV DNA testing in women older than age 30 years at 3 yearly or longer intervals.49 

We are grateful for the assistance provided by the many investigators who planned and executed these large human papillomavirus DNA screening studies. Special thanks go to the principal investigators and coinvestigators who provided raw data from their studies for our further analyses, including Mark Schiffman, MD, MPH, Philip E. Castle, PhD, MPH, and Mark E. Sherman, MD, of the National Cancer Institute team, who were instrumental in the Portland and the Guanacaste studies; Thomas C. Wright, Jr, MD, and Lynette Denny, MMED, of the Cape Town study; Karl-Ulrich Petry, MD, and Thomas Iftner, PhD, of the Hannover-Tubingen study; Jerome Belinson, MD, and Robert G. Pretorius, MD, of the Shanxi study; and Jorge Salmeron, MD, DSc, Mauricio Hernandez-Avila, DSc, Eduardo Lazcano-Ponce, PhD, and Keerti V. Shah, MD, DrPH, of the Morelos study. David Schoenfeld, PhD, provided invaluable expertise and help with some statistical analyses of the Portland study. Mark Del Vecchio, BS, Marshall McCarty, BS, and Michael J. Benecky, PhD, of Digene Corporation, Gaithersburg, Md, assisted by auditing and expertly organizing much of the raw data to permit the additional analyses. Iwona Mielzynska-Lohnas, PhD, and Robert Hallenberg, MS, assisted in the Hybrid Capture 2 testing of the Portland study. Katherine P. Mack, MAT, was of great help in the preparation of the manuscript.

Bosch
,
F. X.
,
A.
Lorincz
,
N.
Munoz
,
C. J. L. M.
Meijer
, and
K. V.
Shah
.
The causal relation between human papillomavirus and cervical cancer.
J Clin Pathol
2002
.
55
:
244
265
.
Lorincz
,
A. T.
,
R.
Reid
,
A. B.
Jenson
,
M. D.
Greenberg
,
W.
Lancaster
, and
R. J.
Kurman
.
Human papillomavirus infection of the cervix: relative risk associations of 15 common anogenital types.
Obstet Gynecol
1992
.
79
:
328
337
.
Bosch
,
F. X.
,
M. M.
Manos
, and
N.
Munoz
.
et al
.
Prevalence of human papillomavirus in cervical cancer: a worldwide perspective.
J Natl Cancer Inst
1995
.
87
:
796
802
.
Walboomers
,
J. M. M.
,
M. V.
Jacobs
, and
M. M.
Manos
.
et al
.
Human papillomavirus is a necessary cause of invasive cervical cancer worldwide.
J Pathol
1999
.
189
:
12
19
.
Herrero
,
R.
,
A.
Hildesheim
, and
C.
Bratti
.
et al
.
Population-based study of human papillomavirus infection and cervical neoplasia in rural Costa Rica.
J Natl Cancer Inst
2000
.
92
:
464
474
.
Thomas
,
D. B.
,
Q.
Qin
, and
J.
Kuypers
.
et al
.
Human papillomaviruses and cervical cancer in Bangkok, II: risk factors for in situ and invasive squamous cell cervical carcinomas.
Am J Epidemiol
2001
.
153
:
732
739
.
Nanda
,
K.
,
D. C.
McCrory
, and
E. R.
Myers
.
et al
.
Accuracy of the Papanicolaou test in screening for and follow-up of cervical cytologic abnormalities: a systematic review.
Ann Intern Med
2000
.
132
:
810
819
.
Bastian
,
L.
,
S.
Datta
,
V.
Hasselblad
,
J.
Hickey
,
E.
Myers
, and
K.
Nanda
.
Evaluation of cervical cytology, 5. 1999. Rockville, Md: Agency for Health Care Policy and Research (AHCPR).
Evidence report/technology assessment prepared by Duke University under contract 290-97-0014. AHCPR publication 99-E010. Available at: http://hstat.nlm.nih.gov/hq/Hquest/db/local.epc.er.cyt/screen/DocTitle/s/48139. Accessed February 25, 2003
.
Fahey
,
M. T.
,
L.
Irwig
, and
P.
Macaskill
.
Meta-analysis of Pap test accuracy.
Am J Epidemiol
1995
.
141
:
680
689
.
Kulasingam
,
S. L.
,
J. P.
Hughes
, and
N. B.
Kiviat
.
et al
.
Evaluation of human papillomavirus testing in primary screening for cervical abnormalities: comparison of sensitivity, specificity, and frequency of referral.
JAMA
2002
.
288
:
1749
1757
.
for the Atypical Squamous Cells of Undetermined Significance/Low-Grade Squamous Intraepithelial Lesion Triage Study (ALTS) Group.
Comparison of 3 management strategies for patients with atypical squamous cells of undetermined significance (ASCUS): baseline results from a randomized trial.
J Natl Cancer Inst
2001
.
93
:
293
299
.
Schiffman
,
M. H.
,
H. M.
Bauer
, and
R. N.
Hoover
.
et al
.
Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasia.
J Natl Cancer Inst
1993
.
85
:
958
964
.
Franco
,
E. L.
,
L. L.
Villa
, and
J. P.
Sobrinho
.
et al
.
Epidemiology of acquisition and clearance of cervical human papillomavirus infection in women from a high-risk area for cervical cancer.
J Infect Dis
1999
.
180
:
1415
1423
.
Sherman
,
M. E.
,
M.
Schiffman
, and
J. T.
Cox
.
Effects of age and human papilloma viral load on colposcopy triage: data from the randomized Atypical Squamous Cells of Undetermined Significance/Low-Grade Squamous Intraepithelial Lesion Triage Study (ALTS).
J Natl Cancer Inst
2002
.
94
:
102
107
.
for the 2001 ASCCP-Sponsored Consensus Conference.
2001 consensus guidelines for the management of women with cervical cytological abnormalities.
JAMA
2002
.
287
:
2120
2129
.
Lorincz
,
A.
and
J.
Anthony
.
Advances in HPV detection by Hybrid Capture®.
Pap Report
2001
.
12
:
145
154
.
Jacobs
,
M. V.
,
P. J. F.
Snijders
,
A. J. C.
van den Brule
,
T. J. M.
Helmerhorst
,
C. J. L. M.
Meijer
, and
J. M. M.
Walboomers
.
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
.
Manos
,
M. M.
,
Y.
Ting
,
D. K.
Wright
,
A. J.
Lewis
,
T. R.
Broker
, and
S. M.
Wolinsky
.
Use of polymerase chain reaction amplification for the detection of genital human papillomaviruses.
In: Furth M, Greaves M, eds. Molecular Diagnostics of Human Cancer. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory; 1989:209–214
.
Castle
,
P. E.
,
M.
Schiffman
, and
P. E.
Gravitt
.
et al
.
Comparisons of HPV DNA detection by MY09/11 PCR methods.
J Med Virol
2002
.
68
:
417
423
.
Kjaer
,
S. K.
,
A. J. C.
van den Brule
, and
G.
Paull
.
et al
.
Type specific persistence of high risk human papillomavirus (HPV) as indicator of high grade cervical squamous intraepithelial lesions in young women: population based prospective follow up study.
BMJ
2002
.
325
:
572
578
.
Belinson
,
J.
,
Y. L.
Qiao
, and
R.
Pretorius
.
et al
.
Shanxi Province cervical cancer screening study: a cross-sectional comparative trial of multiple techniques to detect cervical neoplasia.
Gynecol Oncol
2001
.
83
:
439
444
.
Schiffman
,
M.
,
R.
Herrero
, and
A.
Hildesheim
.
et al
.
HPV DNA testing in cervical cancer screening: results from women in a high-risk province of Costa Rica.
JAMA
2000
.
283
:
87
93
.
Hall
,
S.
,
A.
Lorincz
, and
F.
Shah
.
et al
.
Human papillomavirus DNA detection in cervical specimens by Hybrid Capture: correlation with cytologic and histologic diagnoses of squamous intraepithelial lesions of the cervix.
Gynecol Oncol
1996
.
62
:
353
359
.
Sherman
,
M. E.
,
A. T.
Lorincz
, and
D. R.
Scott
.
et al
.
Baseline cytology, human papillomavirus testing, and risk for cervical neoplasia: a 10-year cohort analysis.
J Natl Cancer Inst
2003
.
95
:
46
52
.
Cuzick
,
J.
,
E.
Beverley
, and
L.
Ho
.
et al
.
HPV testing in primary screening of older women.
Br J Cancer
1999
.
81
:
554
558
.
Ratnam
,
S.
,
E. L.
Franco
, and
A.
Ferenczy
.
Human papillomavirus testing for primary screening of cervical cancer precursors.
Cancer Epidemiol Biomarkers Prev
2000
.
9
:
945
951
.
Wright
,
T. C. Jr
,
L.
Denny
,
L.
Kuhn
,
A.
Pollack
, and
A.
Lorincz
.
HPV DNA testing of self-collected vaginal samples compared with cytologic screening to detect cervical cancer.
JAMA
2000
.
283
:
81
86
.
Clavel
,
C.
,
M.
Masure
, and
J-P.
Bory
.
et al
.
Human papillomavirus testing in primary screening for the detection of high-grade cervical lesions: a study of 7932 women.
Br J Cancer
2001
.
89
:
1616
1623
.
Schneider
,
A.
,
H.
Hoyer
, and
B.
Lotz
.
et al
.
Screening for high-grade cervical intra-epithelial neoplasia and cancer by testing for high-risk HPV, routine cytology or colposcopy.
Int J Cancer
2000
.
89
:
529
534
.
Salmeron
,
J.
,
E. C.
Lazcano
, and
A.
Lorincz
.
et al
.
Comparison of HPV-based assays with Papanicolaou smears for cervical cancer screening in Morelos State, Mexico.
Cancer Causes Control. In press
.
Petry
,
K-U.
,
S.
Menton
, and
M.
Menton
.
et al
.
Inclusion of HPV testing in routine cervical cancer screening for women above 29 years in Germany: results for 8468 patients.
Br J Cancer
2003
.
88
:
1570
1577
.
Liaw
,
K-L.
,
A. G.
Glass
, and
M. M.
Manos
.
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
.
Castle
,
P. E.
,
S.
Wacholder
, and
A. T.
Lorincz
.
et al
.
A prospective study of high-grade cervical neoplasia risk among human papillomavirus infected women.
J Natl Cancer Inst
2002
.
94
:
1406
1414
.
Lorincz
,
A. T.
,
P. E.
Castle
, and
M. E.
Sherman
.
et al
.
Viral load of human papillomavirus and risk of CIN3 or cervical cancer.
Lancet
2002
.
360
:
228
229
.
Bory
,
J-P.
,
J.
Cucherousset
, and
M.
Lorenzato
.
et al
.
Recurrent human papillomavirus infection detected with the Hybrid Capture II assay selects women with normal cervical smears at risk for developing high grade cervical lesions: a longitudinal study of 3,091 women.
Int J Cancer
2002
.
102
:
519
525
.
Jain
,
S.
,
C-J.
Tseng
,
S-G.
Horng
,
Y-K.
Soong
, and
C. C.
Pao
.
Negative predictive value of human papillomavirus test following conization of the cervix uteri.
Gynecol Oncol
2001
.
82
:
177
180
.
Lin
,
C-T.
,
C-J.
Tseng
, and
C-H.
Lai
.
et al
.
Value of human papillomavirus deoxyribonucleic acid testing after conization in the prediction of residual disease in the subsequent hysterectomy specimen.
Am J Obstet Gynecol
2001
.
184
:
940
945
.
Paraskevaidis
,
E.
,
G.
Koliopoulos
,
Y.
Alamanos
,
V.
Malamou-Mitsi
,
E. D.
Lolis
, and
H. C.
Kitchener
.
Human papillomavirus testing and the outcome of treatment for cervical intraepithelial neoplasia.
Obstet Gynecol
2001
.
98
:
833
836
.
Jacobs
,
M. V.
,
D.
Zielinski
, and
C. J. L. M.
Meijer
.
et al
.
A simplified and reliable HPV testing of archival Papanicolaou-stained cervical smears: application to cervical smears from cancer patients starting with cytologically normal smears.
Br J Cancer
2000
.
82
:
1421
1426
.
Zielinski
,
G. D.
,
P. J. F.
Snijders
, and
L.
Rozendaal
.
et al
.
HPV presence precedes abnormal cytology in women developing cervical cancer and signals false negative smears.
Br J Cancer
2001
.
85
:
398
404
.
Allen
,
K. A.
,
S.
Zaleski
, and
M. B.
Cohen
.
Review of negative Papanicolaou tests: is the retrospective 5-year review necessary?
Am J Clin Pathol
1994
.
101
:
19
21
.
Wallin
,
K-L.
,
F.
Wiklund
, and
T.
Angstrom
.
et al
.
Type-specific persistence of human papillomavirus DNA before the development of invasive cervical cancer.
N Engl J Med
1999
.
341
:
1633
1638
.
Nobbenhuis
,
M. A. E.
,
J. M. M.
Walboomers
, and
T. J. M.
Helmerhorst
.
et al
.
Relation of human papillomavirus status to cervical lesions and consequences for cervical-cancer screening: a prospective study.
Lancet
1999
.
354
:
20
25
.
Rozendaal
,
L.
,
J. M. M.
Walboomers
, and
J. C.
van der Linden
.
et al
.
PCR-based high-risk HPV test in cervical cancer screening gives objective risk assessment of women with cytomorphologically normal cervical smears.
Int J Cancer
1996
.
68
:
766
769
.
Hutchinson
,
M. L.
and
C. M.
Cassin
.
Ball HG III. The efficacy of an automated preparation device for cervical cytology.
Am J Clin Pathol
1991
.
96
:
300
305
.
Mandelblatt
,
J. S.
,
W. F.
Lawrence
, and
S. M.
Womack
.
et al
.
Benefits and costs of using HPV testing to screen for cervical cancer.
JAMA
2002
.
287
:
2372
2381
.
Mandelblatt
,
J. S.
,
W. F.
Lawrence
, and
L.
Gaffikin
.
et al
.
Costs and benefits of different strategies to screen for cervical cancer in less developed countries.
J Natl Cancer Inst
2002
.
94
:
1469
1483
.
Goldie
,
S. J.
,
L.
Kuhn
,
L.
Denny
,
A.
Pollack
, and
T. C.
Wright
.
Policy analysis of cervical cancer screening strategies in low-resource settings: clinical benefits and cost-effectiveness.
JAMA
2001
.
285
:
3107
3115
.
Saslow
,
D.
,
C. D.
Runowicz
, and
D.
Solomon
.
et al
.
American Cancer Society guideline for the early detection of cervical neoplasia and cancer.
CA Cancer J Clin
2002
.
52
:
342
362
.

Presented at the College of American Pathologists Strategic Science Series Conference, HPV Testing: Are You Ready for a New Era in Cervical Cancer Screening?, Rosemont, Ill, September 21–22, 2002.

Author notes

Reprints: Attila T. Lörincz, PhD, Digene Corporation, 1201 Clopper Rd, Gaithersburg, MD 20878 ([email protected])