4.1.1. Justification

A strong recommendation was made for using HPV DNA detection as a primary screening test when part of a screen-and-treat approach or a screen, triage and treat approach because a higher value was placed on the greater reductions in cervical cancer and deaths that are likely with HPV DNA detection compared with using VIA or cytology as a primary screening test (moderate-certainty evidence). There may also be fewer harms, such as preterm deliveries, when screening with an HPV DNA test compared with VIA. HPV DNA testing by the provider or by self-sampling may have similar effects, so either method of testing was suggested (low-certainty evidence). HPV DNA testing is largely acceptable to women and providers, is feasible and is more likely to lead to more equitable access to screening.

A conditional recommendation was made to use either HPV DNA detection followed by treatment or HPV DNA detection with a triage test because the balance of benefits and harms may be similar for either approach (moderate-certainty evidence). The benefits and harms may also be similar with any of the triage tests considered (moderate-certainty evidence), but the choice of approach should be made depending on context, because the feasibility and the resources needed for triage tests vary across settings.

On the age at which to start screening, there is evidence from modelling and large databases measuring the incidence of cervical cancer and CIN that supports the initiation of screening at the age of 30 years (moderate-certainty evidence). Starting screening at this age is likely to be acceptable to stakeholders, is feasible and needs fewer resources than starting at an earlier age. There is low-certainty evidence from longitudinal studies of the benefits of screening and of the continued risk of CIN and cervical cancer after the age of 50 years; the evidence suggests there are benefits of continued screening, following regular screening intervals until there have been two consecutive negative screening results after the age of 50. Conditional recommendations were made on the screening intervals and the age at which to stop screening based on modelled evidence showing greater benefits and fewer harms with 5- to 10-year screening intervals with HPV DNA testing, compared with more frequent screening or similar intervals using cytology or VIA (low-certainty evidence).

Conditional recommendations were made for HPV DNA testing 12 months after treatment and 24 months after a negative triage test, if screened initially with an HPV DNA test, or 12 months after a positive cytology test (but negative colposcopy); this is because there may be greater benefits and fewer harms compared with alternative follow-up times (low-certainty evidence based on modelling).

4.1.2. Summary of the evidence

For further details on the following, refer to the summaries in Web annex A (Syntheses of the evidence) and Web annex B (EtD tables) – web links provided at the end of the Table of Contents.

Primary screening tests (Recommendations 1, 3 and 4): The GDG considered evidence for the effects of different primary screening tests and found moderate-certainty evidence from a synthesis of two randomized controlled trials and modelling that HPV DNA testing followed by treatment is more likely to lead to greater reductions in CIN2+ lesions, advanced (stage II+) cervical cancer and cervical cancer deaths at follow-up, compared with VIA testing (Web annex A, IARC handbook materials and Supplementary material 13). VIA has limitations, such as subjectivity and wide variation in accuracy. The modelling also showed that primary HPV DNA testing followed by treatment may result in fewer treatments overall and fewer preterm deliveries in the screened population compared with primary VIA testing, even when assuming favourable VIA test performance.

Evidence from randomized controlled trials and cohort studies indicated a substantially lower 3- to 10-year CIN3+ risk and a lower cancer risk (up to 70% lower) after a negative HPV DNA test compared with the risk after negative cytology (IARC handbook materials).

The modelling showed that discounted costs were similar to HPV DNA testing and VIA testing, and lower compared with cytology (followed by colposcopy). HPV DNA testing was most cost-effective. Although it was not modelled, the sensitivity and specificity of self-collected upper vaginal samples for HPV testing may be similar to those of cervical samples taken by health-care providers for the detection of CIN2+ using clinically validated polymerase chain reaction (PCR)-based high-risk HPV DNA tests. The effects of the HPV screening algorithms may therefore be similar using different sampling methods (low-certainty evidence) (Web annex A, IARC handbook materials and Supplementary material 1).

Screen-and-treat versus screen, triage and treat (Recommendations 2 and 3): Studies measuring outcomes from the published literature did not directly compare screen-and-treat algorithms with screen, triage and treat algorithms. The longitudinal studies assessing HPV DNA screening did, however, include strategies with HPV DNA alone, and with cytology and/or colposcopy as triage, and found that there are likely to be similar benefits across the strategies (moderate-certainty evidence). We also mathematically modelled outcomes. The modelling found that there may be similar reductions in cervical cancer and related deaths with HPV DNA detection as the primary screening test in either a screen-and-treat approach or a screen, triage and treat approach, but there may be slightly more treatments and preterm deliveries with a screen-and-treat approach. When comparing the effects of different triage tests after HPV DNA primary screening, there may be slightly fewer cervical cancers and related deaths when HPV16/18 genotyping is used as a triage compared with the use of colposcopy, VIA or cytology as a triage test.

Age to initiate or stop screening (Recommendations 5, 6 and 7): Evidence to determine the ages at which to start and stop screening was reviewed from both the published literature and from mathematical modelling. Moderate-certainty evidence found that there are likely to be small numbers of women with cervical cancer before the age of 30 years (3/100 000 at age 20; 5/100 000 at age 25; and 12/100 000 at age 30), and a small proportion of women between the ages of 15 and 29 years with CIN2 or CIN3+ (1400/100 000 and 700/100 000, respectively; Web annex A, Supplementary materia 4).

Another review found that a greater proportion (60%) of women younger than 30 with CIN2 may regress back to CIN1 or normal after 24 months of surveillance, compared with 44% of women older than 30. Low-certainty evidence from modelling also found that when screening is started at the age of 30 years using strategies with HPV DNA screening, there are greater reductions in cervical cancer and deaths compared with starting at the age of 35 years, but there may be slightly more harms, such as more treatments and a higher risk of preterm deliveries. There were also greater costs associated with starting at 30 years of age (compared with starting at age 35) since more resources are needed for screening and treatment (Web annex A, Supplementary material 13). Starting screening at 30 or 35 years of age is likely to be acceptable to women.

Moderate-certainty evidence found that, in countries where screening was available, there were fewer women aged 50 years and older with cervical cancer (30/100 000), and in countries where screening was not available, there were more (80/100 000) (Web annex A, Supplementary material 4). Longitudinal studies showed the proportion of women with cervical cancer was similar between the age groups under and over 65 years (and that cervical cancer was primarily in women who had not been screened in the preceding interval); the risk of death from cervical cancer was also similar between age groups under and over 55 years (IARC handbook materials). This provided indirect evidence for screening strategies after the age of 50 years (low-certainty evidence). The GDG agreed that in countries where cervical cancer screening may not have been widely available, women who were older than 50 years would be at greater risk of CIN2/3 and cervical cancer. However, the age to stop screening was dependent on the population risk, the resources available and the frequency of screening between the ages of 30 and 49 years, when the risk is greatest. Barriers to screening for older women may include embarrassment, lack of knowledge about the need for screening (in particular when they have no symptoms) and fear of discomfort (Web annex A, Supplementary material 4).

Screening intervals (Recommendations 7, 8, 9 and 10): Screening intervals were also modelled (low-certainty evidence). A 5-year screening interval may result in greater benefits, fewer harms and lower costs than a 10-year one when providing HPV DNA testing with or without triage. These effects may be similar to those with cytology testing (followed by colposcopy) every three years, but better than those with cytology every five years, and better than VIA every five years (Web annex A, Supplementary material 13). Based on previous modelling to evaluate the impact of the WHO Global strategy to accelerate the elimination of cervical cancer, which demonstrated greater benefit with screening a woman twice in her lifetime compared with once (23,24), the GDG agreed that, at a minimum, twice-in-a-lifetime screening would be feasible and provide benefits, and therefore made it a good practice statement.

Acceptability, equity, feasibility, resources (all recommendations): Studies have identified potential provider perceptions that may impact the acceptability and implementation of recommendations for different screening and treatment approaches (Web annex A, Supplementary material 12). Some providers perceive that screening approaches using HPV DNA detection will increase uptake, lead to more treatment and be more sensitive in detecting lesions, whereas VIA may require more standardized training since it is subjective and dependent on expertise. Some providers also perceive that self-sampling could reduce their opportunities to provide other care for the person. Overall, it may be challenging for providers during the shift from cytology and/or VIA to HPV DNA testing as new infrastructure (e.g. a laboratory) and training are needed. A survey of women found that multiple visits can be difficult for them, and immediate treatment may be preferred (Web annex A, Supplementary material 9). A systematic review of the literature, primarily in high-income countries, found that HPV DNA testing by self-sampling is likely to improve screening uptake and may improve follow-up care slightly (Web annex A, Supplementary material 12).

A survey of GDG members identified concerns that were generally more serious when considering VIA- or cytology-based screening programmes compared with HPV DNA-based screening programmes – these concerns related to maintaining a trained workforce, generating demand for screening and ensuring high compliance with management following a positive screening test (Web annex A, Supplementary material 11). There were greater concerns about cost with HPV DNA and cytology than with VIA-based programmes. Screen-and-treat algorithms were generally more acceptable to stakeholders involved in ongoing service delivery, while screen, triage and treat algorithms were perceived to be more acceptable to health authorities, policy-makers and experts on cervical cancer. When choosing a triage test, providing cytology or colposcopy would be a challenge where it is not currently available and quality controlled and assured. However, if the HPV DNA primary screening test also gave HPV16/18 genotype information (see algorithm 4 in Table 2.2, Chapter 2), there would be no additional test to do, making it important for a self-sampling strategy. HPV DNA testing by self-sampling also has the potential to reduce cultural, socioeconomic and gender barriers to screening, which is likely to increase equity.

Follow-up after negative triage test or after treatment (Recommendations 11, 12, 13 and 14): The evidence for screening after a positive HPV DNA primary screening test and negative triage test was based on modelling of follow-up screening at 12 or 24 months, or both (Web annex A, Supplementary material 13). The modelling found that results may be dependent on the triage test. There is low-certainty evidence showing that after a negative triage test using HPV16/18 genotyping or colposcopy, follow-up screening at 24 months or at 12 months may result in similar benefits, but there may be fewer cancer treatments and lower costs at 24 months. After a negative triage test with VIA or cytology, screening at 24 months leads to slightly fewer benefits than at 12 months, but there may be fewer cancer treatments and lower costs with follow-up at 12 months. For screening at both 12 and 24 months after a negative result with any triage test, there may be slightly better reductions in cervical cancer and deaths, compared with screening at one time point only (12 or 24 months), but there may be greater harms due to more treatments and greater cost. The GDG agreed that the harms outweighed the benefits for screening at both 12 and 24 months, and although the benefits and harms at 12 months and 24 months varied slightly by triage test, a conditional recommendation was made to retest after 24 months.

The evidence on which screening test to use for follow-up after treatment following a positive screening test or after treatment for histologically confirmed CIN2/3 came from literature comparing cures and failures after treatment and on the accuracy of tests to measure failure. It also came from modelling the use of HPV DNA testing at 12 or 24 months follow-up, or co-testing with HPV DNA and cytology at 12 months (Web annex A, IARC handbook materials and Supplementary materials 3 and 13), Systematic reviews providing moderate- and low-certainty evidence on treatment failure in the general population of women suggests it may be around 10% after ablative treatments, and lower after excisional treatment. There is high-certainty evidence from reviews of the accuracy of different tests to predict recurrence or persistence after treatment in women with histologically confirmed CIN2/3 or positive screening results; this evidence indicates that the HPV DNA test is more sensitive and slightly less specific than cytology alone or co-testing with HPV DNA and cytology. High-certainty evidence also shows that positive margin status of an excised cone was less sensitive than the HPV DNA test as a predictor of failure. There is very low-certainty evidence for the sensitivity and specificity of HPV genotyping due to inconsistent results.

Low-certainty evidence from modelling studies showed that retesting women after treatment with an HPV DNA test at 12 or 24 months follow-up, or with HPV DNA/cytology co-testing at 12 months may result in similar reductions in cervical cancer and deaths; using co-testing is likely to increase treatments and costs (Web annex A, Supplementary material 13). Recommendations to extend the length of time before follow-up testing may be of some concern to practitioners, as shorter durations (e.g. 12 months) may be practised in some settings. There may also be a challenging period for a programme when transitioning from VIA or cytology to HPV DNA testing, when women may be screened with VIA or cytology but retested at follow-up with HPV DNA testing.

4.2.1. Justification

For women living with HIV, a strong recommendation was made for using HPV DNA testing as a primary screening test because a higher value was placed on the reductions in cervical cancer and deaths that are likely with this approach than on the potential harm that may occur, such as preterm deliveries (moderate-certainty evidence). When compared with VIA or cytology as a primary screening test, greater benefits are also more likely with HPV DNA testing. HPV DNA testing is acceptable to women and providers, is feasible and is not likely to lead to inequities. In some settings, HPV DNA testing is not yet available, though, and there will be a period when existing quality-assured programmes will need to remain until HPV DNA testing becomes operational.

A conditional recommendation was made to use HPV DNA testing with a triage test rather than HPV DNA testing followed by treatment because providing a triage test may lead to reduced potential harms, with minimal change in benefits (moderate-certainty evidence). The feasibility and resources needed to provide different triage tests vary across settings, thus influencing which test is chosen.

Overall, with all screening and treatment strategies, there are greater reductions in cervical cancer, deaths and CIN2/3 lesions for women living with HIV compared with the general population of women. For women living with HIV on antiretroviral therapy (ART), there were few data regarding the impact of ART on HPV-associated lesions, although the evidence is growing; therefore, recommendations based on use of antiretrovirals were not made.

For the age at which to start screening, there was low-certainty evidence from an individual patient data meta-analysis (IPD-MA), mathematical modelling and studies about cervical cancer incidence and CIN by age that supported the initiation of screening at 25 years of age rather than at age 20 or 30. Starting at this age is likely to be acceptable to stakeholders, is feasible and needs fewer resources than starting screening at an earlier age. There was very low-certainty evidence from the studies mentioned above (given the small numbers of women followed and reporting cervical cancer or CIN lesions) that found that the risk of cervical cancer and lesions may continue after the age of 50 years. Screening was therefore suggested to continue at regular screening intervals, until there have been two consecutive negative screening results after the age of 50. Conditional recommendations were made for screening intervals based on modelled evidence showing greater benefits may occur with three- to five-year screening intervals with HPV DNA testing (or cytology or VIA), though there may be more treatments and therefore harms compared with a longer interval (low-certainty evidence).

Conditional recommendations were made for HPV DNA testing 12 months after treatment and after a negative triage test, regardless of initial screening test, as there may be greater benefits and fewer harms (low-certainty evidence based on modelling).

4.2.2. Summary of the evidence

For further details on the following, refer to the summaries in Web annex A (Syntheses of the evidence) and Web annex B (EtD tables) – web links provided at the end of the Table of Contents.

Primary screening tests (Recommendations 21, 23 and 24): Moderate-certainty evidence from a randomized controlled trial, longitudinal studies and modelling found that HPV DNA as the primary screening test in screening and treatment services is likely to result in greater reductions in CIN2+ than VIA, but may result in similar reductions to cytology (followed by colposcopy). There may be similar harms (including harms from treatments and preterm deliveries) when using either HPV DNA testing or VIA as the primary screening test, but fewer harms with cytology followed by colposcopy. The costs, however, may be greater when using cytology followed by colposcopy (Web annex A, IARC handbook materials and Supplementary materials 2 and 13).

Screen-and-treat versus screen, triage and treat (Recommendations 22, 23, 28, 29 and 30): Studies measuring outcomes found in the published literature did not directly compare different screen-and-treat algorithms against screen, triage and treat algorithms. The longitudinal studies assessing HPV DNA-based screening did, however, include strategies with HPV DNA alone, and with cytology and/or colposcopy as triage, and found that there are likely to be similar benefits across the strategies (moderate-certainty evidence). Outcomes were also mathematically modelled, showing that there may be similar benefits from using HPV DNA screening with or without triage testing, but that harms may be reduced with triage due to fewer treatments for pre-cancer (Web annex A, Supplementary material 13). When screening is provided every five years, HPV DNA without triage may result in slightly greater harm (including more treatments and preterm deliveries) than with triage; and the harms may be similar across different triage tests. Modelled evidence also found that greater benefits may occur with three- to five-year screening intervals with HPV DNA testing (or cytology or VIA) although there may be more treatments and therefore harms when compared with a longer interval (low-certainty evidence). In addition, HPV DNA testing followed by triage at five-years intervals may have similar benefits and harms when compared with cytology (followed by colposcopy) every three years, but it is more cost-effective than cytology and colposcopy at any interval.

Although data were not available for women living with HIV, and this evidence was not modelled, the sensitivity and specificity of clinically validated PCR-based high-risk HPV DNA testing for the detection of CIN2+ on self-collected upper vaginal samples versus cervical samples taken by health-care providers may be similar for women in the general population (Web annex A, IARC handbook materials and Supplementary material 1) and for women living with HIV (low-certainty evidence). Based on previous modelling in 78 low- and middle-income countries (including some with a high burden of HIV) to evaluate the impact of the WHO Global strategy to accelerate the elimination of cervical cancer (23,24), the GDG agreed that, at a minimum, twice-in-a-lifetime screening would be feasible and provide benefits, because the modelling demonstrated benefits with screening twice in a lifetime compared with once. The GDG thus made a good-practice statement to screen twice.

Age to initiate or stop screening (Recommendations 25, 26 and 27): Low-certainty evidence from an IPD-MA found that there may be a substantially lower risk (6 per 100) of histologically confirmed CIN2 or CIN3 among girls aged between 15 and 19 years compared with older age groups. In young adult women aged 20–24 years, the risk may be 32 per 100, and in women aged 45–49 years, 58 per 100. There is low-certainty evidence from modelling that screening starting at 20 years of age in women living with HIV, using strategies with HPV DNA testing, may lead to slightly more reductions in cervical cancer incidence and deaths compared with starting at 25 or 30 years of age. There may be more harms, however, from substantially more pre-cancer treatments and therefore a greater risk of preterm deliveries. There may also be higher costs since more resources are needed for screening and treatment when started at an earlier age (Web annex A, Supplementary material 13). Starting screening at the age of 20, 25 or 30 is likely to be acceptable to women living with HIV.

Only sparse data were available for cervical cancer and CIN2/3 or HPV incidence in women living with HIV after the age of 50. A large database of cervical cancer cases indicated that of all cases of cervical cancer, 20% were diagnosed in women in their 50s and 1% of them were diagnosed in women in their 70s (Web annex A, Supplementary material 5). This very low-certainty evidence provides indirect support for screening strategies continuing after the age of 50. Literature on the general population of women found that barriers for older women included embarrassment, lack of knowledge about the need for screening (in particular when they have no symptoms) and fear of discomfort (Web annex A, Supplementary material 4).

Acceptability, equity, feasibility, resources (all recommendations): Some studies have identified provider perceptions that may impact the acceptability of recommendations for different screening and treatment approaches (Web annex A, Supplementary material 12). Some providers perceive that screening approaches with HPV DNA detection may increase uptake, lead to more treatment and be more sensitive in detecting lesions, whereas VIA may require more standardized training since it is subjective and dependent on expertise. Providers may also perceive that self-sampling could reduce opportunities to provide the person with other care. Overall, it may be challenging for providers during a transition from cytology or VIA-based screening to HPV DNA testing as new infrastructure (e.g. laboratory) and training are needed. A global online survey of women found that multiple visits may be difficult, particularly for those living with HIV, and immediate treatment may be preferred (Web annex A, Supplementary material 9). A systematic review of the literature, primarily in high-income countries, found that HPV DNA testing by self-sampling is likely to improve screening uptake and may improve follow-up care slightly (Web annex A, Supplementary material 12).

A survey of GDG members identified concerns that were generally more serious when considering VIA- or cytology-based screening programmes compared with HPV DNA-based screening programmes – these concerns related to maintaining a trained workforce, generating demand for screening and ensuring high compliance with management following a positive screening test (Web annex A, Supplementary material 11). There were greater cost concerns with HPV DNA and cytology than with VIA-based programmes. Screen-and-treat algorithms were generally more acceptable to stakeholders involved in ongoing service delivery, while screen, triage and treat algorithms were perceived to be more acceptable to health authorities, policy-makers and experts in cervical cancer. When choosing a triage test, providing cytology or colposcopy would be a challenge where it is not currently available and quality controlled and assured. However, if the HPV DNA primary screening test also gave HPV16/18 genotyping information (see algorithm 4 in Table 2.2), there would be no additional test to perform, making it important for a self-sampling strategy. HPV DNA testing by self-sampling also has the potential to reduce cultural, socioeconomic and gender barriers to screening, which is likely to increase equity.

Follow-up after negative triage test or after treatment (Recommendations 31, 32, 33 and 34): The evidence for follow-up screening after a positive HPV DNA primary screening test and negative triage test was based on modelling of follow-up screening at 12 or 24 months, or both (Web annex A, Supplementary material 13). There is low-certainty evidence showing that after a negative triage test, there are greater benefits when retesting at 12 months compared with 24 months, but similar harms; and similar benefits and harms at 12 months compared with retesting at both 12 and 24 months. Therefore, screening at 12 months is suggested.

The evidence on which screening test to use for follow-up after treatment following a positive screening test or after treatment for histologically confirmed CIN2/3 came from literature comparing cures and failures after treatment and literature on the accuracy of tests to measure failure. It also came from modelling the use of HPV DNA testing at 12 or 24 months follow-up or co-testing with HPV DNA and cytology at 12 months (Web annex A, Supplementary materials 2 and 13). Systematic reviews providing moderate- and low-certainty evidence on treatment failure in women living with HIV suggest it may be around 30% after ablative treatment, and lower after excisional treatment. There is high-certainty evidence from reviews of the accuracy, in the general population, of different tests to predict recurrence or persistence after treatment; this evidence indicates that the HPV DNA test is more sensitive and slightly less specific than cytology alone or co-testing with HPV DNA and cytology. High-certainty evidence also shows that positive margin status of an excised cone was less sensitive than the HPV DNA test as a predictor of failure. There is very low-certainty evidence for the sensitivity and specificity of HPV genotyping due to the few studies and inconsistent results.

Low-certainty evidence from modelling studies showed that retesting women with an HPV DNA test at 12 or 24 months after treatment or with HPV DNA and cytology co-testing at 12 months follow-up may result in similar reductions in cervical cancer and deaths; but using co-testing is likely to increase treatments and costs. For women living with HIV, however, there is low-certainty evidence from two studies that the specificity of HPV DNA testing may be lower than for the general population of women, meaning there is less certainty in the applicability of the results from the general population of women. Additional follow-up visits were therefore suggested by the GDG. Recommendations to extend the interval before follow-up may concern some practitioners working in settings that use shorter durations (e.g. 12 months). There may also be a challenging time for programmes when transitioning from VIA or cytology to HPV DNA testing, when women may be screened with VIA or cytology but retested with HPV DNA testing at follow-up.

4.3.1. Justification and evidence summary

Low-certainty evidence from a systematic review of the literature found that there may be little to no difference in the recurrence rate of AIS with CKC or electrosurgical excision, or in the incidence of complications such as major infection and bleeding, and found that more women may have premature deliveries in subsequent pregnancies following a CKC compared with electrosurgical excision. The studies included in the systematic review did not confirm HIV status, but the GDG agreed that the data could be extrapolated to women living with HIV and applied directly. CKC is performed in the operating theatre, so access to CKC may be limited in some settings, more costly and less preferred by women compared with LLETZ. In addition, greater expertise may be needed for successful electrosurgical excision.