In the United States, cotesting with human papillomavirus (HPV) and cytology has been accepted as an alternative method of screening to routine (annual or biennial) cervical cytology alone for women 30 years and older, with a recommendation that women who test negative for both not be rescreened before 3 years.1,2 Even longer intervals after double-negative screening have been recommended based on a recent pooled analysis of cohort data.3
The clinical management of women who test HPV positive with concurrent cytologic abnormalities will fall into established clinical guidelines, but no one yet knows how to manage optimally the many women who test HPV positive and cytology negative. These women are at increased risk of underlying and subsequent cervical intraepithelial neoplasia grade 2 or higher (CIN 2+) and CIN 3+.4 However, they cannot be managed by immediate colposcopy because many of those destined to develop CIN 3+ do not yet have visible colposcopic lesions. Moreover, management by immediate colposcopy results in the excessive detection of CIN 2 that is destined to regress4 but will be treated by current clinical protocols that treat any lesion diagnosed as CIN 2 or more severe for safety.
Until effective triage tests for HPV-positive, Pap-negative women are found, longitudinal follow-up by repeated screening will be needed. This makes sense because it is overtly detectable (persistent) carcinogenic HPV infections that cause CIN 3+, not once-detected (transient) HPV infections.
Previous studies have shown that a sensitive first round of HPV testing significantly reduces the predictive value of the HPV test in the next round of testing.5–7 But previous studies have been limited by small numbers of outcomes in a research rather than in a clinical setting.
To bridge this gap in knowledge, we conducted a study to explore the effect of screening history on the risk of cervical precancer and cancer after an HPV-positive test of HPV-positive women treated at Kaiser Permanente in Northern California, which adopted cotesting for cervical cancer screening in 2003.
MATERIALS AND METHODS
Kaiser Permanente is an integrated delivery system that provides comprehensive care, including inpatient, outpatient, and pharmacy services to more than 3 million members. With the permission of Kaiser Foundation Research Institute and National Cancer Institute Institutional Review Boards, we selected all women aged 30 and older who tested positive for carcinogenic HPV by Hybrid Capture 2 from January 2006 through December 2008 (n=32,057). We selected this time window for the “index” visit because by this time recommendations for management of HPV-positive, Pap-negative women were widely implemented within the Kaiser Permanente Northern California system. As suggested by interim professional guidelines,2 these cases were managed by 1-year follow-up and referral to colposcopy if screen positive (HPV or cytology positive) at the follow-up visit. In addition, many women had previous HPV testing results, which permitted us to examine the effect of the past results on the risk after the index HPV-positive test in the “enrollment” time period. The database repository used to perform the ETL (Extract Transform and Load) task and store the data are a Microsoft SQL Server 2005. Data are sourced from the following: 1) Pap data were extracted from the Laboratory Information System-CoPath Plus, and 2) biopsy and HPV data were extracted from Laboratory Reporting database. Password-protected databases containing patient information, screening results, and pathologic diagnoses were linked through a unique patient identification number.
We further restricted our analysis to women (n=26,914) who had at least one follow-up visit before January 2010, the last update before this analysis of the Kaiser Permanente Northern California databases. Finally, we excluded women who had a CIN 2+ diagnosis before the index visit and who were likely to have undergone excision treatments (Table 1), resulting in a final analytic cohort of 26,799 HPV-positive women. Median, mean, and the interquartile range of enrollment ages for this cohort were 40, 43, and 34–49 years, respectively. Of the 49.8% of women who self-reported their race, 55.5% were caucasian, 15.8% were Asian or Pacific Islander, 13.7% were Latina, 9.8% were black or African American, and 5% were Native American, unknown, or other.
Cotests were excluded if the Pap result was unsatisfactory or “other” (eg, “exfoliated endometrial cells present in a woman [older than] 40”). Pap tests that were interpreted as atypical squamous cells of undetermined significance (ASC-US) or more severe cytology (ASC-US+) were considered screen positives. We used CIN 2+ as our primary end point because CIN 2 is the clinical threshold for treatment. We also used CIN 3+ in some analyses, recognizing that CIN 2 represents an equivocal diagnosis of cervical precancer8,9 whereas CIN 3 is a more definite cervical cancer precursor and is less likely to regress. A sample of 9.9% of all of the biopsy reports was subsequently rereviewed by the investigators to verify the accuracy of the data entry.
We calculated the annual cumulative incidence rates of CIN 2+ or CIN 3+, as a metric of absolute risk, after an index HPV-positive, Pap-negative result as previously described.6,10,11 Briefly, we used a modified Kaplan-Meier method to reflect that diagnosis can occur only during a clinic visit.10 We treated any return visit sooner than 6 months as clinical follow-up of the index screening result. We used 12-month intervals to define subsequent, yearly follow-up intervals from enrollment (6–18 months, 18–30 months, and so on). We estimated the frequency of observing the end point within each interval after the first visit, and calculated the cumulative risk through the end time of a given interval as the complement of the product of the complements of the conditional probabilities of an event at earlier intervals and the conditional probability of an event in the given interval. We estimated the probability of an event from the start of the 1-year interval through the end of interval K as
where p 0=0 and p i, i=1, …, k are the frequency of the event among the women with a visit within the interval i. Each summation term indexed by k is the conditional probability that a woman is diagnosed in interval k, given she had not been diagnosed in intervals 1 to k−1; the product factors are the complements of the probability of diagnosis in the intervals from 1 to k−1. Our modified Kaplan-Meier method requires the assumption that women who had a visit during an interval had the same risk as women who had no visit during their visit, and that those who had an index visit earlier in the enrollment window were similar in terms of risk as those who had a later index visit. Therefore, only those who had an index visit in 2006 contributed to the 4-year risk estimates.
In a sensitivity analysis, women who had an index visit in 2008 were at a slightly lower risk than those who an index visit in 2006 but the relative risks related to screening history remained the similar (data not shown).
We reported the 1-year and 4-year cumulative incidence rates for CIN 2+ and CIN 3+. We reported on the 1-year cumulative incidence rates rather than disease found at the index visit because per current U.S. guidelines, not all women underwent colposcopy immediately. As a consequence, a few disease outcomes after a first HPV-positive, Pap-negative result, particularly CIN 2,12,13 may have gone undetected and regressed during the 1-year follow-up. Thus, we may slightly underestimated the cumulative incidence of CIN 2+ in the HPV-positive, Pap-negative subgroup. However, as recently shown, sending all women testing HPV positive immediately to colposcopy, especially those under the age of 35 years, artificially inflates the detection of CIN 2 with cases that would have otherwise regressed.4 Importantly, our analysis reflects the real-world risks of CIN 2+ and CIN 3+ that face U.S. clinicians managing women who test HPV positive.
We calculated cumulative incidence rates for all women with an index HPV-positive result, and stratified on combinations of the immediate, previous past Pap and HPV test results (unknown, positive, negative). Cumulative incidence rate ratios with 95% confidence interval (CI), as measures of relative risks, were calculated to compare the cumulative incidence rates between subgroups. A cumulative incidence rate ratio with 95% CI that does not encompass 1.0 indicates a statistically significant difference (P<.05) between the reference and the cumulative incidence rate of interest.
Figure 1 shows the cumulative incidence rates for CIN 2+ and CIN 3+ by Pap status (positive compared with negative) for HPV-positive women, irrespective of their screening history. The 1-year and 4-year cumulative incidence rates for CIN 2+ after an HPV-positive, Pap-negative result were 2.83 (95% CI 2.55–3.12) and 7.89 (95% CI 7.00–8.78), and for CIN 3+ were 1.07 (95% CI 0.89–1.25) and 3.63 (95% CI 2.87–4.38).
Although not the main point of this analysis, by comparison, the 1-year and 4-year cumulative incidence rates for CIN 2+ after an HPV-positive, Pap-positive result were 14.96 (95% CI 14.32–15.60) and 20.10 (95% CI 19.27–20.94), and for CIN 3+ were 6.85 (95% CI 6.39–7.30) and 9.37 (95% CI 8.76–9.98).
Among those with an index HPV-positive, Pap-positive result, those with high-grade cytology (high-grade squamous intraepithelial lesion [HSIL], atypical squamous cells cannot rule out HSIL, or atypical glandular cells) were at the greatest 4-year risk of CIN 3+ (cumulative incidence rate=32.60; 95% CI 30.41–34.79). Exclusion of these women did not appreciably change our results because of the rarity of these cytologic interpretations (data not shown). Those with an index low-grade squamous intraepithelial lesion or atypical squamous cells of undetermined significance Pap were at a similar 4-year risk of CIN 3+ (cumulative incidence rates=5.05 and 4.75, respectively), which was only slightly greater risk than after an index HPV-positive, Pap-negative result (cumulative incidence rate=3.64).
Among HPV-positive women regardless of whether the concurrent Pap was positive or negative, the 1-year cumulative incidence rate of CIN 2+ or CIN 3+ differed depending on the previous screening result (Table 2). Focusing on women with an index HPV-positive, Pap-negative result, the 1-year risk of CIN 2+ was greater for women who had a past HPV-positive, Pap-negative result (cumulative incidence rate=5.84, 95% CI 4.88–6.80) compared with those who had HPV-negative, Pap-negative result (cumulative incidence rate=2.32, 95% CI 1.79–2.84) (cumulative incidence rate ratio=2.52, 95% CI 2.24–2.80). The risk differences were even more noticeable for CIN 3+, with a cumulative incidence rate ratio of 4.20 (95% CI 3.68–4.71).
Similar patterns were observed for the 4-year cumulative incidence rate for CIN 2+ and CIN 3+ (Table 3). Using the same comparison as an example, the 4-year risk of CIN 2+ was greater for women who had a past HPV-positive, Pap-negative result (cumulative incidence rate=11.79, 95% CI 10.22–13.36) compared with those who had a HPV-negative, Pap-negative result (cumulative incidence rate=4.56, 95% CI 3.43–5.69) (cumulative incidence rate ratio=2.59, 95% CI 2.30–2.87). Likewise, the 4-year risk of CIN 3+ was greater for women who had a past HPV-positive, Pap-negative result (cumulative incidence rate=5.08, 95% CI 4.01–6.15) compared with those who had HPV-negative, Pap-negative result (cumulative incidence rate=1.12, 95% CI 0.63–1.60) (cumulative incidence rate ratio=4.54, 95% CI 4.05–5.02).
Notably, HPV-positive women whose screening history was unknown had comparable or even greater risks than those who had a preceding HPV-positive test. For example, the 4-year cumulative incidence rate of CIN 2+ and CIN 3+ for index HPV-positive, Pap-positive women with unknown past screening results were 23.34 and 12.57, respectively, similar to what we observed for those who had a past HPV-positive and Pap-positive result.
History of HPV results appeared to stratify the risk of CIN 2+ and CIN 3+ more significantly than history of Pap results. For example, the 4-year cumulative incidence rates of CIN 3+ for HPV-positive, Pap-negative women were 4.67 (cumulative incidence rate ratio=3.43, 95% CI 1.91–6.16) and 4.36 (cumulative incidence rate ratio=3.20, 95% CI 1.95–5.26) for a past HPV test that was unknown or positive, respectively, compared with 1.36 for a past negative HPV test. By comparison, the 4-year cumulative incidence rates of CIN 3+ for HPV-positive, Pap-negative women were 2.43 (cumulative incidence rate ratio=0.60, 95% CI 0.37–0.96) and 5.16 (cumulative incidence rate ratio=1.27, 95% CI 0.73–2.21) for a past Pap result that was unknown or positive, respectively, compared with 4.07 for a past negative Pap result.
Women younger than the median age of 41 were marginally more likely than women 41 and older to develop CIN 2+ (cumulative incidence rate ratio=1.21, 95% CI 1.11–1.31) and CIN 3+ (cumulative incidence rate ratio=1.18, 95% CI 1.00–1.36) over the 4-year follow-up. There were no differences in likelihood of developing CIN 2+ over the 4-year follow-up between those women who had a past screening visit occurred within 880 days compared with longer (data not shown).
Here, we show the significant variability in the clinical meaning (risk of CIN 2+ or CIN 3+) of a positive HPV test. Women with evidence of a persisting HPV infection, based on a past positive HPV test or abnormal Pap, were at a significantly higher risk than those with a newly detected HPV infection. The HPV testing history appeared to be qualitatively a more important modifier of the risk than the Pap history. Notably, those with an unknown screening history had a risk profile that most resembles women with persistent HPV infections. Some women with an unknown screening were those who had not been previously screened at Kaiser Permanente in Northern California. Many of those with no past HPV test but a past abnormal Pap appeared to be women around the age of 30 who transitioned from routine Pap screening under the age of 30 to cotesting when they became 30 and older (median age of 32 years).
There is overwhelming evidence that negative cervical cytology adds little to the reassurance against cervical precancer and cancer once a woman has tested HPV negative.3,4,10 However, as shown here, cytology provides very useful risk stratification among HPV-positive women aged 30 and older. These and other data provide the rationale for a shift from using HPV testing to triage ASC-US cytology to cotesting or HPV test–based primary screening with cytology triage of HPV-positive women aged 30 and older. Adding HPV testing into routine cervical cancer screening provides durable safety against cervical precancer and cancer afforded by a negative HPV test, thereby permitting screening interval extension, whereas cytology provides the necessary specificity and risk stratification. HPV testing aptly identifies a subset of HPV-positive, Pap (cytology)–negative women 30 and older who are at future risk of CIN 3+ and warrant closer surveillance. Good follow-up of the HPV-positive, Pap-negative women, as observed at Kaiser Permanente in Northern California, is critical to maximizing the benefits of adding HPV testing into routine cervical cancer screening.14,15
Current guidelines recommend that women with an HPV-positive, Pap-negative result be followed for a year and rescreened, with a positive screening result triggering referral to colposcopy. We observed that the women with two consecutive HPV-positive, Pap-negative results (ie, past and index HPV-positive, Pap-negative result) had relatively low risk of CIN 3+ over 4 years, much lower than women who had index abnormal Pap, suggesting that these women with repeat HPV-positive, Pap-negative result might be managed less aggressively by continued annual surveillance rather than immediate colposcopy. Yet their management represents a clinical dilemma considering that these women are still at risk for cervical cancer.4 A better solution would be the use of additional biomarkers, such as HPV genotyping16 or p16INK4a immunocytochemistry,17 to identify those within this subgroup at greater risk for CIN 3+.
Finally, current clinical algorithms do not account for a number of important modifiers of risk after a positive HPV test result, such as duration of HPV infection. In the future, HPV vaccination will further complicate risk assessment by providing variable, age-of-vaccination dependent protection against the two most carcinogenic HPV genotypes, HPV16 and HPV18.18 The variability in the clinical meaning of a HPV test due to factors such as screening and HPV vaccination history as well as concurrent Pap result underscores the need to move from algorithm-based management19 to risk estimation that integrates complex clinical information to optimize clinical management.20,21 Management and follow-up of screening results is most rationally defined by risk of current CIN 2+ or CIN 3+. Women with similar risk should be managed in a similar fashion regardless of how that risk is determined.20,21 Optimal management, which balances the risks and harms, can be designed using increasingly extensive real-world and accurate measurements of risk by test results and history, using real-life databases from Kaiser Permanente Northern California and elsewhere.
1. Wright TC Jr, Massad LS, Dunton CJ, Spitzer M, Wilkinson EJ, Solomon D; 2006 ASCCP-Sponsored Consensus Conference. 2006 consensus guidelines for the management of women with abnormal cervical screening tests. J Low Genit Tract Dis 2007;11:201–22.
2. Wright TC Jr, Schiffman M, Solomon D, Cox JT, Garcia F, Goldie S, et al. Interim guidance for the use of human papillomavirus DNA testing as an adjunct to cervical cytology for screening. Obstet Gynecol 2004;103:304–9.
3. Dillner J, Rebolj M, Birembaut P, Petry KU, Szarewski A, Munk C, et al. Long term predictive values of cytology and human papillomavirus testing in cervical cancer screening: joint European cohort study. BMJ 2008 Oct 13;337:a1754. doi: 10.1136/bmj.a1754.
4. Ronco G, Giorgi-Rossi P, Carozzi F, Confortini M, Dalla Palma P, Del Mistro A, et al. Efficacy of human papillomavirus testing for the detection of invasive cervical cancers and cervical intraepithelial neoplasia: a randomised controlled trial. Lancet Oncol 2010;11:249–57.
5. Kjaer S, Hogdall E, Frederiksen K, Munk C, van den Brule A, Svare E, et al. The absolute risk of cervical abnormalities in high-risk human papillomavirus-positive, cytologically normal women over a 10-year period. Cancer Res 2006;66:10630–6.
6. Castle PE, Rodriguez AC, Burk RD, Herrero R, Wacholder S, Alfaro M, et al. Short term persistence of human papillomavirus and risk of cervical precancer and cancer: population based cohort study. BMJ 2009 Jul 28;339:b2569. doi: 10.1136/bmj.b2569.
7. Rodriguez AC, Schiffman M, Herrero R, Hildesheim A, Bratti C, Sherman ME, et al. Longitudinal study of human papillomavirus persistence and cervical intraepithelial neoplasia grade 2/3: critical role of duration of infection. J Natl Cancer Inst 2010;102:315–24.
8. Castle PE, Stoler MH, Solomon D, Schiffman M. The relationship of community biopsy-diagnosed cervical intraepithelial neoplasia grade 2 to the quality control pathology-reviewed diagnoses: an ALTS report. Am J Clin Pathol 2007;127:805–15.
9. Schiffman M, Castle PE, Jeronimo J, Rodriguez AC, Wacholder S. Human papillomavirus and cervical cancer. Lancet 2007;370:890–907.
10. Sherman ME, Lorincz AT, Scott DR, Wacholder S, Castle PE, Glass AG, 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.
11. Rodriguez AC, Schiffman M, Herrero R, Wacholder S, Hildesheim A, Castle PE, et al. Rapid clearance of human papillomavirus and implications for clinical focus on persistent infections. J Natl Cancer Inst 2008;100:513–7.
12. Castle PE, Schiffman M, Wheeler CM, Solomon D. Evidence for frequent regression of cervical intraepithelial neoplasia-grade 2. Obstet Gynecol 2009;113:18–25.
13. Stoler MH, Vichnin MD, Ferenczy A, Ferris DG, Perez G, Paavonen J, et al. The accuracy of colposcopic biopsy: analyses from the placebo arm of the Gardasil clinical trials. Int J Cancer 2010 May 20 [Epub ahead of print].
14. Kitchener HC, Almonte M, Thomson C, Wheeler P, Sargent A, Stoykova B, et al. HPV testing in combination with liquid-based cytology in primary cervical screening (ARTISTIC): a randomised controlled trial. Lancet Oncol 2009;10:672–82.
15. Sasieni P, Castle PE, Cuzick J. Further analysis of the ARTISTIC trial. Lancet Oncol 2009;10:841–2.
16. Khan MJ, Castle PE, Lorincz AT, Wacholder S, Sherman M, Scott DR, et al. The elevated 10-year risk of cervical precancer and cancer in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of type-specific HPV testing in clinical practice. J Natl Cancer Inst 2005;20:97:1072–9.
17. Carozzi F, Confortini M, Dalla Palma P, Del Mistro A, Gillio-Tos A, De Marco L, et al. Use of p16-INK4A overexpression to increase the specificity of human papillomavirus testing: a nested substudy of the NTCC randomised controlled trial. Lancet Oncol 2008;9:937–45.
18. Kjaer SK, Sigurdsson K, Iversen OE, Hernandez-Avila M, Wheeler CM, Perez G, et al. A pooled analysis of continued prophylactic efficacy of quadrivalent human papillomavirus (Types 6/11/16/18) vaccine against high-grade cervical and external genital lesions. Cancer Prev Res (Phila) 2009;2:868–78.
19. Wright TC Jr, Massad LS, Dunton CJ, Spitzer M, Wilkinson EJ, Solomon D; 2006 American Society for Colposcopy and Cervical Pathology-sponsored Consensus Conference. 2006 consensus guidelines for the management of women with cervical intraepithelial neoplasia or adenocarcinoma in situ. J Low Genit Tract Dis 2007;11:223–39.
20. Castle PE, Sideri M, Jeronimo J, Solomon D, Schiffman M. Risk assessment to guide the prevention of cervical cancer. J Low Genit Tract Dis 2008;12:1–7.
21. Katki HA, Wacholder S, Solomon D, Castle PE, Schiffman M. Risk estimation for the next generation of prevention programmes for cervical cancer. Lancet Oncol 2009;10:1022–3.