We conducted this targeted systematic evidence review of five key questions to assist the U.S. Preventive Services Task Force (USPSTF) in updating its 2003 recommendation on cervical cancer screening.

We conducted literature searches of the Database of Abstracts of Reviews of Effects, the Cochrane Database of Systematic Reviews, PubMed, the Health Technology Assessment database, MEDLINE, the Cochrane Collaboration Registry of Clinical Trials, and PsycINFO from January 2000 through September 2010. We also wrote trial authors for unpublished data and searched for updated publications from trials of human papillomavirus (HPV) screening.

We reviewed a total of 4,262 abstracts and 641 complete articles. We included 35 studies reported in 66 articles (only one of which was published at the time of the previous USPSTF review): five related to initiating cervical cancer screening, four comparing liquid-based and conventional cytology, 12 evaluating HPV for primary cervical cancer screening, four evaluating the use of HPV plus cytology screening, one evaluating cytology triage of primary HPV testing, six evaluating HPV for triage of abnormal cytology to colposcopy, and four evaluating the harms of HPV testing.

Two investigators independently reviewed all abstracts against a set of a priori inclusion criteria for all key questions. One investigator abstracted data from included studies into evidence tables and a second reviewer checked these data. At least two investigators critically appraised each study using design-specific quality criteria from the USPSTF, supplemented by the National Institute of Health and Clinical Excellence criteria for randomized controlled trials and systematic reviews and the QUADAS tool for quality assessment of diagnostic accuracy studies. Per the USPSTF methods, studies rated as poor quality were excluded.

Our results focus on trials and studies conducted in countries that have well-developed approaches to cervical cancer screening and are summarized primarily using qualitative synthesis due to incomplete reporting and clinical heterogeneity among included studies.

Key Question 1: Initiation of cervical cancer screening. The incidence of invasive cervical cancer (ICC) peaks among U.S. women aged 40 to 44 years, and few cases of cervical cancer are detected in women younger than age 20 (age-adjusted incidence rate of squamous cell carcinoma, 0.05 cases per 100,000 U.S. women). In contrast, HPV infection is most prevalent among women younger than age 20 years, occurring in about 20 percent of women, and is primarily transient in nature (median duration, 13.7 months), as are cytologic abnormalities (median duration, 8.7 months). Women younger than age 25 years have a higher proportion of false-positive Pap smears (age 15 to 19, 3.1%; age 20 to 24, 3.5%) than women aged 25 to 39 years (age 25 to 29, 2.1%; age 30 to 39, 2.6%). A large case-control study in the United Kingdom including 4,012 women with invasive cancer and 7,889 controls found that cervical cancer screening among women younger than age 25 was not associated with a decreased incidence of cervical cancer diagnosis prior to the age of 30, although an impact on stage IB+ cervical cancer in women aged 25 to 27 years could not be ruled out. An overall protective effect of screening was not demonstrated until age 32 years, at which time screening was associated with a 45 percent reduction in the incidence of ICC diagnosis between the ages of 35 and 39 years (odds ratio, 0.55 [95% CI, 0.44 to 0.69]).

Key Questions 2 and 4: Liquid-based cytology compared to conventional cytology. Liquid-based cytology (LBC) and conventional cytology (CC) did not differ significantly in measures of relative sensitivity or absolute sensitivity or specificity for detection of cervical intraepithelial neoplasia (CIN)2+ or CIN3+ at any cytologic threshold. In two large randomized trials (n=134,162; age 25 to 64 years), LBC yielded a lower proportion of unsatisfactory slides than conventional cytology (0.4% and 2.6% of LBC slides vs. 1.1% and 4.1% of CC slides).

Key Question 3: HPV primary screening alone or followed by cytology triage.

Women aged 35 years and older. In a large fair-quality Italian randomized controlled trial (RCT) (NTCC Phase II) testing Hybrid Capture 2 (HC2) high-risk HPV screening against CC in 35,471 women aged 35 to 60 years, about twice as many CIN3+ or CIN2+ cases were detected in the HPV arm relative to CC after a single round, with relatively decreased CIN3+ in the second screening round (RR, 0.23 [95% CI, 0.07 to 0.82]). Cumulative relative CIN3+ detection was increased after a second screening round (which included cytology only) and 3.5 median years of followup (RR, 1.57 [95% CI, 1.03 to 2.40]), with about the same number of invasive cancer cases detected in both arms. Since women with a positive HPV test or atypical squamous cells of undetermined significance (ASC-US) cytology were immediately referred for colposcopy, baseline colposcopies were much higher in the HPV arm (5.8%), compared with cytology (2.5%). Trial investigators pooled invasive cancer from these primary HC2 results (NTCC Phase II) with HC2-CC co-testing results (NTCC Phase I) due to insignificant statistical heterogeneity between trials. Pooled results suggested decreased invasive cancer in women aged 35 years and older who were screened with HPV (6 total ICC cases in the HPV screening arms compared to 15 in the CC only arms; p=0.052). However, cancer outcomes would ideally come from comparable screening strategies and reflect clearly similar opportunities for diagnosis through comparable delivery of colposcopies and/or long enough followup with registry linkages to allow disease ascertainment outside the screening program. Reported data on cumulative burden or relative harms were lacking, since neither cumulative colposcopies nor cumulative relative positive predictive value (PPV) over the screening rounds were reported, nor compared between HPV and cytology screening. In absolute test performance studies, HC2 was much more sensitive (about 40% or higher relative sensitivity), but less specific (3 to 5% relatively less specific) than CC for CIN2+ or CIN3+ at a threshold of ASC-US or low-grade squamous intraepithelial lesion (LSIL) in women aged 30 years and older.

A very large fair-quality trial in 59,757 Finnish women aged 35 to 65 years compared primary HC2 screening (followed by CC triage for positive HPV tests) to CC screening alone at a colposcopy referral threshold of LSIL+. HPV with cytology triage tended to identify about one-third more CIN2+ or CIN3+ cases than CC alone after a single screening round (and at least 2 years of followup). However, extended followup (mean, 3.3 years) after this first screening round with linkage to registry data was required to demonstrate a significant increase in CIN3+ (RR, 1.77 [95% CI, 1.16 to 2.74], including 11 ICC/adenocarcinoma in situ (ACIS) cases in HPV arm and 6 ICC/ACIS cases in CC only arm). In terms of colposcopy, cytology-triaged HPV screening and cytology screening alone resulted in about the same number of immediate referrals (about 1%), with slightly more women identified for retesting (and possible colposcopy referral in the future) in the HPV-cytology triage arm (7.2%), compared with CC (6.6%). Data for total colposcopies and compliance with colposcopy and retesting referrals for the entire first screening round are not yet reported, but will be important, since about half of CIN3+ cases found during extended followup came from those recommended for retesting. A second screening round at 3 years is planned. As more data from this trial are reported, differences with U.S. practice, including cytology referral and CIN treatment thresholds, will also need to be considered.

Women younger than age 35 years. In the fair-quality Italian NTCC Phase II trial in 13,725 women aged 25 to 34 years, HC2 screening detected about four times the amount of CIN2+ and CIN3+ cases as CC after a single round, with relatively decreased CIN3+ in the second screening round (0.20 [95% CI, 0.05 to 0.93]). Cumulative detection of both CIN2+ and CIN3 was at least doubled in the HPV arm relative to CC (after a second round of CC screening only in both arms), with almost no invasive cancer cases in either arm. Pooled results for invasive cancer across the NTCC Phase I and II trials in younger women were not considered due to significant heterogeneity in age-specific protocols and statistical tests of between-trial results in younger women. Only baseline colposcopy referrals were reported, and these were markedly increased in the HPV primary screening arm (13.1%), compared with CC (3.6%). In the single study reporting absolute test performance for HPV alone in women younger than 30, sensitivity was relatively increased for CIN2+ or CIN3+ (23 to 27% higher than cytology at ASC-US+ threshold), while specificity was decreased to a much greater degree (11% relatively lower than cytology) than in older women.

Among 11,580 women aged 25 to 34 years old in the Finnish trial, HC2 with CC triage was little different from cytology in either CIN3+ detection or immediate colposcopy (2.8 vs. 2.7%), despite a higher percentage (16.7%) of HPV positive results initially. Complete colposcopy referrals for the entire first screening round will likely be greater in the HPV-cytology triage arm, since 15.8 percent of younger women—about twice the percentage in the cytology arm—were targeted for repeat testing. A second screening round at 3 years is planned.

Key Question 3: Combination HPV and cytology screening (co-testing).

Women aged 30 or 35 years and older. Four large fair-quality RCTs (NTCC Phase I, POBASCAM, Swedescreen, ARTISTIC) compared combined HPV-cytology (co-testing) to cytology screening alone in 82,390 European women aged 30 to 64 years. Cumulative relative CIN3+ detection was the same between HPV-cytology co-testing and cytology alone after two screening rounds in all the RCTs, and most co-testing trials report differences in round-specific relative CIN detection (e.g., more CIN2+ with co-testing after Round 1, and less CIN3+ with co-testing after Round 2). Cumulative invasive cancer detection was similar or slightly higher in cytology alone compared with co-testing, with findings limited due to incomplete reporting of full followup for all participants, particularly after the second round of screening. Three of four co-testing trials (POBASCAM, Swedescreen, ARTISTIC) had a high threshold for colposcopy referral, generally referring women for high-grade squamous intraepithelial lesion (HSIL+) cytology, with colposcopy referral for HPV positive results (with normal cytology, ASC-US, or LSIL) only after repeat testing for persistent HPV positivity and/or abnormal cytology. Also, none of these three trials has complete reporting for Round 2 (and therefore cumulatively) for a substantial proportion of trial participants (POBASCAM), for the complete followup period (Swedescreen), or for both (ARTISTIC). Data from a third screening round reported in 2011 from ARTISTIC do not correct these deficiencies, but address 6-year cumulative rates of CIN2+ and CIN3+ development by baseline screening test results.

In the only co-testing trial that found a cumulative increase in relative CIN detection for any CIN measure (NTCC Phase I), women were referred to colposcopy immediately at a lower cytology threshold (ASC-US+) or if HPV positive. Relative to cytology alone, this strategy increased both CIN2+ and CIN3+ after one screening round and cumulative CIN2+ overall; however, it did not significantly reduce CIN3+ in Round 2 or affect cumulative CIN3+. Invasive cancers were higher in the cytology arm in both rounds, and therefore cumulatively, but small numbers complicate interpretation. Cumulative CIN2+ was increased (RR, 1.50 [95% CI, 1.13 to 1.98]), perhaps reflecting overdiagnosis of regressive disease. Indirect comparisons between NTCC Phase I and II in older women suggest no additional benefit from co-testing above HPV primary screening alone, but possible increases in false positives. In NTCC Phase I, immediate colposcopies were much higher (10.6%) with co-testing than with cytology alone (3.0%), and neither phase of NTCC has reported cumulative colposcopies beyond those from immediate referral after initial screening in Round 1. Cumulative colposcopies are reported for only two trials (POBASCAM, ARTISTIC). Cumulative colposcopies were slightly higher in the co-testing arm (3.4%) of POBASCAM, compared with cytology (2.8%), although both arms received HPV testing with polymerase chain reaction (PCR) in Round 2, which might minimize differences. For women aged 30 to 64 years, cumulative colposcopy referrals after two screening rounds were 6.0 percent in the co-testing (HC2-LBC) arm in ARTISTIC, compared with 4.9 percent in the LBC only arm. Results from a third screening round are expected from at least one trial (ARTISTIC), which could be important, since ARTISTIC varied somewhat from other trials in several round-specific findings. Age-specific ARTISTIC data are not completely reported by rounds, thus some of these data include the 21 percent of women younger than age 30 years.

Two fair- or good-quality studies reported absolute sensitivity and specificity for HC2-CC co-testing among 17,885 women aged 30 to 60 years in countries with established cervical cancer screening programs. Studies used a positive definition from either co-test, so that all HPV positives met the threshold. For the detection of CIN3+ or CIN2+, HC2 plus cytology was 44 to 56 percent more sensitive than ASC-US+ cytology alone, but was 4.2 to 4.8 percent less specific. In these studies, the combination of HC2 plus cytology did not differ significantly from the use of HC2 alone in sensitivity (100% vs. 97 to 98%) or specificity (93 to 94% vs. 94 to 95%) for the detection of CIN2+ or CIN3+ lesions.

Indirect comparisons in trials and absolute test performance studies suggest that adding cytology to primary HPV screening (HPV-CC co-testing) does not significantly improve sensitivity but may decrease specificity compared to HPV alone. More rounds of screening could help determine if there may be other values for co-testing, such as identification of a cohort negative on both tests, that are appropriate for prolonged intervals before rescreening.

Women younger than age 30 or 35 years. Two co-testing trials included women younger than age 30 or 35 years (NTCC Phase I and ARTISTIC). Because complete age-specific results are not available from ARTISTIC, it is discussed with results for women older than 30, who represent almost 80 percent of the total sample. Among 11,810 women aged 25 to 34 in NTCC Phase I, women were referred for ASC-US+ cytology but retested for HPV positive-cytology normal results. In contrast to other co-testing trials, no impact on CIN3+ in any round or cumulatively was seen in younger women. CIN2+ detection was relatively greater after Round 1 and cumulatively with co-testing, perhaps reflecting overdiagnosis of regressive disease. No cancer was found in the co-testing arm, although three cases were found in the cytology only arm. Although cumulative colposcopies are not yet reported, much higher initial colposcopies after co-testing compared with cytology (11.9 vs. 4.1%) are consistent with likely increased false positives and related harms in a co-testing strategy, compared with cytology alone, in younger women. Very limited absolute test performance data in younger women suggest a single co-test (with positive defined as LSIL+ or both ASC-US+/HPV+) decreased sensitivity relative to HPV testing alone, but remained similar to cytology alone, while specificity improved relative to either HPV testing alone or cytology alone. This strategy mimics triage if either test is positive (HPV+ or ASC-US+) using the other, and requires both to be positive for colposcopy referral.

KQ3: HPV for triage of ASC-US or LSIL cytology. Three cross-sectional (two of fair quality and one of good quality) and one prospective cohort study (of fair quality) compared HC2 with repeat cytology for the triage of women aged 15 to 78 years with ASC-US cytology results to colposcopy, two of which also compared HC2 with repeat cytology for triage of LSIL. Pooled estimates for the detection of CIN2+ among women with ASC-US cytology results demonstrated a 12 percent higher relative sensitivity for HC2 compared to repeat cytology (95% CI, 0 to 24) at a threshold of ASC-US, but no difference in specificity. One study evaluated HPV triage of ASC-US for the detection of CIN3+ and found no difference between HPV and repeat cytology. HPV testing strategies showed very poor absolute specificity for triaging LSIL (29.9 to 44.0% for CIN2+, 27.1% for CIN3+). In one small study (n=749) of ASC-US only, age-specific sensitivity for CIN2+ did not differ by age among women older and younger than age 35 years. However, in women aged 35 years and older, specificity for HC2 was better than for repeat cytology (84.8 vs. 74.7%), while specificity for HC2 in women younger than age 35 years tended to be lower than repeat cytology (60.4 vs. 65.5%).

Two good-quality RCTs evaluated HPV testing and repeat cytology versus repeat cytology alone for triage of ASC-US and LSIL Pap smears. Women were referred for HPV+ in either trial, for cytology of HSIL+ in the ALTS trial, or for ASC-US+ in the Swedish trial. Among women aged 18 to 35+ years (78% younger than age 35 years) in the ALTS trial, those triaged with HPV and repeat cytology for ASC-US screening results showed a nonsignificant increase in CIN3+ detection (RR, 1.24 [95% CI, 0.88 to 1.73]), compared to repeat cytology alone every 6 months for 2 years. Due to high prevalence of HPV in women with LSIL, 85 percent of women in the HPV-enhanced triage arm were referred to colposcopy, which was therefore discontinued as an unsuccessful triage strategy. The smaller Swedish trial mixed outcomes for women referred for either LSIL or ASC-US, but showed a similar impact on CIN3+ detection (RR, 1.20 [95% CI, 0.88 to 1.63]). Relative CIN3+ detection may be better in women aged 30 years or older compared to women younger than age 30 years. Both trials increased relative colposcopies. Neither trial exactly mimics current U.S. practice or guidelines.

Key Question 5: Harms of HPV testing. Four studies that examined the psychological impact of HPV testing found increased levels of immediate anxiety and distress in women testing positive for HPV compared to HPV negative women. These differences, however, were resolved at 6-month followup.

The evidence we reviewed indicates that a reasonable age at which to initiate cervical cancer screening in women is age 21. Screening before this age is complicated by relatively high rates of transient HPV and regressive cervical abnormalities, with very few actual cancer cases. Current data cannot assure that beginning screening after this age is clearly safe, particularly in the United States, which has no centralized national cervical cancer screening program.

For cytology-based screening, LBC does not differ from CC in sensitivity, specificity, or relative CIN detection, but may yield a lower proportion of unsatisfactory slides. Cost, overall screening strategy, and other considerations may also pertain to local decisions on which approach to use for collecting cytology samples.

In women older than age 30 years, a single HC2 test is clearly more sensitive for CIN2+ and CIN3+ (about 40% greater) than cytology alone. However, a single HC2 test is also 3 to 5 percent less specific than cytology. Thus, while HPV-enhanced screening strategies offer a potential disease detection benefit compared with cytology alone, the potential burden due to increased false-positives is critical to understand, particularly given the relatively low incidence of cervical cancer and the established practice of repeated cervical screening.

Based on large trials, primary screening using a clinically validated HPV test, such as HC2, appears very promising in women aged 35 years and older, particularly when coupled with reflex cytology to triage positive HPV results before colposcopy. HPV testing enhances the detection of CIN3+, but also increases CIN2+ detection and immediate colposcopy referrals, compared with cytology alone. Cytology triage of positive HPV results identifies women with milder abnormalities for further followup and retesting, thus reducing the proportion immediately referred for diagnostic colposcopy. Eventually, after repeated screening rounds are reported, this strategy may be shown to reduce overall colposcopies and false-positive related harms—including some of the overdiagnosis and treatment of regressive disease—which both absolute test performance and existing trial data suggest are likely with primary HPV screening alone. While not yet reported, cumulative colposcopy requirements, treatments, and related harms are essential to determine net benefit from any enhanced CIN detection/cancer prevention with primary HPV screening. Thus, the net impact of primary HPV screening (with or without cytology triage) remains to be determined through completion of ongoing trials and more detailed reporting of potential harms, as well as benefits from completed trials.

Screening with combined HPV/cytology (co-testing) in women 30 years and older is much more sensitive than cytology alone, but may represent a strategy that adds little to HPV screening. Based on indirect comparisons between trials and on test performance data, one-time HPV/cytology co-testing appears to be very similar to HPV testing alone for the detection of CIN2+ or CIN3+, with similar (or slightly reduced) specificity. Compared with cytology alone, co-testing trials of repeated screening did not clearly report a consistent disease detection pattern indicating benefit, although most reported reduced relative CIN3+ detection in the second screening round, which may suggest benefit. Determination of net program impact is not possible, since most trials have not yet reported complete cumulative outcomes ascertainment for the entire study population, nor cumulative colposcopy requirements and related harms. Once trial data are more completely reported, judgment as to their applicability will be required, since co-testing trials to date used screening and retesting protocols that are not entirely relevant to U.S. practice.

In women younger than 30 years, there are much less data on primary HPV screening (with or without cytology triage) or co-testing with HPV-cytology. Where available, these indicate using HPV in any primary screening strategy is associated with a substantially inferior specificity (about 10 to 11% less specific) compared with cytology. Thus, HPV screening in younger women is likely to result in substantially more colposcopy referrals, greater regressive CIN2+ detection and treatment, and increased treatment-related harms, particularly compared with older women. Current data are inadequate to determine whether, and how much, cytology triage might mitigate specificity concerns with primary HPV screening performance in younger women, but caution is warranted.

For all HPV-enhanced primary screening approaches, results to date raise questions about possible overdiagnosis of regressive (or non-progressive) lesions and/or a high burden for a small net benefit in the context of frequently repeated screening as is typically done for cervical cancer in the United States. No available trials report adequate cumulative data on the proportion of women undergoing repeat testing, resulting colposcopy referrals, rates of treatment and diagnosis, or treatment-related harms, all of which are critical to addressing the issues of relative burden and harms of newer strategies relative to cytology. Thus, the net impact of HPV-enhanced primary screening remains elusive, but may become clearer after more in-depth reporting from trials reviewed here and reports from ongoing trials and studies, as well as international efforts to pool results of HPV-enhanced cervical cancer screening trials. Modeling exercises may also be useful.

A major benefit of HPV-enhanced primary screening could be identification of a low-risk cohort in whom a prolonged screening interval would be appropriate. Risk-stratifying approaches have not been directly incorporated into trials to date, and safety data for prolonged screening intervals in low-risk women based on baseline HPV testing (with or without cytology) are still accruing from trials and cohort studies. However, ensuring the acceptability of overall program requirements and feasibility would be important considerations in cervical cancer screening policy, even after it is shown that a large proportion of women could be safely risk-stratified to longer screening intervals. Ongoing research in HPV subtypes, as well as HPV-related biomarkers, could further advance efforts in risk stratification for appropriately targeted screening.

For the triage of women with ASC-US cytology to colposcopy, a single HC2 test has a higher sensitivity and similar specificity compared to single repeat cytology at a threshold of ASC-US for the detection of CIN2+. No additional benefit occurs when HC2 triage is combined with cytology, but this strategy increases false positives. HC2 does not appear useful for the triage of women with LSIL cytology because such a high proportion of women will test positive. HPV testing has few unique harms compared with cytology screening, but a positive HPV test may increase anxiety and distress, in the short-term only. Further research could be useful.

The most thoroughly studied HPV test for use in cervical cancer screening or triage is HC2. Data reported in this review primarily reflect results using HC2 at a positive threshold of 1 pg/ml and, to a lesser extent, PCR GP5+/6+. Careful consideration of all aspects of other tests’ performance characteristics (sensitivity, specificity, PPV, negative predictive value) in screening settings is warranted before substituting tests, particularly in a population-based screening program.