Abstract
Cervical cancer screening has undergone a transformation in recent decades. Historically, programs were based on cervical cytology (i.e., “Pap smear”), which had to be repeated often because of its limited sensitivity and reproducibility. In more recent years, the discovery of human papillomavirus (HPV) as the necessary cause of virtually all cervical cancers has led to the introduction of HPV testing into clinical practice, first as a triage test for minor cytologic abnormalities, then in conjunction with cervical cytology (cotesting), and most recently, as a standalone screening test. Multiple randomized trials have shown that HPV-based screening has higher sensitivity compared with cytology, providing great reassurance against cervical precancer and cancer for women testing HPV-negative for many years. Analyses have also been conducted in support of the recent U.S. Preventive Services Task Force guidelines that show that primary HPV screening achieves the greatest balance of benefits and harms compared with other strategies. An added benefit of primary HPV testing is the ability to conduct it from self-collected samples, which is critical for extending coverage among hard-to-reach individuals and could provide a safe and effective alternative to in-person screening visits during the COVID-19 pandemic.
See related article by Liang et al., p. 474
The primary goal of cervical cancer screening is to detect cervical precancers that can be treated to prevent invasion. Cervical screening programs have led to substantial reduction of cervical cancer incidence and mortality. For decades, cervical cytology (i.e., “Pap smear”) has been the mainstay of cervical screening. However, cytology has limited sensitivity and reproducibility, and requires frequent repetition to achieve high program sensitivity. The recognition that almost all cervical cancers are caused by human papillomaviruses (HPV) has led to introduction of HPV assays in clinical practice, first to triage minor cytologic abnormalities, then as an adjunct to cytology in primary screening (cotesting), and most recently, as standalone screening test (1, 2). HPV testing has much higher sensitivity for cervical precancer, and a much longer lead-time compared with cytology, because it detects HPV infection, the first, necessary step in cervical carcinogenesis. Development of cervical cancer after an initial HPV infection can take decades (2). Therefore, women with negative HPV test results can be reassured for a long time against cancer risk, and screening intervals can be safely extended compared with cytology.
Current screening recommendations in the United States include all three screening modalities (cytology, cotesting, and HPV alone; refs. 3, 4). While it is widely accepted that HPV-based screening approaches are better than cytology alone, there is an ongoing debate about the benefit of adding cytology to HPV screening. Large studies have shown that the small increase in sensitivity of cotesting versus HPV alone is outweighed by a lower specificity and higher cost because two tests are conducted in the whole-screening population (5). Consequently, recently published American Cancer Society screening guidelines give a preference for HPV alone over cotesting, after weighing the benefits and harms (3). Despite the strong evidence in support of primary HPV screening, a recent analysis of cervical screening performance in a large commercial laboratory system challenged screening guidelines and stressed the added value of cytology to HPV testing for detecting prevalent cervical cancers within a year (6). However, the analysis and interpretation of the large dataset had some shortcomings (7). First, the primary goal of cervical cancer screening is to detect and treat precancers, not to diagnose cancer. Furthermore, it is important to recognize that many of these prevalent cancers are already symptomatic and the cotest is conducted as part of the clinical evaluation of symptoms, not as a screening test. In some cases, HPV testing may be negative when advanced cancers are detected by microscopic signs of necrosis. HPV-negative cervical cancers are exceedingly rare and most likely represent misclassified cancers from other sites (8, 9).
The article by Liang and colleagues, starting on p. 474, addresses this question through a comparative analysis of cytology, cotesting, and HPV alone in a population-based cervical cancer screening study (1). The MARZY study was a randomized trial evaluating screening attendance following a screening invitation compared with a control group without invitation in an opportunistic screening setting. MARZY was powered to detect differences in screening attendance between two different types of invitations and the control group. The current secondary study reports on data from intervention group participants, who had a conventional Pap, a liquid-based cytology test, and two different HPV assays performed, allowing to directly compare cytology, HPV alone, and cotesting in a population-based sample of women eligible for cervical screening. The authors showed similar sensitivity and specificity of HPV alone compared with cotesting. Strengths of the study include the population-based sample and the fact that all tests were conducted in all women who participated. The investigators attempted to account for verification bias by inviting a subset of women with negative screening results to colposcopy, although the merit of verification bias adjustment in HPV screening studies has been questioned previously (10). An important limitation of the study is that the number of cervical precancers in this population was very low (only 19 CIN2+), and thus the study was not adequately powered to compare sensitivity between HPV alone and cotesting. In addition, follow-up among the screen positives (atypical squamous cells of undetermined significance or worse, or HPV positive) was 65%, and only 36% of the small random sample of screen negatives attended colposcopy; it is not clear whether follow-up was differential by screening result.
Despite these limitations, the findings from Liang and colleagues' study align with other reports, including our own data from KPNC that show only a minor difference in risk of CIN3 and cancer in women negative for HPV alone versus cotesting (11). The 2018 U.S. Preventive Services Task Force (USPSTF) guidelines for cervical cancer screening were supported by an extensive data synthesis and modeling effort that also addresses the HPV alone versus cotesting question (5, 12). The modeling approach allows to evaluate the number of tests needed, colposcopies performed, disease endpoints (CIN3 or greater) detected, and cervical cancer–related deaths over a woman's lifetime for different screening approaches, including cytology alone, primary HPV testing, and cotesting at various ages and screening intervals (12). The ability to generate these estimates over a lifetime rather than a single screening round is important, because the three screening strategies have different management approaches, surveillance intervals for screen positives, and screening intervals for screen negatives, complicating cross-sectional comparisons at individual timepoints.
According to the USPSTF model, primary HPV testing with cytology triage at 5-year intervals beginning at age 30 years had the best tradeoff of benefits and harms. Cotesting at 5-year intervals beginning at age 30 years was associated with 60% more lifetime total tests and 12% more colposcopies with similar number of cases detected compared with the analogous strategy with primary HPV testing and cytology triage. There was no difference in number of cervical cancer–related deaths between the two strategies (Fig. 1).
Comparison of lifetime cervical cancer screening outcomes for cytology and HPV-based screening strategies. Data used for this figure were obtained from the USPSTF modeling study evaluating different strategies for cervical cancer screening (12). Outcomes are shown for various screening strategies, including (1) cytology alone beginning at age 21 years at 3-year intervals; (2) HPV and cytology cotesting beginning at age 30 years at 5-year intervals; and (3) HPV alone, with cytology triage beginning at age 30 years at 5-year intervals. Strategies 2 and 3 assume screening with cytology alone at 3-year intervals for women ages 21–29 years. For all strategies, screening is assumed to end at age 65 years. Outcomes are calculated from age 20 to 100 years and are all expressed as numbers per 1,000 women over a lifetime of screening. Total number of tests does not include colposcopies.
Comparison of lifetime cervical cancer screening outcomes for cytology and HPV-based screening strategies. Data used for this figure were obtained from the USPSTF modeling study evaluating different strategies for cervical cancer screening (12). Outcomes are shown for various screening strategies, including (1) cytology alone beginning at age 21 years at 3-year intervals; (2) HPV and cytology cotesting beginning at age 30 years at 5-year intervals; and (3) HPV alone, with cytology triage beginning at age 30 years at 5-year intervals. Strategies 2 and 3 assume screening with cytology alone at 3-year intervals for women ages 21–29 years. For all strategies, screening is assumed to end at age 65 years. Outcomes are calculated from age 20 to 100 years and are all expressed as numbers per 1,000 women over a lifetime of screening. Total number of tests does not include colposcopies.
Worldwide, HPV-alone testing is widely accepted as the preferred strategy for cervical screening. In 2017, the Netherlands became the first country to implement nationwide primary HPV screening, followed later by Australia and several other countries (13). Given its superior sensitivity compared with cytology and visual inspection with acetic acid, HPV-based screening is also an ideal strategy for low-resource settings, providing good protection with fewer rounds of screening over a woman's lifetime. However, there is currently a lack of robust low-cost HPV assays that can be implemented in resource-constrained settings, limiting the introduction of HPV screening in many places.
A key benefit to HPV testing is that, unlike cytology, it can be successfully performed on self-collected specimens. Self-sampling with HPV testing obviates the need for an in-person pelvic exam and therefore, has the potential to extend screening coverage to underscreened/unscreened women in both high- and low-resource settings. Results from a large meta-analysis have demonstrated similar accuracy of HPV testing using PCR-based HPV tests from self-collected samples and clinician-collected samples (14).
In addition to expanding coverage to hard-to-reach populations, self-sampling provides an opportunity to serve as an alternative to clinic-based screening in the COVID-19 and post-COVID-19 eras. While the long-term impact of COVID-19–related shutdowns on cervical cancer incidence and mortality is currently unknown, a short-term delay among women with a negative screening history is certainly less concerning than longer delays and lack of follow-up of positive screening tests. Screening rates in 2020 thus far have been well-below levels reported in recent years (15). As health systems begin to reopen to preventive care visits, concerns among patients and providers regarding SARS-CoV-2 exposure may result in fewer in-person visits for cervical cancer screening for the foreseeable future. Offering self-sampling as an alternative option would be a safe and efficient strategy to ensure the continuation of effective cervical cancer screening, while reducing the risk of exposure to SARS-CoV-2 through in-person clinic visits. Importantly, ensuring that underserved and vulnerable populations have equal access to self-sampling and follow-up care is critical to prevent cervical cancer disparities from widening, especially as these populations have been disproportionately affected by the pandemic. While HPV self-sampling has been introduced in several screening programs worldwide, it currently lacks regulatory approval in the United States. Accelerating regulatory approval of HPV self-sampling will have important immediate public health benefit by extending the reach of cervical screening and making cervical screening programs more resilient against disruptions like the current pandemic.
When introducing HPV-based screening, it is important to consider the need for triage of HPV-positive women, because most HPV infections are transient and do not require treatment (16). Currently, approved options for triage include cytology, partial genotyping (in combination with cytology), and dual stain (13, 16, 17). More work is needed to develop triage strategies that can be implemented in low-resource settings that do not have expertise and infrastructure for cervical cytology. Furthermore, triage strategies that can be performed from self-collected specimens could further improve screening for underscreened and hard-to-reach populations. Promising approaches include host and viral methylation assays that are amenable to self-collected specimens (18, 19).
In summary, the transition of cervical screening from cytology to primary HPV testing will lead to a further reduction in cervical cancer–related deaths. While the addition of cytology to HPV testing increases the number of tests performed and the number of false positives substantially, it does not translate into a meaningfully higher detection of precancer or reduction of cancer. HPV-based primary screening has great promise for low-resource settings that currently have no screening options, but robust and affordable HPV tests are needed for widespread implementation. Research and development work are needed for triage assays that can be conducted from self-collected specimens among HPV-positive women.
Authors' Disclosures
No disclosures were reported.