|Home | About | Journals | Submit | Contact Us | Français|
An update to the American Cancer Society (ACS) guideline regarding screening for the early detection of cervical precancerous lesions and cancer is presented. The guidelines are based on a systematic evidence review, contributions from six working groups, and a recent symposium cosponsored by the ACS, American Society for Colposcopy and Cervical Pathology (ASCCP), and American Society for Clinical Pathology (ASCP), which was attended by 25 organizations. The new screening recommendations address age-appropriate screening strategies, including the use of cytology and high-risk human papillomavirus (HPV) testing, follow-up (e.g., management of screen positives and screening interval for screen negatives) of women after screening, age at which to exit screening, future considerations regarding HPV testing alone as a primary screening approach, and screening strategies for women vaccinated against HPV16 and HPV18 infections.
Cervical cancer screening has successfully decreased cervical cancer incidence and mortality. The ACS Guideline for the Early Detection of Cervical Cancer was last reviewed and updated in 2002; for the first time, those recommendations incorporated human papillomavirus (HPV) DNA testing.1 Since that time, numerous studies have been published that support changes to recommended age-appropriate screening as well as the management of abnormal screening results, as summarized in Table 1.2
High-quality screening with cytology (Pap testing) has markedly reduced mortality from squamous cell cervical cancer, which comprises 80–90% of cervical cancers.3–5 Since the introduction of cervical cytology in United States in the mid-20th century, cervical cancer, once the most frequent cause of cancer deaths in women, now ranks 14th for cancer deaths.6 This reduction in mortality through screening is due to 1) an increase in detection of invasive cancer at early stages, where the five-year survival rate is approximately 91 percent7 and 2) detection and treatment of pre-invasive lesions, which reduces the overall incidence of invasive cancer. In 2012, an estimated 12,170 cases of invasive cervical cancer will be diagnosed, and an estimated 4,220 women will die.6
It is now understood that persistent cervical infection with high-risk human papillomavirus (HPV) genotypes (“types”) is necessary for the development of cervical cancer and its immediate precursor lesion (“precancer”), cervical intraepithelial neoplasia (CIN) grade 3 (CIN3). Epidemiologic case series have shown that nearly 100% of cervical cancer cases test positive for HPV.8 HPV16 is the most carcinogenic HPV genotype and accounts for approximately 55%-60% of all cervical cancers.8–10 HPV18 is the next most carcinogenic HPV genotype, and accounts for 10–15% of cervical cancers.8–10 Approximately ten other HPV genotypes cause the remaining 25–35% of cervical cancers. HPV causes all common and most rare histologic types of cervical cancer. HPV18 causes a greater proportion of glandular cancers, adenocarcinoma and adenosquamous carcinoma, than squamous cell carcinoma (approximately 32% vs. 8%, respectively.9 The establishment of the causal link between HPV and cervical cancer, along with an understanding of the epidemiology and natural history of HPV infection, has led to a new model for cervical carcinogenesis: HPV acquisition, HPV persistence (versus clearance), progression to precancer, and invasion,11,12 which helps guide age-appropriate interventions to prevent cervical cancer.
Genital HPV is acquired through sexual and genital skin-to-skin contact. In most populations, prevalence peaks within a few years after the median age of sexual debut, which in the U.S. is 17 years.13 Most (~90%) HPV infections are transient, becoming undetectable within one to two years.14,15 Women whose infections persist are at significant risk of developing precancerous lesions. One-year16 and two-year HPV persistence,17 especially by HPV16, strongly predict CIN3 or more severe diagnoses (CIN3+) in the subsequent years (e.g., 20–30% risk of CIN3+ over 5 years for one-year or two-year persistent HPV16). Untreated CIN3 has a 30% probability of becoming invasive cancer over a 30-year period, although only about 1% of properly treated CIN3 will become invasive.18
The fundamental goal of cervical cancer screening is to prevent morbidity and mortality from cervical cancer. The optimal screening strategy should identify those cervical cancer precursors likely to progress to invasive cancers (maximizing the benefits of screening) and avoid detection and unnecessary treatment of transient HPV infection and its associated benign lesions that are not destined to become cancerous (minimizing the potential harms associated with screening). Cytology (Pap test) screening has been very successful in lowering cancer incidence and mortality in countries where good quality screening is available, yet false-positive results are common, since most abnormal (atypical squamous cells of undetermined significance or more severe) cytology is not associated with concurrent CIN3 or cancer, and therefore still a concern.19, 20 Increased understanding of the association between HPV and cervical cancer risk has led to the development of molecular tests for HPV* that offer increased sensitivity albeit lower specificity compared to cytology. HPV tests may better forecast which women will develop CIN3+ over the next 5–15 years than cytology.21–23 Incorporation of HPV testing into cervical cancer screening strategies has the potential to allow both increased disease detection (improving benefits) and increased length of screening intervals (decreasing harms such as psychosocial impact of screening positive, additional clinical visits and procedures, and treatment of lesions destined to resolve). In the development of this evidence-based review and guideline update, we considered the tradeoffs of benefits and harms of screening while considering different screening modalities and ages.
Of note, approximately half of the cervical cancers diagnosed in the United States are in women never screened, and an additional 10 percent of cancers occur among women not screened within the past five years.24–26 The current opportunistic approach to cervical cancer screening in the U.S. fails to reach sub-populations of women mainly living in low-resource, medically underserved regions, and thus invasive cervical cancer is one among a complex of diseases strongly linked to socioeconomic, geographic, and/or racial disparities. Annual rates of cervical cancer incidence and mortality in these populations are several-fold higher than in the rates in the general U.S. population and are similar to the rates observed in some lower-income countries.24–26
Technologic improvements in screening are unlikely to have a substantial impact on mortality if they do not reach this population.27 While this new ACS-ASCCP-ASCP screening guideline includes a review of molecular screening tests and strategies, perhaps the largest immediate gain in reducing burden of cervical cancer incidence and mortality could be attained by increasing access to screening (regardless of the test used) among women who are currently unscreened or screened infrequently. Incorporation of HPV testing may offer advantages over what is already a successful screening strategy if utilized, i.e., cytology. For example, HPV testing provides longer-term safety following a negative test than cytology, a useful characteristic for the infrequently screened.
From 2009 to 2011, the ACS, ASCCP, and ASCP worked collaboratively to convene an expert panel to develop new screening recommendations based on a systematic review of the available evidence. The process was overseen by a Steering Committee composed of representatives from the sponsoring organizations, other stakeholder organizations and agencies, and experts representing multiple disciplines (see page 52 for names of all committee members). An independent Evidence Evaluation Committee (the “Data Group”) composed of experts in literature reviews, evidence evaluation, and data analysis had primary responsibility for overall development and implementation of the guidelines process, and for providing feedback and guidance to the Working Groups. The six topic areas to be addressed by the recommendations were identified by the Steering Committee. A Working Group consisting of experts on a particular topic and representing different disciplines was assigned to each area, with each Working Group having a member of the Data Group serving as a liaison. Each group met regularly via teleconference, including web-based conferences for all participants to review specific methodologic issues.
The six working groups addressed the following topic areas:
The working groups were instructed to propose evidence-based cervical cancer prevention strategies that best serve women, specifically balancing benefits and harms of screening and, in some cases, management of screening results. They were specifically directed not to consider financial cost in making their recommendations.
In planning this workshop, the Steering Committee critically examined some of the issues involved in defining conflict-of-interest (COI) and recognized that all interests – whether directly financial or more indirect such as an affiliation with a company, the success of one’s clinical practice, or the prominence of a professional specialty – represent potential conflicts. Steering Committee members, Working Group and Data Group co-chairs, and members of the Writing Committee were required to not have any financial ties to companies that market or sell screening tests or devices (e.g., methods to visualize the cervix such as colposcopes). All participating individuals were required to disclose all real or potential conflicts of interest. Employees or representatives of industry and insurance companies were not invited to participate in the development of these guidelines due to their significant, direct financial interests in the outcome of these guidelines. The complete COI policy can be found in the online supplement.
The six Working Groups independently considered a series of screening and management questions. We recognized that different groups of experts could evaluate the same data for related questions and reach different conclusions because of differences in weighing benefits and harms of screening. Therefore we harmonized the main outcomes for benefits and harms across Working Groups as defined below.
Ideally, the screening test(s) should efficiently and accurately identify women with precancer who are at significant risk of developing cancer, so that appropriate intervention will prevent progression to invasive cancer. We used detection of CIN3 as the measure of a screening test’s sensitivity for precancer because a substantial proportion of women with CIN3 would develop invasive cervical cancer if left untreated.18 Also, given the natural history of cervical carcinogenesis and the relative rarity of cervical cancer in screened populations, it is the most feasible directly measurable outcome in controlled clinical studies. By contrast, CIN2 is an equivocal diagnosis that includes some precancer lesions (CIN3) but also some lesions (e.g., CIN1) that would regress on their own.28–30 Although CIN2 is the widely accepted threshold for treatment2 -- to provide an additional margin of safety -- we posited that CIN2 should not be the primary target of cervical cancer screening.
Ideally, the screening interval for a particular testing modality should be chosen such that development of invasive cancer is highly unlikely before the next screen. Because few studies have sufficient numbers of cancer cases to assess cancer risk directly, we considered the absolute risk of CIN3, including the rare cases of cancer (CIN3+) between or at the next visit as our best measure of incident cervical cancer risk. When available and appropriate, we also noted the risk of invasive cancer, especially in relationship to screening intervals, following a negative screening test.
Thus, for these guidelines, we judged that a screening test or modality provided greater benefit if MORE CIN3 was detected immediately by the screening test, and risk of CIN3+ was reduced in the interval before the next screening test.
Most episodes of HPV infection and many CIN1 and CIN2 cases are transient and will not develop into CIN3 or cancer.28–30 The potential harms associated with detecting these transient lesions include the anxiety associated with a “positive” cancer screening test, potential stigmatization from the diagnosis of a sexually transmitted infection, discomfort from additional diagnostic and treatment procedures, bleeding from treatment, and, longer term, an increased risk of pregnancy complications such as preterm delivery due to treatment31. Having a positive test at any point in one’s life may contribute to a perception of an increased risk of cancer, and a subsequent desire for more testing, further increasing the likelihood of another positive test.32 Although any false positive test has the potential for inducing anxiety or other psychological distress, quality-of-life instruments are rarely included in controlled clinical trials of screening. Because of this, we used the number of colposcopies, both alone and relative to CIN3+ and cancer detected, as the primary measure of harm, since colposcopies themselves are associated with physical discomfort, and are a necessary pre-requisite to more invasive treatments with greater short- and long-term risks of harms. Since the number of subjects undergoing colposcopy is usually reported in controlled studies, and more screening leads to more screen positives and therefore more colposcopy, it provides a surrogate for potential harm of screening analogous to the use of the detection of CIN3 as a surrogate for cervical cancer for potential benefits of screening.
Our basic tenets regarding risk and risk-based interventions were as follows:
We recognize that the women at different ages may have different tradeoffs in benefits and harms from screening. These differences are addressed through the development of age-specific screening recommendations.
Based on good evidence showing similar sensitivity and specificity of conventional and liquid-based cytology for CIN grade 2 or more severe diagnoses (CIN2+),35,36 we included studies that used either cytology method. We found no data to suggest a need to analyze data from studies using liquid-based cytology separately from those using conventional cytology.
The hallmarks of HPV testing are greater sensitivity but lower specificity for CIN3+ and CIN2+37–43 and better reproducibility than cytology.44–46 Benchmarks for clinical performance of HPV testing are described in detail elsewhere.47,48 Our general assumptions are that the sensitivity of HPV testing for CIN3+ and CIN2+ should be ≥90% and the percent of women in the general population who test (screen) positive, as a measure of false positives, should be less than or equal to established thresholds from well-validated HPV DNA tests.47,48
Several U.S. Food And Drug Administration (FDA)-approved HPV tests are commercially available, though none is yet approved for primary, stand-alone screening. The performance characteristics vary among these HPV tests, and comparability cannot be assumed. For use in the U.S., HPV tests should both be FDA approved (for validity) and meet specific criteria for clinical performance as described above.47,48 Other well-validated tests (e.g., GP5+/6+ EIA) are commercially available in Europe, and data using these tests were included in our review,39,40 but these are not FDA-approved. HPV tests not meeting these standards of performance (including FDA-approved tests) should not be used. In particular, excessive analytic sensitivity is a significant concern, as it will be unlikely to improve already very high clinical sensitivity for CIN3+ but may increase harms due to poorer specificity.49 The updated guidelines for cervical cancer screening described here were developed based on HPV tests that have performance characteristics similar to those of the HPV tests used in the supporting evidence. The guidelines cannot be expected to perform as designed (i.e., to balance benefits and harms) when using HPV tests with different performance characteristics.
Laboratory-developed tests (LDTs), which are currently exempt from the FDA regulatory oversight, rarely have undergone the necessary evaluation using clinical endpoints of CIN3+ and CIN2+ in properly designed studies. Therefore, we recommend against the use of LDTs for cervical cancer screening.
HPV tests should be used in accordance with their package labeling. Professional medical organizations with clinical expertise in gynecological cancer may recommend off-label applications of FDA-approved tests (e.g., HPV testing for post-treatment follow-up as recommended by the ASCCP).2 Laboratory standard operating procedures and robust quality assurance programs should accompany the use of any HPV test. Inter-laboratory or proficiency testing to ensure quality results across laboratories should be established.50 While well validated in the research setting, additional studies of inter-laboratory comparability of HPV testing in the clinical laboratory setting would be helpful.
We utilized the Grading Recommendations Assessment, Development, and Evaluation (GRADE) system51–56 to provide a framework for the guidelines development process. An initial literature search for terms relevant to all the Working Groups was performed, and abstracts were reviewed by Data Group members. Papers meeting initial inclusion criteria were retrieved and distributed to each Working Group as appropriate. The search included articles from 1995 through July 5, 2011. (See online supplement, Figure 1 and Figure 2 for details.)
Each Working Group took the initially defined areas and formulated specific questions using the GRADE framework.51–56 From an initial list of potential outcomes identified by the Data Group, each Working Group defined 3–4 outcomes as “Critical,” 3–4 as “Important,” and 3–4 as “Useful” (see online supplement for list of outcomes). Members of the Working Groups then reviewed each article to determine if data were available on critical or important outcomes. We did not perform formal data synthesis or meta-analysis to create single summary estimates for each outcome/intervention pair. Instead, summary data from each included article, along with a quality grade of “high, moderate, low, or very low” were presented to the group, with a subsequent quality grade for the entire body of evidence for a given outcome/intervention pair.
The GRADE system does not specifically address modeling studies, which were frequently the only evidence available for comparing alternatives, particularly different screening intervals. Because modeling integrates evidence from a wide range of sources of varying quality, we considered individual modeling studies as “low” quality evidence, but, if the individual studies followed best practices for model-based analyses, and the results were consistent across studies done by different groups using different methods, the rating of the overall body of evidence based on modeling could be graded as of “moderate” quality.
Based on the initial grading of evidence, each Working Group formulated an initial summary recommendation, graded as “Strong” or “Weak,” based on the overall quality of the evidence for outcomes considered “Critical,” as well as additional criteria such as variation in patient preferences (if data were available) and feasibility of obtaining additional evidence. A “strong” recommendation means that the group felt confident that further research would be unlikely to change the recommendation, based on the overall quality of the available evidence, the prospect of obtaining better evidence, and the balance between benefits and harms. A “weak” recommendation means that there is substantial uncertainty surrounding the balance of benefits and harms, and further research is needed to increase confidence in the results, or that benefits and harms are closely balanced, with decisions based largely on individual preferences and values.* Members of the Steering Committee and Data Group, as well as the other Working Groups, reviewed these recommendations and corresponding rationale and provided feedback. After revision, the draft recommendations and rationale were posted on the ASCCP website for public comment from October 19, 2011 to November 9, 2011. The public comments were distributed to each Working Group, and revisions were made to address or clarify issues raised. However, each Working Group had the ultimate authority and responsibility for the (revised) draft recommendations presented at the symposium for consideration.
A symposium was held November 17–18, 2011 to discuss, revise as necessary and vote on the final recommendations. In addition to the members of the Steering Committee, Data Group and Working Groups, representatives from other stakeholder organizations were invited (see online supplement for list). Each Working Group presented their evaluation of the evidence and draft recommendations. After the presentation, there was an open discussion, followed by voting on the recommendations, including both the wording of the recommendation and the strength of the recommendation. A two-thirds majority was required for a recommendation to be accepted—if this threshold was not achieved, the recommendation was revised by the Working Group and brought back to the plenary participants for voting.*
These guidelines were developed to address cervical cancer screening in the general population. These guidelines do not address special, high-risk populations who may need more intensive or alternative screening. These special populations include women 1) with a history of cervical cancer, 2) who were exposed in utero to diethylstilbestrol (DES), and 3) who are immune-compromised (e.g., infection with human immunodeficiency virus).57
The following recommendations are based on review and assessment of the published peer-reviewed literature available at the time of the symposium. It is anticipated that they will be reviewed on an ongoing basis and revised as new evidence becomes available about the impact of alternative strategies on the balance of benefits and harms associated with cervical cancer screening.
The question of when to begin screening was addressed at the 2009 Practice Improvement in Cervical Screening and Management (PICSM) Symposium on Management of Cervical Abnormalities in Adolescents and Young Women33 and was not part of the current review. The following recommendation has been previously endorsed by the several organizations that participated in that meeting.
Cervical cancer screening should begin at age 21 years. Women under the age of 21 should not be screened regardless of the age of sexual initiation or other risk factors.
Cervical cancer is rare in adolescents and young women58 and may not be prevented by cytology screening.59 The incidence of cervical cancer in this age group has not changed with increased screening.58 Screening adolescents leads to unnecessary evaluation and potentially to treatment of pre-invasive cervical lesions that have a high probability of regressing spontaneously. This overtreatment, and subsequent increased risk of reproductive problems, represents a net harm.33
Adolescent cervical cancer prevention programs should focus on universal HPV vaccination, which is safe, highly efficacious, and when used in adolescents before becoming sexually active, highly effective and cost-effective.60,61 Even without cervical cancer screening, it is critical that adolescents continue to have access to appropriate health care, including assessment of health risks, family planning and contraception, and prevention counseling, screening and treatment of sexually transmitted infections.33
Over time, growing evidence and the improved understanding of the natural history of cervical cancer have led to growing recognition that earlier recommendations for annual screening were excessive and led to an increased rate of harms. Today, there is little evidence to support annual screening of women at any age by any screening test, method, or modality. Annual screening leads to a very small increment in cancers prevented, at the cost of a very large excess of unnecessary procedures and treatments62,63 due to the high prevalence of transient, benign HPV infections and associated lesions, most of which will regress within a year or two or, of those that do not, are many years on average from causing cancer. Women at any age should NOT be screened annually by any screening method; rather, recommended screening intervals for women at average risk are based on age and clinical history.
For women 21–29 years of age, screening with cytology alone every 3 years is recommended. For women 21–29 years of age with 2 or more consecutive negative cytology results, there is insufficient evidence to support a longer screening interval (i.e. >3 years).
HPV testing should not be used to screen women in this age group, either as a stand-alone test or as a cotest with cytology.
For women under 30 years of age, there are few studies specifically addressing the optimal interval for cytology-based screening. Those studies meeting selection criteria were mainly modeling studies. While affording slightly greater cancer risk reduction, annual screening results in twice the number of colposcopies compared to screening every 3 years.63 Only one study modeled the trade-offs between cancers detected and colposcopies for screening every 2 years versus every 3 years in this age group.62 The results for both intervals conducted over a 10-year interval were similar for reducing the lifetime cancer burden. Combining these results with findings of other studies64,65 that showed no significant difference in cancer reduction between a 2- and 3-year screening interval, we determined that screening every 3 years provided the best balance of benefits and harms of screening in this age group.
Table 2(62,63,68–70) shows patient outcomes, number of studies and quality of the evidence. In the absence of screening, modeling predicts a lifetime risk of cervical cancer in the U.S. of approximately 31–33 incident cancers per 1000 women.63,70 In addition, these studies predict a lifetime risk of cancer associated with screening every 3 years of approximately 5 to 8 incident cancers per 1000 women.62,63,68–70 Screening every 2 years is associated with a lifetime cancer risk ranging from 4 to 6 incident cancers per 1000 women;63,69 screening annually is associated with a lifetime risk of about 3 per 1000 women.63,69 Early stage cervical cancer has a very high 5-year survival rate of 90%. The predicted lifetime risks of death due to cervical cancer associated with screening every 3 years, every 2 years and annually are even lower: 0.05, 0.05, and 0.03 per 1000 women, respectively.
With respect to harm, screening every 3 years is associated with a lifetime prediction of about 760 colposcopies per 1000 women, screening every 2 years with about 1080 per 1000 women (a 40% increase vs. screening every 3 years), and screening every year is predicted to double the total number of lifetime colposcopies to close to 2000 per 1000 women, or nearly three times the rate expected from screening every 3 years.
A modeling study that examined outcomes for women 20 years of age screened over a subsequent 10-year time period,62 predicted that there would be a doubling of the colposcopies (using ASC-US as the threshold for referral) per 1000 women with annual screening compared with screening every 3 years. These results are similar to those reported by Kulasingam63 who examined outcomes associated with screening every 2 years. Compared to screening every 3 years, screening every 2 years (starting screening at any age between 15 and 25) was associated with little additional patient benefit in terms of reduced lifetime risk of cancer modeled over a slightly shorter time period than reported by Stout (9 versus 10 years).62
There is insufficient high quality evidence from RCTs to increase screening interval based on prior negative cytology results for any age group. Miller71 calculated the risk of invasive cervical cancer associated with different intervals since the last negative cytology test. The odds ratio comparing a 3-year vs. 2-year interval was 1.20 (95% confidence interval (0.65, 2.21). Adjusting for a history of an abnormal cytology or prior consecutive negative cytology tests did not substantially change the results. The authors also reported an incremental rise in cancer risk (≥3.16) over time as the interval from the last negative cytology test moved beyond three years72 and did not find a significantly reduced risk of CIN3+ associated with increasing numbers of prior negative cytology tests after controlling for time since last normal cytology test.
Women ages 30–65 years should be screened with cytology and HPV testing (“cotesting”) every 5 years (preferred) or cytology alone every 3 years (acceptable). There is insufficient evidence to change screening intervals in this age group following a history of negative screens.
For women 30–65 years of age, even with a history of negative cytology tests, the limited available evidence does not support a longer screening interval than 3 years. Studies of screening intervals in women with a history of negative cytology results report an increased risk of cancer after 3 years even after controlling for prior number of negative cytology tests.71 Furthermore, the only modeling study that examined the screening interval among US women with a history of prior normal cytology results compared screening every year to screening every 3 years.73 A longer interval was not examined in this review, although some countries (e.g., The Netherlands) use a 5-year interval. Modeling studies have shown a stepwise increase in cancer risk with increasing interval from 1 year to 3 years to 5 years.62,63 As such, we concluded that a 3-year interval for cytology provides an appropriate balance of benefits and harms and that there was insufficient evidence to support a longer interval than every 3 years utilizing cytology alone in women aged 30 years or older, even with a screening history of consecutive negative cytology tests.
In the majority of studies reviewed, the addition of HPV testing to cytology resulted in an increased detection of prevalent CIN3 with a concomitant decrease in CIN3+ or cancer detected in subsequent rounds of screening.39–41 This increase in diagnostic lead-time with cotesting translates into lower risk following a negative screen, permitting a longer interval between screens with incident cancer rates similar to or lower than screening with cytology alone at shorter intervals. If the incident cervical cancer rates associated with cytology at 3-year intervals are acceptable, cotesting at 5-year intervals provides similar or lower cancer risk.22,74
The addition of HPV testing to cytology also enhances the identification of women with adenocarcinoma of the cervix and its precursors.42,74 Compared to squamous cell cancers, cytology has been relatively ineffective in decreasing the incidence of invasive adenocarcinoma of the cervix.75,76 A strategy of cotesting may become increasingly important based on evidence of increasing incidence of adenocarcinoma, which has been observed in several European countries76 and the U.S.77 that have exclusively or primarily used cytology-only screening.
Cotesting at a 3-year interval, as recommended in interim guidelines from 2002 and 2004,1,78 resulted in a significantly smaller diagnostic yield of CIN3+ in the second round of cotesting following a prior negative cotest,74 supporting the recommendation to use a longer interval between cotests. Modeling studies have shown that cotesting in 40-year-old women at a 3-year versus a 5-year interval over a 10-year period only slightly decreases lifetime cervical cancer risk (0.39% vs. 0.61%, respectively) while significantly increasing the number of colposcopic evaluations.62 In the same modeling study, the lifetime cancer risk estimates for 40-year-old women undergoing 3-year cytology versus 5-year cotesting over a 10-year period were comparable (0.69% vs. 0.61%, respectively). Cotesting more frequently than at 3-year intervals, and especially annual screening, is predicted to further exacerbate the harms of increases in colposcopic referrals and treatments.62 Unfortunately, there is evidence that cotesting is being used at shorter intervals than 3 years.79 The lack of greater benefits and the increase in potential harms associated with screening more frequently support a recommendation of cotesting every 5 years.
Compared with cytology, HPV testing is more sensitive but less specific for identifying women with prevalent CIN3+.80 In a meta-analysis,80 the sensitivity of HPV testing for CIN3+ was 37% greater than that of cytology using a positive cut point of LSIL (i.e., LSIL or more severe cytologic abnormalities were considered screen positive), while the specificity of HPV testing was 7% lower. The sensitivity of HPV tests for CIN3+ was 28% greater than that of cytology at a positive cutpoint of ASC-US, while the specificity of HPV tests and cytology were the same.
When compared with women with negative cytology, those with negative HPV tests have a lower subsequent risk of CIN3+22,74 and, more importantly, cancer.41 Results from FDA-approved or well-validated HPV tests are also more reproducible (intra-assay reliability) than cytology.45,46,81
There are four RCTs39–41,82 in which two rounds of screening are reported comparing cotesting with cytology and HPV testing to cytology alone; three of those trials provided adequate evaluation of HPV-positive, cytology-negative results. Each of the trials had a complex protocol and each differed in the way that HPV-positive women were evaluated. ARTISTIC (a randomized trial of HPV testing in primary cervical screening) differed from the other three in that the lower age limit of eligible women was 20 years, while for the other three trials, it was 32 years, 29 years, and 35 years.39–41 The ARTISTIC trial82 also differed in that women in ARTISTIC who tested HPV positive, cytology negative were referred to one-year follow-up, similar to the interim guidelines in the U.S.,1,78 rather than being immediately referred to colposcopy as was done in the other trials.41 As the added benefit of HPV testing is only realized with thorough follow-up of and disease ascertainment in women with HPV-positive, cytology-negative results, it is not surprising that ARTISTIC did not show a benefit to cotesting versus cytology alone, with only half of the HPV-positive, cytology-negative women returning in a year. Saseini83 pointed out that if all of the HPV-positive, cytology-negative women had been evaluated and disease rates in those who were lost to follow-up had been comparable to those found among the women evaluated, ARTISTIC would have demonstrated increased sensitivity with the addition of HPV testing to cytology, as observed in other studies, compared to cytology alone.83 The other three trials39–41 were powered to detect differences in the rate of CIN3+ in the second round of screening, but not powered to detect differences in the rate of cancer in the second round of screening. Their protocols and results are described in detail in the online supplement.
In each of the three trials considered,39–41 the cotesting arm detected a greater proportion of CIN3+ in the first round of screening compared to cytology alone. The difference in the incidence of cancer in the second round of screening was not stated in Naucler et al.,39 showed a trend towards declining incidence (not statistically significant) in Bulkmans et al. (0.08% vs. 0.02%),40 and showed a statistically significant decrease in Ronco et al. (0.03% vs. 0%, p=0.004).41 The number of colposcopy referrals in the three studies was not clearly stated, so the increased number of colposcopies must be inferred based on modeling.
Based on the significant increase in sensitivity in detection of CIN3+ at the first round of screening and the reduction in invasive cancer in the second round of screening in Ronco et al,41 we concluded that the addition of HPV testing to cytology is beneficial. The main harms associated with adding HPV testing-- the increased referral to colposcopy and diagnosis of CIN2, some of which would regress without intervention-- can be mitigated by extending the screening interval to 5 years (as discussed below) and thereby reducing the detection of transient HPV infections and related lesions that would trigger clinical follow-up in low-risk women.62
While cotesting is preferred to cytology alone based on risks and harms assessment, such a strategy might not be feasible in all clinical settings in the U.S. due to a lack of payment for cotesting or due to local policies. With regards to the primary goal of cervical cancer screening, which is prevention of cervical cancer, a cytology-based screening strategy in women over 30 has been and continues to be an acceptable option. As mentioned previously in this document, a more frequent, cytology-only strategy does lead to more colposcopy and other harms including the potential need for prescribing shorter screening intervals due to equivocal cytology results that have minimal cancer risk.
Cotesting has increased sensitivity for detecting CIN3+ compared to cytology. Consequently, women screened with cotesting also have a lower subsequent risk of CIN3+ and invasive cancer, permitting a lengthening of screening intervals. Seven observational studies involving 24,295 women were pooled to examine the long-term predictive values of HPV testing and cytology.22 The 6-year risk of CIN3+ following a negative HPV test was 0.27%, compared to 0.28% among cotest negatives. By comparison, the 6-year risk of CIN3+ following a negative cytology alone was significantly greater at 0.97%. The authors also noted that the risk of CIN3+ at a 3-year screening interval, the most commonly used screening interval in Europe, after negative cytology was 0.51%. In a retrospective observational study of 330,000 women aged 30 years and older undergoing cotesting in routine clinical practice,74 the 3-year risk of CIN3+ following negative cytology alone (regardless of the HPV result) was 0.17%; the 5-year risk of CIN3+ following a negative HPV test alone (regardless of the cytology result) was 0.17%; and the 5-year risk of CIN3+ following a negative cotest was 0.16%, essentially comparable results across each testing strategy. Likewise, the risks of cancer also were comparable (0.018%, 0.019%, and 0.016%, respectively).
In the same analysis,74 women who were cotest negative at the initial screening and HPV and/or cytology positive 3 years later were at a lower risk of CIN3+ or cancer than women with a positive HPV and/or cytology result at the initial screen. This lower risk associated with previously normal findings presumably is due to the prolonged period of HPV carriage (chronic infections) required for invasive cancer to develop. Taken together, these reports indicate that healthcare providers can rely on the negative predictive value of the HPV test to assure women who cotest negative that they are at very low risk for CIN3 and cancer for at least 5 years after negative cotesting.
Modeling from several sources indicates that there is a dramatic increase in colposcopy rate with minimal change in invasive cancer incidence as screening intervals decrease below 3 years, regardless of the modality employed.63,84,85 Despite differing assumptions, all three analyses indicated that the number of colposcopies more than tripled with annual cytology starting at age 21, in comparison to annual cytology for ages 21–29 and cotesting at 5-year intervals starting at age 30. The models also agreed that cotesting of women aged 30 and older at 5 years intervals involves fewer colposcopies with similar or slightly lower cancer risk compared with 3-year cytology alone.
Case control studies in Australia and Italy demonstrated that cytologic screening provides only modest protection against adenocarcinoma.86,87 More recently, the International Collaboration of Epidemiological Studies of Cervical Cancer Group pooled screening data from 12 studies involving 1,374 women with adenocarcinoma and concluded that risk reduction of a preceding cytology test was greater for squamous cell carcinoma than for adenocarcinoma.88
In Castellsague et al,.89 HPV was detected in 93% of 167 adenocarcinomas cancers of the cervix (including 55 adenosquamous carcinomas). A case control study with these cases and 1881 controls was also reported. Testing HPV positive (vs. negative) was strongly associated (odds ratio = 81.3) with a diagnosis of cervical adenocarcinoma.89 From Katki et al.74 63% of the adenocarcinomas diagnosed over a 5-year period followed an HPV-positive, cytology-negative cotest.
Women cotesting HPV positive, cytology negative should be followed with either (as noted in the interim ASCCP guidelines:78 (1) repeat cotesting in 12 months (Option 1), or (2) immediate HPV genotype-specific testing for HPV16 alone or for HPV16 and HPV18 (Option 2). If cotesting is repeated at 12 months, women testing positive on either test* should be referred to colposcopy; women testing negative on both tests** should return to routine screening. If immediate HPV genotype-specific testing is used, women testing HPV16 positive or HPV16 and/or HPV18 positive should be referred directly to colposcopy; women testing HPV16 negative or HPV16 and HPV18 negative should be cotested in 12 months, with management of results as described in option 1.
Women cotesting HPV positive, cytology negative should not be referred directly to colposcopy. Furthermore, they should not be tested for individual HPV genotypes other than HPV16 and HPV18. The use of HPV genotype-specific testing for HPV16 or HPV16 and HPV18 is recommended only for the management of HPV-positive, cytology-negative women. Currently, there is insufficient evidence to support the use of non-HPV biomarkers.
There are no RCTs that directly compare different management strategies for women cotesting HPV positive, cytology negative. Consistent, high-quality evidence from prospective observational studies indicates that the short-term risk of CIN3 in this population is far below the risk threshold of HPV positive ASCUS and LSIL used for referral to colposcopy (i.e., 2-year risk of 8–10% in the ASCUS/LSIL Triage Study).20,90 For this reason, immediate colposcopy for all HPV-positive women (including HPV-positive, cytology-negative women) was strongly dismissed by the consensus conference participants as a potential management strategy. Repeat cotesting at 12 months is the current recommended management2 for women testing HPV positive, cytology negative. This is supported by evidence from cohort studies showing the majority of transient infections clear by 12 months,15,91 returning the majority of women to routine screening without excessive risk.
Where available, HPV genotype-specific testing for HPV16 or HPV16/HPV18 may be performed following HPV-positive, cytology-negative results. Large cohort studies17,92 and one industry-sponsored trial93 have shown that HPV16-positive or HPV16- and HPV18-positive results are associated with clinically relevant short-term risk of CIN3 or cancer in HPV-positive, cytology-negative women, supporting immediate referral to colposcopy. In all studies, HPV16 conferred much higher absolute risk than any other carcinogenic type and was most commonly followed by HPV18. Other types such as HPV31 and HPV58 were associated with short-term risks similar to those of HPV18 in some populations.17,94 However, large international case series studies of type attribution to cervical cancers have demonstrated that the etiologic fraction of HPV18 is much higher than that of any other type (except for HPV16),9,10,95,96 and the etiologic fraction is even higher in adenocarcinomas. Thus, including HPV18 in genotype-specific assays appears warranted.
Aside from management of HPV-positive, cytology-negative women, no other clinical indications have sufficient evidence to recommend HPV genotype-specific testing for HPV16 or HPV16 and HPV18. Further studies are likely to further refine the risk estimates of specific test result combinations. There is also a lack of evidence to support the use of other molecular markers in HPV-positive, cytology-negative women. However, studies are ongoing, with results anticipated within 2–3 years.
Table 4(74,93,97–105) shows patient outcomes, number of studies and quality of the evidence. The prevalence of HPV-positive, cytology-negative screening results was reported in nine studies (online supplement, Working Group 3a report Table 1)74,93,97–102,106,107 and ranged from 3.4% to 8.2% in women age 30 years and older. In a screening population of women age 30 years and older,74 the proportion of HPV-positive, cytology-negative results (3.7%) was more than twice that of HPV-positive, cytology-positive results (1.4%), implying a significant increase in referral to colposcopy if HPV-positive, cytology-negative women were referred for immediate colposcopy.
Cumulative risks of CIN2 or CIN3 among HPV-positive, cytology-negative women have been reported from 11 prospective studies with heterogeneous populations, varying disease ascertainment, and length of follow-up ranging from 1–16 years (online supplement, Working Group 3a report Table 2). The estimated 12-month risk of CIN3+ following an HPV-positive, cytology-negative cotest, relevant for management decisions, ranged from 0.8%74 to 4.1% .93 The estimated 12-month risk of cancer was 0.08%.74
For HPV16 or HPV16 and HPV18 genotype-specific testing in HPV-positive, cytology-negative women, risk of CIN3 reaches 10% over 1–4 years for HPV16 positivity and over 2–5 years for HPV18 positivity.17,92 One industry-sponsored trial reported risk of prevalent disease (within 12 weeks) for HPV16- and HPV18-positive results as 11.4% for CIN2+ and 9.8% for CIN3+, and for HPV16-positive results as 13.6% for CIN2+ and 11.7% for CIN3+.93 The short-term risks associated with non-16/18 oncogenic HPV genotypes were considerably lower. 17,92, 93 While these risk estimates for non-16/18 oncogenic HPV genotypes do not warrant immediate colposcopy, they are elevated compared to those for women cotesting negative; therefore, a reasonable approach for women who test HPV16 or HPV16 and HPV18 negative following an initial HPV-positive, cytology-negative result is to follow with repeat cotesting at 12 months to identify women who are likely to have persistent HPV infection and are at an elevated risk of CIN3+ over many years.92,100
The potential utility of p16INK4A immunocytochemistry for managing HPV-positive women has been demonstrated in an Italian screening trial,108 but this study did not directly evaluate women testing HPV positive, cytology negative. As more data from HPV-positive, cytology-negative populations become available for other biomarkers, revisions to recommendations may be warranted.
Women with ASC-US cytology and a negative HPV test result should continue with routine screening as per age-specific guidelines.
The cytologic interpretation of ASC-US represents a category of morphologic uncertainty. The definition of ASC-US is “some, but not all” of the features of a LSIL and as such, the differential diagnosis includes both poorly sampled and poorly represented LSIL and the many morphologic mimics of LSIL. An ASC-US interpretation does not represent a specific diagnosis. Because of its morphologically equivocal nature, the inter- and intra-observer reproducibility of an ASC-US interpretation is less than that for the reliable, unequivocal diagnostic categories of LSIL and high-grade squamous intraepithelial lesion (HSIL).81 The current ASCCP recommendation2 of HPV testing for the management of ASC-US cytology tests allows for the use of this more objective test to stratify the risk for the development of cervical cancer precursor lesions. The introduction of HPV testing for management of ASC-US and cotesting for primary screening over the last decade has led to an increased number of women being identified with HPV-negative, ASC-US-cytology results. The key question to provide rationale for this recommendation is as follows: Does the risk of precancerous lesions in women with HPV-negative, ASC-US-cytology results warrant increased surveillance in comparison to that of women who are negative for intraepithelial lesion or malignancy (NILM) and HPV negative?
Data from published studies have shown that the risk of precancerous lesions following an HPV-negative, ASC-US cytology result is very low, and not different from a negative cotest. Because of the very low cervical cancer risk observed in the ASC-US/HPV-negative population, continued routine screening is recommended for this group: 3-year interval for cytology screening of women ages 21–29 or 30–65 years old, and 5-year interval for cotesting of women ages 30–65 years old.
Women with abnormal cytology more severe than ASC-US (LSIL or more severe) should be referred to colposcopy regardless of their HPV status.2 The risks of CIN3+ and cancer following HPV-negative, LSIL+ cytology results are too great to warrant a return to routine screening.19
Table 4 shows patient outcomes, number of studies and quality of the evidence. The risk of CIN3+ following HPV-negative, ASC-US cytology results is very low. In the largest study, the risk of CIN3 at enrollment in ASC-US/HPV-negative women was 0.28%.103 In a longitudinal follow-up study, the risk of CIN3+ in this population at 5 years was 0.54%.74 Analyses from the ASCUS LSIL Triage Study showed that the 2-year cumulative risk of CIN3+ in ASC-US/HPV-negative was less than 2% (1.4% to 1.9% depending on the HPV testing method used).43,104 For comparison, two studies showed follow-up data on cotest-negative women 30 years and older. In these studies, the risks of CIN3+ ranged from 0.3% (prevalent)93 to 0.16% with 5 years follow-up.74
Overall, the absolute risk of a true precancerous lesion in the HPV-negative, ASC-US cytology population is very low (less than 2% overall, and less than 1% when based on the most robust studies). This level of risk does not warrant more frequent screening.
In most clinical settings, women ages 30–65 years should not be screened with HPV testing alone as an alternative to cotesting at 5-year intervals or cytology alone at 3-year intervals.
Primary HPV testing has been prospectively assessed as a replacement for current acceptable modalities of cervical cancer screening. RCTs of HPV testing alone have demonstrated that when compared to standard cytologic screening, HPV testing has increased sensitivity for detection of CIN3+ and CIN2+ after a single screening round. Greater sensitivity also means greater negative predictive value over a longer time period, because the absence of positive HPV findings is an indication of low risk of developing CIN3+. RCTs have been less successful at defining the specificity of HPV testing, and so the potential harms of primary HPV testing are poorly quantified.
Although primary HPV-based screening approaches appear promising, the lack of a well-defined and evaluated management strategy for positive tests precludes their practical implementation in the majority of clinical settings in the U.S. at this time. There are no data to estimate how the clinical performance of cytology (as a follow-up test) would be affected by a priori knowledge of positive HPV status. The lack of an internal standard for specimen adequacy for some HPV assays may provide false reassurance among a small number of women whose negative screening results may be a function of specimen inadequacy rather than true absence of disease. Such an event is less common with cytology since specimen adequacy assessment is a routine component of the evaluation and inadequacy prompts intervention and follow-up on the part of the clinician and patient. Thus, the inclusion of cytology with HPV testing, i.e. cotesting, provides some additional reassurance against testing errors due to specimen inadequacy, although the benefits in terms of sensitivity and negative predictive values are only incremental. Implications, such as cost effectiveness of and adherence to implementing such a major change in the current US opportunistic screening setting, require further evaluation and planning.
Table 5(38,74,109,110) shows patient outcomes, number of studies and quality of the evidence. HPV testing for primary screening appears promising in women aged 30 years and older, as this group may be at greatest risk for developing CIN3+. In single round screening studies, HPV testing is more sensitive for detection of CIN2+ and CIN3+ than cytology alone and is almost as sensitive as cotesting (2–5% additional CIN3+ are detected among women with HPV-negative, cytology-positive results, primarily those with LSIL or more severe cytology).74,80,105 And a negative HPV test provided greater reassurance against CIN3+ in the subsequent 5–7 years than cytology alone and is nearly as reassuring as a negative cotest. Therefore an acceptable screening interval for use of HPV testing alone should be comparable to that of cotesting.
However, the published studies of HPV primary screening are limited by a lack of long-term follow-up, with only one reporting the second round of screening.41 In that study, referral of all HPV-positive women to colposcopy led to a reduction in cancers in the second round of screening 4 years later compared to cytology screening. But more data are needed regarding the long-term impact of using HPV primary testing for cervical cancer screening.
HPV testing for primary screening is less specific than cytology alone and may identify clinically insignificant disease destined to spontaneously regress.41 Thus, a strategy of immediate colposcopy of all HPV-positive women can be associated with significant harms, due to unnecessary diagnostic procedures or treatment, which may outweigh the benefits of the increased sensitivity. For this reason primary screening using HPV testing alone requires yet-to-be-defined appropriate tests for assessing a positive HPV result.
Currently, there are no published large-scale or population-based studies evaluating management strategies of HPV-positive women in an HPV primary screening setting. A recent systematic review of the available published evidence concluded that HPV testing is very promising for primary screening of women aged 30 years and older, particularly when coupled with cytology testing (for follow-up) of HPV positive results, which may reduce the increase in false positives (and their related harms) that would result from HPV testing alone.111 There are no direct data to estimate the performance of cytology in a triage (follow-up) setting, although a simulation analysis using data from a RCT found an improved positive predictive value using HPV testing followed by cytology compared to other combinations of cytology and/or HPV testing.38 Specifically, it is unclear whether the interpretation of cytology in a real-world setting is affected by a priori knowledge of HPV positivity and what impact this may have in a general population screening setting. In addition, as discussed in the Management of Women with HPV Positive, Cytology Negative Cotests section, rational clinical follow-up of HPV-positive, cytology-negative women is critical to realizing the (sensitivity) benefits of using HPV testing (although the HPV-negative patients would still benefit from the added safety). Assessment of the full impact of a primary screening strategy using HPV with or without cytology follow-up may be possible only after implementation in selected clinical settings in a Western or high-resource setting and/or using modeling analyses.111
Other strategies have aimed to improve specificity and reduce harm by interposing secondary testing for management decisions between a positive HPV test and colposcopy. Potential secondary biomarkers included HPV genotyping (for HPV16 or HPV16/18),92,100 HPV mRNA testing,112 and/or detection of other non-HPV biomarkers (e.g. p16INK4A).108 Although promising, there are limited data regarding the test performance of these markers. Specifically, the cross-sectional and archival nature of most available molecular marker studies as well as the heterogeneity of clinical endpoints examined (CIN2+ vs. CIN3+) limits the current usefulness of these data. Finally, there are no direct comparisons of these various triage strategies and the specificity of such an approach, and the consequential potential harms (or benefits) have not yet been well defined.
Women over 65 years of age with evidence of adequate negative prior screening and no history of CIN2+ within the last 20 years should not be screened for cervical cancer with any modality. Once screening is discontinued it should not resume for any reason, even if a woman reports having a new sexual partner.*
Following spontaneous regression or appropriate management of CIN2, CIN3, or adenocarcinoma in situ (AIS), routine screening should continue for at least 20 years (even if this extends screening past age 65).
In well-screened women older than the age of 65 in the United States, CIN2+ prevalence is low29,113 and cervical cancer is rare.1 In the US, cervical cancer is most commonly diagnosed in unscreened and under-screened women.114–116 In contrast, the potential harms from screening of this population, in addition to those already specified, include discomfort during cytology sampling and false-positive screening tests. Based on the extended natural history of the disease, it is improbable that incident HPV infections and newly detected CIN3 after the age of 65 will have sufficient time to progress to invasive cancer in the woman’s lifetime but it is unlikely that there will ever be a clinical trial to demonstrate this directly. Finally, one modeling study concluded that for women who have been screened every 3 years prior to age 65 years, the ratio of colposcopies to years of life gained associated with further screening was large (or the years of life gained per colposcopy small) because of the small gains in life expectancy.63
The age at which to discontinue screening is based on the opinions of the expert panel members and was chosen to balance the benefits and harms of screening older women. Women who discontinue cervical cancer screening should continue to obtain age-appropriate health care.
While women with adequate negative prior screening have a very low risk of cervical cancer, those who have been treated for CIN2+ in the past 20 years (or had it resolve spontaneously) remain at approximately a 5- to 10-fold higher risk for cervical cancer than the general population.117,118 (We note that these studies were based on cytology alone; future studies incorporating HPV testing may yield different risks.) We endorse the ASCCP guidelines for continued regular screening of these women for 20 years after an initial period of more intense surveillance, even if that extends screening past age 65. We define regular screening as screening every five years using cotesting (preferred) or every three years using cytology alone (acceptable).
Recent evidence suggests that the natural history of incident HPV infections is unaffected by a woman’s age at acquisition.94,119 A new carcinogenic HPV infection in a woman aged 65 years or older with a cervix should clear spontaneously in most cases, and only a small percentage of women should have a persistent infection. Since the transformation zone of older women is smaller and less accessible than in younger women, and since cervical cancer develops many years after an incident infection, screening this population would detect a very small number of new cases of CIN2+ and prevent very few cervical cancers and even fewer cancer deaths.
Table 6 (63) shows patient outcomes, number of studies and quality of the evidence. Mathematical modeling63 among women screened with cytology every three years prior to age 65 demonstrates that continued screening even to age 90 prevents only 1.6 cancer cases and 0.5 cancer deaths per 1000 women. For every 1000 women, continued screening extends life expectancy by only one year, while resulting in 58 extra false positives, 127 extra colposcopies and 13 extra CIN2/3 diagnoses requiring treatment.
With respect to newly acquired HPV infection in women who have discontinued screening, indirect evidence regarding the risk of not resuming screening in this population is found in the report by Chen et al. In a large-scale community-based cohort of women followed for up to 16 years after receiving cytology and HPV testing at baseline and two years later, newly detected infections were associated with very low absolute risks of persistence and CIN3+ regardless of the woman’s age. Furthermore, in women aged 55 and older after two negative HPV tests two years apart, the risk of subsequently developing CIN3 or cervical cancer was only 0.08%, with only one woman developing CIN3 after 9.6 years.94 In another large 7-year, population-based cohort study, newly detected infections were associated with very low absolute risks of persistence or progression. The rate of progression to CIN2+ (or CIN3+) after 3 years of follow-up was not higher for women aged 34 years and older than for younger women.119 Therefore, most new carcinogenic HPV infections in women age 65 years or older should clear spontaneously, and only a small percentage is likely to persist. Since the transformation zone of older women is smaller and less accessible than in younger women and since cervical cancer develops at a median 15–25 years after an incident infection, screening this population would detect a very small number of new cases of CIN2+ and prevent very few cancers and even fewer cancer deaths. The risks associated with over-treatment in the elderly population outweigh the benefits.
Women at any age following a hysterectomy with removal of the cervix who have no history of CIN2+ should not be screened for vaginal cancer using any modality. Evidence of adequate negative prior screening is not required. Once screening is discontinued it should not resume for any reason, including if a woman reports having a new sexual partner.
In women who have undergone hysterectomy with removal of the cervix for reasons other than CIN2+, vaginal cytology screens for primary vaginal cancer. Vaginal cancer is an uncommon gynecologic malignancy. Its age-specific incidence is similar to or less than that of other cancers for which screening is not performed, such as breast cancer in men. Abnormal vaginal cytology is rarely of clinical importance. Therefore, there is no justification for continuing to screen these women for lower genital tract malignancies. Women who have had a hysterectomy for cervical intraepithelial lesions may be at increased risk of vaginal cancer, but the data are limited. Women who discontinue screening should continue to obtain age-appropriate preventive health care.
The incidence rates for all vaginal cancers combined were 0.18 per 100,000 female population for in situ cases and 0.69 for invasive cases.120 A retrospective cohort study of vaginal cuff cytology in 5,862 women post-hysterectomy for benign disease reported abnormal cytology among 79 women (1.1% of all smears). The mean length of time from hysterectomy to abnormal cytology result was 19 years. The positive predictive value of vaginal cuff cytology for detection of vaginal cancer was 0 (95% CI 0 to 33%).121 A 10-year retrospective study among 697 women after hysterectomy for benign disease found that 663 vaginal cuff cytology tests were needed to detect one case of vaginal dysplasia.122 A retrospective study of 220 women selected at random from 2,066 women with a previous hysterectomy for benign conditions and followed for an average of 89 months identified seven patients (3%) with intraepithelial cytologic abnormalities, but no vaginal cancers. No benefit in patient outcomes was observed.123 A cross-sectional study of 5,330 screening cytology tests in women after a hysterectomy found one case of dysplasia and no cancers.124
In a study of 193 women with CIN at hysterectomy, the incidence of abnormal vaginal cuff cytology at least two years after hysterectomy was 0.7 per 1,000; at 20 years, 96.5 percent of the women continued to have normal cytology.125 Thus, even if women with hysterectomy were at an increased risk of vaginal cancer, there is no proven method to effectively intervene before vaginal cancer develops.
Recommended screening practices should not change on the basis of HPV vaccination status.
Two HPV vaccines have been licensed in the US; both are highly effective at preventing infection with the two most carcinogenic HPV types, HPV16 and HPV18, which cause about 70% of all cervical cancers. Randomized clinical trials have also shown that HPV16/18 vaccination is highly effective in preventing CIN2 and CIN3 among women not previously exposed to these types. In vaccinated populations, it is plausible that women protected by vaccination could have less intensive screening and also start screening at a later age, since they will likely experience a lower risk of cervical cancer in the future. However, a number of arguments preclude a more permissive screening policy at this time among a vaccinated cohort in the US. About 30% of cervical cancers will continue to occur, because the first generation of vaccines covers only HPV16 and HPV18. As Advisory Committee on Immunization Practices (ACIP) recommendations include vaccinating women up to 26 years, many women may be vaccinated after HPV infection has already occurred, when efficacy declines. Moreover, coverage of HPV vaccination in the US has yet to reach levels comparable to those of countries, like Australia and the United Kingdom, which have publicly-funded, school-based vaccination programs that guarantee high coverage of pre-adolescents and young women. On average for all states in the U.S. in 2010, only 32% of eligible girls and women had received all 3 doses of the vaccines, and HPV vaccination is largely opportunistic, not necessarily targeting girls and young women before the onset of sexual activity.126 There are also geographic and socioeconomic disparities in vaccination coverage.
Thus, there are no data at this time that support changes in the age when screening is to be initiated or in the screening interval for U.S. women that have been vaccinated. The same recommendation applies to the individual woman who reports having been vaccinated. Overall practice recommendations for age at initiation of screening, interval, and acceptable screening technologies are described elsewhere in this article and should be followed in populations with access to HPV vaccination as well as for individual women with known vaccination.
Table 7(126–151) shows patient outcomes, number of studies and quality of the evidence. The low HPV vaccine coverage in the U.S. remains a barrier for proposing population-based changes in cervical cancer screening among those receiving the vaccine. Moreover, the coverage threshold at which changes would be cost-effective for the population and yet safe for the individual remains unknown. At low coverage, herd immunity will not occur and there will be little impact on HPV transmission rates and consequently on the incidence of CIN3+. Even in countries with high HPV vaccination coverage, changes to cervical screening practices are not immediately anticipated. Surveillance systems, such as the New Mexico HPV and Pap Registry,152 the establishment of sentinel CIN3 registries, and HPV Vaccine Impact Monitoring Project Across Connecticut (HPV-IMPACT), a sentinel surveillance system for monitoring HPV vaccine impact on HPV type-specific CIN2+ established by the Centers for Disease Control,153 will be critical for monitoring the impact of vaccination.
Although mathematical models suggest that current cervical cancer screening recommendations could be modified following HPV vaccination, there is no consensus on screening recommendations in a vaccinated cohort. There is agreement among modelers that it will take more than a decade to see the full impact of vaccination on screening outcomes. As a result, changes in screening recommendations in the HPV vaccination era, although attractive, will likely not be warranted in the immediate future. It is also important to ensure that the benefits of vaccination are not offset by reductions in screening coverage due to complacency or an erroneous belief that vaccination eliminates the need for screening. One study suggested that with vaccine coverage of 84% among12-year-old girls, a reduction of screening from about 80% coverage to about 60% coverage could lead to reductions in life expectancy compared to no HPV vaccination with 80% screening coverage.147 Likewise, another study found that the quality-adjusted life expectancy was lower for certain age groups under conditions in which there was lower participation in screening (<70%) and incomplete vaccine coverage (<75%) compared to current screening practice without vaccination.148
Another important argument against modifying screening recommendations based on introduction of HPV vaccines is the lack of empirical data on the performance of screening tests among a vaccinated cohort. Mathematical models149,150 indicate that vaccination is expected to reduce the prevalence of high-grade cervical lesions over time, which will have a deleterious influence on the positive predictive value of screening tests, thus increasing the proportion of false positive results. Although not empirically based, these models provide insights concerning the role of cytology or other screening technologies and raise awareness about the need to reassess future screening practices to guarantee acceptable performance quality and safety. The guidelines presented elsewhere in this report emphasize the need for less intensive screening, setting the stage for reducing the harms that would come from the expected loss in screening test performance.
Despite the rationale for changes in screening practices among vaccinated women, agreeing on a recommendation would have to be based on high quality evidence on the critical outcomes, including duration of protection and reduction of risk of CIN3+. A key question is the duration of protection following HPV vaccination, especially in girls ages 11–12, and the impact on age-specific cancer risks. In addition, reliable documentation of fully vaccinated status at an age likely to be prior to HPV exposure would be needed. Evidence is also needed on (1) the effect of vaccination on the HPV genotype distribution, (2) the impact of vaccination on the performance of cytology and HPV testing (the two methods recommended in the updated guidelines), and (3) the effect of vaccination on screening adherence. It is expected that epidemiologic surveillance via linkage of vaccination registries with screening and HPV testing databases or electronic medical records from managed care organizations, and collection and reporting of screening data by vaccination status will permit comparisons of HPV type distributions, screening behaviors, and lesion prevalence between vaccinated and unvaccinated individuals. Having such data sources would permit tailoring of screening recommendations for women with a documented history of HPV vaccination. In addition to registries, clinic-based systems and large screening programs such as Title X Family Planning, the CDC’s National Center of Chronic Disease Prevention and Health Promotion, Planned Parenthood, and managed care organizations should begin reporting screening data by HPV vaccination status. Such guidance is already being provided by the Public Health Agency of Canada.154
These updated guideline recommendations were motivated by an increased understanding of the natural history of HPV infection and cervical carcinogenesis, and by an expanding knowledge of the relative performance of different screening tests. Evident and important remaining research priorities include the following:
The process used to develop these recommendations represents a transitional stage in guidelines development for the American Cancer Society. Previous guidelines have been developed using a consensus process involving experts in the field along with key stakeholders; although these recommendations were based on evidence, there was not a formalized process for evaluating the evidence and incorporating it into the recommendations. The group developing these guidelines also consisted of experts and stakeholders; the key difference was in the use of the principles of the GRADE guideline development process to more formally evaluate the evidence and incorporate the quality of the evidence into the recommendations. Beginning in 2012, the ACS will be using a new guidelines process, involving a standing group of non-specialists and a formal process for evidence review;157 this change comes in response to Institute of Medicine recommendations for improving guideline transparency, clarity, and reducing potential conflict of interest.
One of the key principles outlined in the IOM report was the need for timely updating of guidelines as new evidence becomes available. Particularly for areas where uncertainty remains, there are large ongoing trials that either were published after our final update for the conference,158 or will publish results in the next 1–2 years. One advantage of using a structured evidence evaluation process such as GRADE is that it facilitates identifying the key research needs that will lead to changes in recommendations; this should help expedite the updating process.
Until the next update, these recommendations reflect the participants’ judgment of the best evidence-based practice for the prevention of cervical cancer morbidity and mortality through currently available screening tests that maximizes protection against cervical cancer while minimizing the potential harms associated with false-positive results and overtreatment.
We would like to thank the following members of the Working Group on the Impact of HPV vaccination on future screening practices: Dr. Mona Saraiya for serving as Co-chair of this working group, Dr. Harrell Chesson for his important modeling contributions, and Dr. Donatus Ekwueme for serving on this group. We thank Dr. Tom Wright and Dr. Ronald Luff for their invaluable contribution to the initial planning of the Guidelines effort; Kim Andrews for her contribution to the initial literature search; Lisa Oliver for her assistance in preparing the manuscript and online supplement; and Debbie McClain for her work with the web-based bulletin board and onsite audience voting system.
Working Group 1: Maureen Killackey, MD (Co-Chair), Memorial Sloan Kettering Cancer Center, New York, NY; Shalini L. Kulasingam, PhD (Co-Chair), University of Minnesota, Minneapolis, MN; Patricia Fontaine, MD, MS, HealthPartners Research Foundation, Bloomington, MN; Richard S. Guido, MD, Magee-Womens Hospital of the UPMC Health System, University of Pittsburgh, Pittsburgh, PA; Abbe Herzig, PhD, Health Ratings Center, Consumer Reports, Yonkers, NY; Herschel W. Lawson, MD, Emory University School of Medicine, Atlanta, GA; Dina R. Mody, MD, The Methodist Hospital, Houston, TX; Jeffrey Waldman, MD, Planned Parenthood, University of California, San Francisco, CA; Mark H. Stoler, MD (liaison), University of Virginia Health System, Charlottesville, VA
Working Group 2: Joanna M. Cain, MD, FACOG (Co-Chair), University of Massachusetts School of Medicine, Worcester, MA; Walter Kinney, MD (Co-Chair, liaison), University of California, The Permanente Medical Group, Sacramento, CA; George Birdsong, MD, Emory University School of Medicine, Grady Health System, Atlanta, GA; Wendy R. Brewster, MD, PhD, University of North Carolina, Chapel Hill, Lineberger Comprehensive Cancer Center, Chapel Hill, NC; David Chelmow, MD, Virginia Commonwealth University School of Medicine, Richmond, VA; Valerie J. King, MD, MPH, Oregon Health & Science University, Portland, OR; Robert G. Pretorius, MD, Southern California Permanente Medical Group, Fontana, CA; Cosette M. Wheeler, PhD, University of New Mexico Health Sciences Center, Albuquerque, NM; Barbara A. Winkler, MD, Mount Kisco Medical Group, Mount Kisco, NY
Working Group 3a: Alan G. Waxman, MD, MPH (Chair), University of New Mexico School of Medicine, Albuquerque, NM; Jane J. Kim, PhD, Harvard School of Public Health, Boston, MA; Nicolas Wentzensen, MD, PhD, MS, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, MD; Philip E. Castle, PhD, MPH (liaison), American Society for Clinical Pathology, Washington, DC
Working Group 3b: David C. Wilbur, MD (Chair), Massachusetts General Hospital, Harvard Medical School, Boston, MA; J. Thomas Cox, MD, University of California, Santa Barbara, CA; Isam A. Eltoum, MD, MBA, University of Alabama at Birmingham, Birmingham, AL; Philip E. Castle, PhD, MPH (liaison), American Society for Clinical Pathology, Washington, DC
Working Group 4: Levi S. Downs, Jr, MD (Co-Chair), Masonic Cancer Center, University of Minnesota Medical School, Minneapolis, MN; Mark Spitzer, MD (Co-Chair), Weill Medical College of Cornell University, Brookdale University Hospital and Medical Center, Brooklyn, NY; Teresa M. Darragh, MD, University of California, San Francisco, CA; Shirley E. Greening, MS, JD, CFIAC, Jefferson School of Health Professions, Philadelphia, PA; Hope K. Haefner, MD, The University of Michigan Hospitals, Ann Arbor, MI; E.J. Mayeaux, Jr., M.D., DABFM, FAAFP, Louisiana State University Health Sciences Center, Shreveport, LA; Laurie Zephyrin, MD, MPH, MBA, New York-Presbyterian/Columbia, New York, NY; Debbie Saslow, PhD (liaison), American Cancer Society, Atlanta, GA
Working Group 5: Anna-Barbara Moscicki, MD (Co-Chair), University of California, San Francisco, CA; Kevin A. Ault, MD, Emory University, Atlanta, GA; Myriam Chevarie-Davis, MD, McGill University, Montreal, Canada; Eduardo L. Franco, DrPH, McGill University, Montreal, Canada; Michael A. Gold, MD, Vanderbilt University Medical Center, Nashville, TN; Warner K. Huh, MD; University of Alabama, Birmingham, AL; Diane Solomon, MD (liaison), Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Rockville, MD
Working Group 6: Francisco A. R. Garcia, MD, MPH (Co-Chair); Center of Excellence in Women’s Health, University of Arizona, Tucson, AZ; Ann T. Moriarty, MD (Co-Chair), Ameripath, Indianapolis, IN; Terence J. Colgan, MD, Mount Sinai Hospital, Toronto, Ontario, Canada; Mark H. Einstein, MD, MS, Montefiore Medical Center and Albert Einstein College of Medicine, Bronx, NY; Michael R. Henry, MD, Mayo Medical Laboratories, Rochester, MN; L. Stewart Massad, MD, Washington University School of Medicine, St. Louis, MO; Kate Simon, PhD, Microbiology Division, OIVD, CDRH, Food and Drug Administration, Silver Spring, MD; Patti Gravitt, PhD, MS (liaison), Johns Hopkins Bloomberg School of Public Health, Baltimore, MD
Data Group: Herschel W. Lawson, MD (Co-Chair), Emory University School of Medicine, Atlanta, GA; Debbie Saslow, PhD (Co-Chair), American Cancer Society, Atlanta, GA; Philip E. Castle, PhD, MPH, American Society for Clinical Pathology, Washington, DC; Jack Cuzick, PhD, Wolfson Institute of Preventive Medicine, Queen Mary University of London, London; Patti Gravitt, PhD, MS, Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; Walter Kinney, MD, University of California, The Permanente Medical Group, Sacramento, CA; Evan R. Myers, MD, MPH, Duke University Medical Center, Durham, NC; Kathleen G. Poole, MALS, American Society for Colposcopy and Cervical Pathology, Hagerstown, MD; Mark Schiffman, MD, Clinical Genetics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD; Diane Solomon, MD, Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Rockville, MD; Mark H. Stoler, MD, University of Virginia Health System, Charlottesville, VA
Steering Committee: Diane Solomon, MD (Co-Chair), Division of Cancer Prevention, National Cancer Institute, National Institutes of Health, Rockville, MD; Debbie Saslow, PhD (Co-Chair), American Cancer Society, Atlanta, GA; Philip E. Castle, PhD, MPH, American Society for Clinical Pathology, Washington, DC; Carmel J. Cohen, MD, Mount Sinai School of Medicine; Ruttenberg Cancer Center, New York, NY; Mitchell I. Edelson, MD, Hanjani Institute for Gynecologic Oncology, Abington Memorial Hospital, Abington, PA; Francisco A. R. Garcia, MD, MPH, Center of Excellence in Women’s Health, University of Arizona, Tucson, AZ; E. Blair Holladay, PhD, SCT(ASCP)CM, American Society for Clinical Pathology, Chicago, IL; Walter Kinney, MD, University of California, The Permanente Medical Group, Sacramento, CA; Herschel W. Lawson, MD, Emory University School of Medicine, Atlanta, GA; Kenneth L. Noller, MD, MS, American Board of Obstetrics and Gynecology, Dallas, TX; Edward E. Partridge, MD, University of Alabama at Birmingham Comprehensive Cancer Center, Birmingham, AL; Kathleen G. Poole, MALS, American Society for Colposcopy and Cervical Pathology, Hagerstown, MD; Carolyn D. Runowicz, MD, Florida International University, Herbert Wertheim College of Medicine, Miami, FL, Robert A. Smith, PhD, American Cancer Society, Atlanta, GA; Alan G. Waxman, MD, MPH, University of New Mexico School of Medicine, Albuquerque, NM
Writing Committee: Philip E. Castle, PhD, MPH (Co-Chair), American Society for Clinical Pathology, Washington, DC; Evan R. Myers, MD, MPH (Co-Chair), Duke University Medical Center, Durham, NC; Debbie Saslow, PhD (Co-Chair), American Cancer Society, Atlanta, GA; David Chelmow, MD, Virginia Commonwealth University School of Medicine, Richmond, VA; Eduardo L. Franco, DrPH, McGill University, Montreal, Canada; Francisco A. R. Garcia, MD, MPH, Center of Excellence in Women’s Health, University of Arizona, Tucson, AZ; Abbe Herzig, PhD, Health Ratings Center, Consumer Reports, Yonkers, NY; Jane J. Kim, PhD, Harvard School of Public Health, Boston, MA; Walter Kinney, University of California, The Permanente Medical Group, Sacramento, CA; Herschel W. Lawson, MD, Emory University School of Medicine, Atlanta, GA; Mark Schiffman, MD, Clinical Genetics Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD; Mark Spitzer, MD, Weill Medical College of Cornell University, Brookdale University Hospital and Medical Center, Brooklyn, NY; Jeffrey Waldman, MD, Planned Parenthood, University of California, San Francisco, CA; Nicolas Wentzensen, MD, PhD, MS, Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Rockville, MD; David C. Wilbur, MD, Massachusetts General Hospital, Harvard Medical School, Boston, MA
*HPV refers only to high-risk HPV. Other HPV types are unrelated to cervical cancer and should not be used in cervical cancer screening. Testing for low-risk HPV types has no clinical role in cervical cancer screening or evaluation of women with abnormal cytology.
*The majority of recommendations are “strong.” The strength of each recommendation is noted in the individual working group reports in the online supplement.
*The majority of recommendations are “strong.” The strength of each recommendation is noted in the individual working group reports in the online supplement.
*HPV positive OR LSIL or more severe cytology
**HPV negative AND ASCUS or negative cytology
*Adequate negative prior screening is defined as 3 consecutive negative cytology results or 2 consecutive negative cotests within the 10 years before ceasing screening, with the most recent test occurring within the past 5 years.
Disclaimers: The contents of the paper are solely the responsibility of the authors and do not necessarily represent the official views of the Centers for Disease Control and Prevention, National Institutes of Health, the U.S. Government, or any medical or academic institutions.
The following reported no financial relationships or potential conflicts of interest to disclose: D Saslow, E Partridge, B Holladay, W Kinney, H Lawson, K Noller, E Myers, A Waxman, K Poole, R Smith, P Fontaine, A Herzig, M Killackey, S Kulasingam, D McCoy, W Brewster, J Cain, D Chelmow, V King, R Pretorius, B Winkler, I Eltoum, J Kim, N Wentzensen, L Downs, S Greening, H Haefner, L Zephyrin, M Chevarie-Davis, D Ekwueme, T Colgan, M Henry, S Massad, K Simon.
P Castle receives payment for service on the Data Monitoring and Safety Board for Merck Sharp & Dohme. C Cohen serves as a speaker for Merck, Inc and receives honoraria. M Edelson’s spouse is employed and receives salary from Merck, Inc. F Garcia is employed by the University of Arizona, which holds contracts for the performance of research with Roche Pharmaceutical/Roche Molecular, Hologic, Third Wave, MTM, Qiagen, Becton Dickson, and Medispectra/Luma. Dr. Garcia also serves on the speakers’ list for Qiagen and receives honoraria. J Cuzick serves on advisory boards and as an ad-hoc consultant for Qiagen, Roche, Gen-Probe, BD, and Abbott, with research funds provided to his institution from Qiagen, Roche, Gen-Probe, BD, and Abbott. P Gravitt provides service on the scientific advisory board from Qiagen and for which she received honoraria. E Myers received research support for investigations for Gen-Probe, Inc. and from GSK, Inc. He served as a speaker for and received honoraria from Gen-Probe, Inc., and has served as a consultant for Merck, Inc., for which he received an honorarium. M Schiffman holds a research agreement to serve as a Medical Monitor in the NCI vaccine trial through GSK; Dr. Schiffman also receives research support from Qiagen for CareHPV research in Nigeria. D Solomon serves as a Medical Monitor for the National Cancer Institute’s HPV Vaccine Trial in Costa Rica: the trial receives vaccine from GlaxoSmithKline. M Stoler received fees for serving as a consultant to Merck Research Labs, Roche, Ventana Medical Systems, BD, Hologic, MTM, and Gen-Probe. D Mody conducted lectures/workshops for CAP, ASCP and ASC for which she received honoraria and/or travel expenses. G Birdsong’s employer receives funding for contracted research performed by Dr. Birdsong for BD Diagnostics. C Wheeler is an employee of University of New Mexico, which is contracted by GlaxoSmith Kline, Inc. for its vaccine trials and receives equipment/reagents from Roche Molecular Systems for HPV genotyping. D Wilbur serves on the scientific advisory board for Corista, LLC. T Darragh received Thin Prep supplies for research from Hologic. She serves on an advisory board from OncoHealth and has received stock options as payment. E Mayeaux serves on the speakers’ advisory board for both Merck, Inc., and Pharmaderm and he receives honoraria from both companies for his service. M Spitzer serves as a speaker for both Merck, Inc and Qiagen and receives honoraria. K Ault received clinical research grants from NIAID, GenProbe, Merck, Inc., and Roche and served as a site principal investigator for the research. All grants were provided to his employer, Emory University. E Franco received honorarium as a Study Steering Committee member for GlaxoSmithKline; Dr. Franco serves on the advisory boards of Merck, Inc., Roche, and Gen-Probe and from which he receives honoraria. M Gold received honorarium for serving as a speaker and consultant for Hologic. W Huh serves as a consultant to Roche, Qiagen, Merck, Inc., and Inovio and receives honoraria from all four companies. A-B Moscicki received honorarium for serving as a consultant to an advisory board for Merck, Inc. M Einstein has advised or participated in educational speaking activities, but does not receive an honorarium from any companies. His employer, Montefiore Medical Center, has received payment for his time spent on activities for Merck, Inc., GlaxoSmithKline, Roche, Bristol-Myers Squibb, Hologic, Advaxis, Aura Biosciences, Inovio, Photocure, Neodiagnostix, and PDS Biotechnologies. A Moriarty received honorarium as a speaker for the American Society of Cytopathology.