Evidence of Decreased Long-term Risk of Cervical Precancer after Negative Primary HPV Screens Compared with Negative Cytology Screens in a Longitudinal Cohort Study

Cervical cancer is a highly preventable disease owing to effective primary prevention [human papillomavirus (HPV) vaccination] and secondary prevention (screening for cervical precancer) strategies (1–3). Accordingly, the World Health Organization (WHO) has called for the elimination of cervical cancer (4). To attain this goal, many jurisdictions (5, 6) are in various stages of implementation of highly sensitive primary HPV screening in place of cytologic screens, which have lower cervical precancer detection sensitivity (7, 8). However, critical questions remain for how to best implement primary HPV screening.

Many high-income countries have longstanding cytology-based cervical cancer screening programs that have led to significant decreases in incidence and mortality among women and individuals with a cervix (WIC) over the past 70 years (9). Despite this, incidence rates in most high-income countries remain above the 4 per 100,000 WHO elimination threshold goal [GLOBOCAN 2020 (cited April 18, 2024); available at https://gco.iarc.fr/]. The effects of HPV vaccinations have yet to have a large impact, and vaccine coverage remains suboptimal in many regions; thus, to decrease rates in the immediate future, improved rates of screening and treatment adherence among certain high-risk populations are needed (10), as well as increased use of high-performance tests with higher sensitivity than cytology to detect precancerous cervical lesions (11). Primary HPV screening has the potential to reduce barriers to screening by allowing for reduced numbers of lifetime screens due to the test’s high negative predictive value and longer interval between negative screens (12, 13). However, primary HPV screening has lower clinical specificity than cytology: it detects HPV infections that may or may not cause precancerous lesions, necessitating triage testing of HPV-positive results. There is thus concern about the effects of overtreatment with primary HPV screening on both WIC [e.g., result-related stress (14) and pregnancy outcomes (15)] and health systems [e.g., increased referral to colposcopy; (15)]. These risks can be reduced by appropriately extended screening intervals combined with effective triage strategies (16, 17).

The FOCAL-DECADE cohort followed participants who received 1 to 2 rounds of primary HPV screening in the Human Papillomavirus for Cervical Cancer (HPV FOCAL) trial (7) followed by reentry into the provincial screening program in which participants were subsequently screened with cytology. Detection of precancerous lesions was monitored for 14 years after baseline trial screening through the British Columbia (BC) provincial Cervix Screening Registry. This study investigated the long-term risk of cervical intraepithelial neoplasia grade 2 or worse (CIN2+) and grade 3 or worse (CIN3+) among participants who had negative HPV results in the HPV FOCAL trial compared with a cohort from the general screening population who had one or two consecutive negative cytology results. The results will help identify optimal implementation strategies for primary HPV screening.

This longitudinal cohort study used data from the HPV FOCAL trial to investigate the cumulative detection rate of cervical precancer among four screening populations. The populations selected for inclusion in this study were (i) participants from HPV FOCAL who had one negative HPV test result (HPV1 cohort); (ii) a nonoverlapping group of participants from HPV FOCAL who had two consecutive negative HPV test results 4 years apart (HPV2 cohort); (iii) WIC receiving cervical cancer screening through the BC provincial screening program with one negative cytology result and those were not participants in the HPV FOCAL trial (BCS1 cohort); and (iv) a subset of BCS1 with two consecutive negative cytology results (BCS2 cohort). All follow-up was done with cytology (the standard of care in the BC Cervix Screening Program at the time of this analysis) and occurred after the final negative screen in each respective population.

The main outcome of interest was the cumulative risk of CIN3+, which is defined as the appropriate endpoint for investigating cervical cancer (18, 19). However, the cumulative risk of CIN2+ detection was also examined as CIN2 is the clinically actionable threshold for treatment in most screening programs, including in BC (20). The HPV FOCAL trial (7) and the FOCAL-DECADE cohort (21) have both been previously described, and details pertinent to this analysis are described below. Strengthening the Reporting of Observational Studies in Epidemiology study guidelines were followed in the reporting of this cohort study (Supplementary Materials and Methods).

The HPV FOCAL randomized controlled trial [isrctn.org Identifier: ISRCTN79347302; (7, 22, 23, 24)] recruited women, ages 25 to 65 years, from Metro Vancouver and Greater Victoria, BC, Canada, from January 2008 through May 2012. The trial included a control group (N = 9,457), which received cytology (liquid-based) at baseline and 24 months and cotesting (HPV and cytology) at 48-month exit; an intervention group (N = 9,552), which received primary HPV screening at baseline and cotesting at 48-month exit; and a safety group (N = 6,214), which received primary HPV screening at baseline and cytology at 24-month exit. All groups received a baseline demographic survey. The primary analysis from HPV FOCAL compared rates of CIN2+ between the control and intervention groups at baseline and 48 months, demonstrating a significantly lower risk of precancer in the intervention group at exit (7). A subset of HPV FOCAL participants from the intervention and safety groups who were HPV-negative at baseline (intervention and safety) and exit (intervention) were included in this analysis. Participants from the safety group with one negative HPV screen at baseline comprised HPV1, and participants from the intervention group with two consecutive negative HPV test results 4 years apart (baseline and exit) comprised HPV2. Participants from both groups reentered the provincial screening program after completion of the trial, in which the screening recommendations were conventional cytology screening every 24 months (up until June 2016) or every 36 months [after June 2016; (7)].

FOCAL-DECADE (21, 25) is a long-term evaluation of data obtained from original participants in the HPV FOCAL trial linked to their screening data in the provincial program. Data from FOCAL-DECADE have been previously used to investigate the long-term safety of one negative HPV screen using data from the safety group (21). This study provides updated results on the extended long-term safety of one negative HPV screen and extends previous findings by investigating the additional long-term safety gained from a second consecutive negative HPV screen. HPV1 and HPV2 were followed for 14 years after entry into HPV FOCAL to track any CIN3+ and CIN2+ detected by the screening program.

Written informed consent was obtained for all HPV FOCAL participants. Participants consented to data linkages to provincial health registries, allowing for the development of FOCAL-DECADE. Ethics approval was obtained from the University of British Columbia Clinical Research Ethics Board (HPV FOCAL: H06-04032; FOCAL-DECADE: H18-02063).

The BC Cancer Cervix Screening Program [BC Cervical Cancer Screening 2018 Program Results (cited 2024 Apr 18); available at http://www.bccancer.bc.ca/screening/Documents/Cervix-Program-Results-2018.pdf.] manages cervical cancer screening for the province, issuing guidelines and coordinating a centralized laboratory where provincial cervical screen specimens are processed. It also maintains a centralized registry of cytology, colposcopy, and treatment history (including histopathology) for all WIC in the province. Screening and treatment recommendations in the provincial program were different from HPV FOCAL, as the provincial standard of care for screening at the time of this analysis was cytology (BC transitioned to HPV screening in January 2024). In the provincial program, prior to the transition to primary HPV screening, WIC with equivocal (atypical cells of undetermined significance) or low-grade abnormal cytology results (low-grade squamous intraepithelial lesion) were recommended to repeat screening at 6-month intervals for up to a year, and those with persistent atypical cells of undetermined significance/low-grade or initial high-grade cytology results (atypical glandular cells, atypical squamous cells that cannot exclude high-grade squamous intraepithelial lesion, high-grade squamous intraepithelial lesion, adenocarcinoma in situ, or invasive carcinoma) were referred to immediate colposcopy.

To allow for a direct comparison of risk of cervical precancer after negative HPV results versus negative cytology results, two comparison cohorts were constructed from the provincial screening program (BCS1 and BCS2). To ensure comparable populations, members of the BCS cohorts met the same age eligibility criteria as HPV FOCAL participants (i.e., ages 25–65 years). Furthermore, all BCS members had a negative cytology-based screen during the HPV FOCAL recruitment interval (2008–2012). Members of BCS2 had a second consecutive negative screen so as to be comparable with HPV2 members, who had two consecutive negative HPV screens. After the final negative screen, BCS members were followed through the provincial screening program in an identical manner as HPV1 and HPV2.

The incidence rates (per 1,000) of CIN2+ and CIN3+ for the follow-up period were calculated for each cohort, as well as by age group within the cohort. Age groups were selected to allow for the identification of risk in priority groups (i.e., those less than 30 and 50+ years). Cumulative risks of CIN2+ and CIN3+ by cohort were assessed at intervals over follow-up. Prior to the transition to HPV screening in BC, cervical cancer screening with cytology was recommended at 3-year intervals; thus, this interval was considered to have adequate safety. To generate a comparison with the HPV cohorts (HPV1 and HPV2), we looked at the interval length with HPV screening that gave a CIN2+ detection rate similar to the 3-year detection rate among the cytology cohorts (BCS1 and BCS2).

Baseline characteristics using medians, IQRs, frequencies, and proportions were compared between HPV1 and HPV2 participants using Kruskal–Wallis and χ² tests for continuous variables and categorical variables, respectively. 1 − S(t) actuarial curves based on the Kaplan–Meier technique were used to plot the cumulative incidence of CIN2+ and CIN3+ detection over the follow-up period. Follow-up started at the date of the final negative HPV screen (baseline screen for HPV1 and exit screen for HPV2) in the HPV cohorts or the date of the first negative cytology for BCS1 and the second consecutive negative cytology for BCS2. Time to event was calculated as the difference between the date follow-up started and the date of the first CIN2+ or CIN3+ detection. If CIN2+ or CIN3+ was not detected, the follow-up was censored at the most recent test (i.e., cytology or colposcopy). If CIN2 later developed to CIN3, WIC were followed up to the date of CIN2 detection for the calculation of CIN2+ cumulative risk and to the date of CIN3+ detection for the calculation of CIN3+ cumulative risk. The actuarial curves of HPV1 and HPV2 were compared with BCS1 and BCS2. Multivariable Cox proportional hazard models, adjusted for age at each test, were constructed to compare the hazards between primary HPV-screened cohorts and cytology-screened cohorts. Separate models were built comparing (i) BCS1 with HPV1 and HPV2 and (ii) BCS2 with HPV1 and HPV2. The hazards between HPV1 and HPV2 were also compared to assess the additional protection from a second negative HPV screen spaced 4 years after the first test. A P value < 0.05 was considered statistically significant. All analyses were conducted using R 3.6.3 [R Foundation; (26)].

The data generated in this study are not publicly available but are available upon reasonable request from the corresponding author.

Figure 1 shows the inclusion criteria for this study. A total of 12,170 eligible participants from HPV FOCAL were included in this analysis. The safety and intervention groups of HPV FOCAL included 6,214 and 9,552 participants, respectively. Of those, HPV1 included 5,546 participants (attrition due to: 10 were lost to follow-up before exit testing, 519 were HPV-positive at baseline, and 139 had no recorded screen during the follow-up) and HPV2 included 6,624 participants (attrition due to: 14 became ineligible during the trial, 907 were lost to follow-up before exit testing, 1,093 had at least one HPV-positive screen, and 914 did not have a recorded screen during the follow-up). For the comparison cohorts, BCS1 included a total of 782,297 eligible WIC with one negative cytology result during HPV FOCAL recruitment years, among which BCS2 included 673,778 WIC who had two consecutive negative cytology results in the provincial screening program.

Although the median age at recruitment was higher in HPV1 than in HPV2 (P < 0.001), the difference was small [HPV1: 46 (IQR, 38–53); HPV2: 45 (IQR, 38–52); Table 1]. Other baseline characteristics were not significantly different between HPV1 and HPV2. The median age at the start of follow-up was 46 (IQR, 38–53) years for HPV1, 50 (IQR, 42–57) years for HPV2, 43 (IQR, 34–52) years for BCS1, and 45 (IQR, 36–54) years for BCS2 cohorts. The median length of follow-up was 10 (IQR, 8–12) years in HPV1, 6 (IQR, 4–7) years in HPV2, 10 (IQR, 6–11) years in BCS1, and 8 (5–10) years in BCS2.

During the 14-year follow-up, there were 26 CIN2+ detections in HPV1 after one negative HPV screen [4.7/1,000 (95% CI, 3.2–6.9)] and 15 in HPV2 after two negative HPV screens [2.3/1,000 (95% CI, 1.4–3.7); Table 2]. The total detection of CIN3+ was 14 [2.5/1,000 (95% CI, 1.5–4.2)] in HPV1 and 10 [1.5/1,000 (95% CI 0.8–2.8)] in HPV2. CIN2+ and CIN3+ detection rates in the BCS cohorts were higher than those in the HPV cohorts. The total detection of CIN2+ was 10,639 [13.6/1,000 (95% CI, 13.3–13.9)] and 6,548 [9.7/1,000 (95% CI, 9.5–10.0)] in BCS1 and BCS2, respectively. The total detection of CIN3+ was 7,139 [9.1/1,000 (95% CI, 8.9–9.3)] in BCS1 and 4,424 [6.6/1,000 (95% CI, 6.4–6.8)] in BCS2. The incidences of CIN2+ and CIN3+ by age group in each cohort are shown in Table 2 and are lower in the HPV cohorts than in the BCS cohorts across all age groups. The cumulative risk of CIN2+ and CIN3+ remained lower over the follow-up period for both HPV cohorts compared with cytology cohorts, and after 10 years of follow-up, the risk of precancer in the HPV cohorts was about a third of that in the cytology cohorts (Table 3).

Prior to the January 2024 transition to primary HPV screening, cervical cancer screening with cytology was recommended every three years in BC. Thus, we set the level of acceptable risk of CIN2+ detection as the cumulative risk in the BCS cohorts 3 years after their final negative screen. The risks were 3.3/1,000 (95% CI, 3.1–3.4) for BCS1 and 2.5/1,000 (95% CI, 2.4–2.6) for BCS2 (Table 3). This was similar to the risk in the HPV cohorts 8 years after their final negative screen [HPV1: 3.2/1,000 (95% CI, 1.6–4.7); HPV2: 2.7/1,000 (95% CI, 1.2–4.2)].

Figure 2 shows the CIN2+ cumulative incidence curves of HPV1 and HPV2 compared with those of BCS1 and BCS2 (Panel A). The CIN2+ curve was significantly higher for BCS1 compared with HPV1 (P < 0.001) and HPV2 (P < 0.001). The hazard adjusted for age at screens was significantly lower among HPV1 [HR = 0.48 (95% CI, 0.33–0.70); P < 0.001] and HPV2 [HR = 0.51 (95% CI, 0.31–0.84); P = 0.008] compared with BCS1. When the HPV cohorts were compared with BCS2 (Table 4), the adjusted hazard for CIN2+ detection remained significantly lower in HPV1 [HR = 0.41 (95% CI, 0.28–0.61); P < 0.001] and HPV2 [HR = 0.52 (95% CI, 0.31–0.86); P = 0.01]. In addition, when HPV1 and HPV2 were compared with each other, the CIN2+ curves were not statistically significant [HR = 1.00 (95% CI, 0.49–2.03); P > 0.999]; results not shown. Similar results were demonstrated in the analyses of CIN3+ detection (Panel B).

Among the 15 CIN2+ detections in HPV2, 2 (13.3%) were not detected by primary HPV screening but by cytology in the exit cotesting, of which 1 was CIN2 and 1 was CIN3. It is assumed that these two CIN2+ missed by HPV screening would have been detected at the first cytology screen upon reentry into the provincial screening program (approximately 3 years after trial exit). A sensitivity analysis was conducted by extending the date of CIN2+ detection to 3 years after cotesting for the two “missed” CIN2+. The cumulative risks had minimal changes, and the results of the survival model remained the same (Supplementary Tables S1 and S2).

Here, we investigated the long-term safety of primary HPV screening for cervical precancer detection. Participants who had 1 to 2 negative HPV screens had a third the risk of a future cervical precancer compared with those who had 1 to 2 negative cytology screens for more than a decade of follow-up. The risk of cervical precancer 8 years after a negative HPV screen was similar to the risk 3 years after a negative cytology screen. Similar to the US Preventive Services Task Force recommendations (27), the screening program in BC currently recommends cytology screening at 3-year intervals, suggesting that the risk of CIN2+ (the threshold for treatment) associated with this interval is acceptable. Thus, the results from our study suggest that the appropriate screening interval in a program using primary HPV screening can be extended well past the 5-year recommendation currently used in most existing programs (28). This extension in screening interval for those who are HPV negative, along with effective triage strategies for those who are HPV positive, could potentially reduce unnecessary referrals to colposcopy and treatment, thus lessening the burden on healthcare systems and minimizing risks for perinatal issues associated with treatment (15).

Additionally, we found that over the duration of the follow-up period, cervical precancer detection after negative HPV screens remained low and was significantly lower than detection after normal cytology screens. The cumulative risk of CIN2+ 14 years after one negative HPV screen was comparable with that 6 to 7 years after normal cytology screens. These findings were consistent across all age groups investigated and are in line with prior research, indicating that primary HPV screening is a more effective screening tool than cytology for the early detection of precancerous lesions (7, 29–31). The low long-term risk of cervical precancer after a negative HPV screen allows for the consideration of fewer recommended lifetime screens. Remaining resources could be focused on reaching underscreened populations and supporting navigation to appropriate follow-up for those who have abnormal screens. The well-extended assurance against the future development of precancer offered by HPV screening compared with cytology demonstrated in this study provides additional evidence to support prior findings that suggest that the implementation of primary HPV screening for cervical cancer could be more feasible (32) and cost-effective (33–35) compared with cytology-based screening.

This study was conducted using over a decade of robust data from a large, randomized trial and a comprehensive screening registry. The provincial cervix screening program in BC maintains a registry that tracks and compiles 100% of cervix screens and related treatments conducted in the province, allowing for complete follow-up of participants. Furthermore, we used data from the only randomized trial in North America that compared primary HPV screening with cytology for cervical cancer. However, the results of this study should be interpreted in light of its limitations. Although trial baseline screens were conducted using primary HPV screening, exit screens utilized cotesting to obtain a complete census of cases, leading to two (13.3%) CIN2+ detections in HPV2 being detected by cytology and not HPV testing in exit cotesting. It is unclear if and when these cases would have been detected in a primary HPV screening program, yet prior work from HPV FOCAL demonstrated that >8 times more cases were missed with cytology compared with primary HPV screening (36). A sensitivity analysis assuming that these cases would have been detected at the following screen did not affect our main findings. Furthermore, although the safety group (HPV1) and intervention group (HPV2) of the FOCAL trial were randomized at trial entry, differences in drop-out and screening rates may have occurred over the follow-up period after they exited the study, leading the groups to no longer be comparable.

The BC Cervix Screening Program offers active outreach and follow-up to improve screening rates. Eligible WIC and their healthcare providers are notified when screening is due, and the registry tracks adherence to screening recommendations and monitors the performance of the program [BC Cervical Cancer Screening 2018 Program Results (cited 2024 Apr 18); available at http://www.bccancer.bc.ca/screening/Documents/Cervix-Program-Results-2018.pdf]. These resources are notably absent in many jurisdictions, including most of the United States (37). High screening rates and linkage to follow-up for those who are screen-positive are critically linked with the ability to effectively screen for cervical cancer, regardless of the screening and follow-up interval lengths recommended. Thus, accessible and organized screening and program surveillance are key components to the success of cervical cancer screening programs.

Prior studies have posited that cotesting only minimally increases the sensitivity of primary HPV screening, suggesting that the added cytology test is unneeded in the context of an HPV screening program (37–39). These studies were conducted using data from programs implementing cotesting; thus, our findings enhance previous results by demonstrating similar outcomes in the context of a primary HPV screening program. Results from the COMPASS trial in Australia (30), which compared primary HPV screening with cytology in a highly vaccinated population, are expected to be published later this year and will add additional insights into the discussion of cotesting versus primary HPV screening.

This study includes over a decade of comprehensive cervix screening data among WIC who received negative HPV screens or negative cytology screens that were used to compare the long-term risk of CIN2+ detection based on the screen strategy. In our population, the risk of future precancer detection in the cohorts that had negative HPV screens was a third of that in the cohorts that had negative cytology screens during more than a decade of follow-up. Furthermore, it took 8 years of follow-up to reach the acceptable risk of precancer detection in HPV-negative participants based on an only 3-year interval with cytology screening. Thus, primary HPV screening programs may be feasibly extended well beyond the current 5-year recommendation. Finally, the risk of precancer detection after a negative HPV screen remained low throughout the duration of the 14-year follow-up, suggesting the potential impact of even one lifetime primary HPV screen to reduce cervical cancer risk in underscreened populations, in line with the WHO’s current minimum recommendations [WHO Cervical Cancer Elimination Initiative (cited 2024 Apr 18); available at https://www.who.int/initiatives/cervical-cancer-elimination-initiative]. These results could be used to assist decision-makers in creating updated cervical cancer screening recommendations to meet the WHO’s cervical cancer elimination objectives.

Future studies will continue to follow this robust cohort from the HPV FOCAL trial and the corresponding comparison cohort from the provincial screening program to better understand their risk for the development of precancerous and cancerous cervical lesions and to better inform screening recommendations for primary HPV screening programs.

A. Gottschlich reports grants from Michael Smith Health Research BC (RT-2921-1595) during the conduct of the study. M. Lee reports grants from the NIH and Canadian Institutes of Health Research (CIHR) during the conduct of the study, as well as personal fees from Merck and grants from the CIHR outside the submitted work. J. Melinikow reports grants from the NCI during the conduct of the study. E.L. Franco reports grants from Merck and personal fees from Merck outside the submitted work, as well as a patent for methylation markers issued to self. G.S. Ogilvie reports grants from the NIH and CIHR during the conduct of the study. No disclosures were reported by the other authors.

The funders did not play a role in the study’s design, conduct, or reporting.

A. Gottschlich: Conceptualization, formal analysis, writing–original draft, writing–review and editing. Q. Hong: Formal analysis, writing–original draft, writing–review and editing. L. Gondara: Data curation, formal analysis, writing–review and editing. M.S. Alam: Writing–review and editing. D.A. Cook: Conceptualization, funding acquisition, writing–review and editing. R.E. Martin: Conceptualization, funding acquisition, writing–review and editing. M. Lee: Conceptualization, writing–review and editing. J. Melinikow: Conceptualization, funding acquisition, writing–review and editing. S. Peacock: Conceptualization, funding acquisition, writing–review and editing. L. Proctor: Writing–review and editing. G. Stuart: Conceptualization, funding acquisition, writing–review and editing. E.L. Franco: Conceptualization, funding acquisition, methodology, writing–review and editing. M. Krajden: Conceptualization, funding acquisition, writing–review and editing. L.W. Smith: Conceptualization, writing–review and editing. G.S. Ogilvie: Conceptualization, funding acquisition, writing–review and editing.

We would like to acknowledge the contribution of all BC healthcare providers and thousands of BC women who participated in the HPV FOCAL trial. This work was supported by the NIH (R01 CA221918, G.S. Ogilvie), the Canadian Institutes of Health Research (MCT82072, G.S. Ogilvie), and the Michael Smith Health Research (RT-2021-1595, A. Gottschlich).