Acceleration of cervical cancer diagnosis with human papillomavirus testing below age 30: Observational study

Several international cervical screening guidelines advise against using high‐risk human papillomavirus (HR‐HPV) testing in women younger than 30. The rationale for this in young women, lies in the potential for additional detection of both low‐grade and high‐grade cervical intraepithelial neoplasia (CIN) leading to unnecessary treatments without reducing the burden of cervical cancer. We studied 56 544 women screened at 24 to 29 with HR‐HPV testing and 116 858 screened with liquid‐based cytology (LBC) in the English HPV screening pilot. They were compared to 528 460 women screened at the age of 30 to 49. We studied the detection of cervical cancer and CIN2/3 across two consecutive screening rounds 3 years apart. At 24 to 29, a positive HR‐HPV test detected more cases of cervical cancer in the prevalence round than did a positive LBC test (1.36/1000 screened vs 0.82/1000, ORadj: 1.61, 95% CI: 1.18‐2.19). In women with a negative HR‐HPV test, cervical cancer was diagnosed before or at the incidence round in 0.07/1000. After a negative LBC test, cancer detection reached 0.47/1000 and 40% of these cases were diagnosed at FIGO stage IB+. HR‐HPV testing increased the detection of CIN2/3 diagnoses in two consecutive rounds combined by 30% (71.9/1000 vs 55.2/1000). The patterns of detection of cervical cancer and CIN2/3 were almost identical at older ages. These data support using HR‐HPV testing for screening of women younger than 30, which not only accelerates the diagnosis of cervical cancer but leads to a similar relative increase in CIN2/3 diagnosis to that found in women aged 30 to 49.


| INTRODUCTION
High-risk human papillomavirus (HR-HPV) testing is more sensitive than cytology for cervical screening. 1 This test allows extended screening intervals, 2 and is more rational in a post-vaccination era. Its use, however, remains controversial for women younger than 30. In the United States, the American Cancer Society endorsed stand-alone HR-HPV testing for these women in late 2020 following encouraging findings from an unpublished modelling study 3 ; The American College of Obstetricians and Gynecologists accepted HR-HPV testing for young women in 2021 but explicitly stated that cytology remains the preferred test 4 ; while the US Preventive Services Taskforce continues to advise against HR-HPV testing before the age of 30. 5 In Europe, HR-HPV testing is used for screening of women younger than 30 in countries like England, Scotland and Wales. 6,7 In other organised screening programmes that offer screening to women younger than 30, for example, in Italy, 8 Denmark 9 and most of Sweden, 10 HR-HPV testing remains reserved for women older than 30. The latter is supported by the guidelines published by the European Commission. 11 This controversy stems from the lack of high-quality observational data for the age group, particularly in terms of cancer development. 12 So far, most of the evidence supporting the use of HR-HPV testing has been derived from randomised controlled trials that reported outcomes across two consecutive screening rounds. 1 These trials included $94 000 women tested for HR-HPV and $81 000 women tested with cytology, but very few of those were younger than 30. Within this context, some authoritative voices have argued that HR-HPV positive but cytologically negative high-grade cervical intraepithelial neoplastic lesions (CIN2/3) are unlikely to progress to cancer before age 30, so their detection could be delayed until women turn 30. 5,10,11,13 The perceived absence of clinical benefit has been compounded by the concern about the high proportions of women with positive HR-HPV tests undergoing colposcopy and the high likelihood of CIN "overdiagnosis." The latter has been based on the findings from a single randomised trial, where HR-HPV testing at age 25 to 29 detected almost four times as many CIN2 and more than twice as many CIN2/3 in two consecutive rounds than did (conventional) cytology. 14 It is, however, unclear whether these findings are representative of other screening programmes, as the trial showed the highest relative detection with HR-HPV testing at any age out of all trials, that is, its cytology missed the highest numbers of CIN2+. 15 We used data from a large English screening pilot to study the outcomes of HR-HPV testing and liquid-based cytology (LBC) across two consecutive screening rounds in 173 402 women younger than 30. 16 These women were compared to 528 460 women aged 30 to 49, for whom HR-HPV testing has been universally endorsed.

| The English HPV pilot
The pilot has been previously described in detail. [16][17][18] Briefly, six National Health Service (NHS) LBC screening laboratories in England using ThinPrep (Hologic) and SurePath (BD) systems converted about a third of their operations to HR-HPV testing using cobas 4800 (Roche), APTIMA (Hologic) and RealTime (Abbott) HR-HPV assays. These HR-HPV assays target the following genotypes: 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68. Each laboratory used both LBC and HR-HPV testing for primary screening, and selected administrative areas for the conversion to HR-HPV testing in a non-randomised process to ensure consistency of clinical recommendations for each health care provider. Generally, the allocation of screening tests was preserved over two consecutive rounds. Some LBC areas, however, converted to HR-HPV testing earlier than planned to resolve national backlogs owing to an increasing shortage of cytology workforce. 19,20 Women aged 25 to 49 were invited for screening every 3 years. They were managed as described in Figure 1. The pilot followed the English Cervical Screening Programme's (CSP) screening and colposcopy quality assurance guidelines.

| Data sources
Data on screening tests, colposcopies and the associated screendetected diagnoses in 2013 to 2019 were retrieved from the laboratories' information systems. Data on cervical cancer diagnoses and the registered CIN3 diagnoses in 1995 to 2018 were retrieved from the English National Cancer Registration and Analysis Service (NCRAS). 21 English NHS numbers were used for linkage.

| Statistical analysis
Women were categorised according to their age (24 to 49 years) at screening in the pilot's first (prevalence) round. We included women aged 24 because the first CSP screening invitation is sent at 24.5 years. Women older than 49 years were not included as their routine recall interval is 5 years.
We included screening episodes in the pilot's prevalence round in 2013 to 2015, and any subsequent (incidence) episodes that started by the end of 2018 after negative prevalence-round screening tests. High-grade cytology, any cytology worse than LSIL. At the 12-month early recall, three out of six laboratories used HR-HPV 16/18 genotyping as an additional triage test for women with persistently HR-HPV positive and cytologically negative tests. As described in an earlier report, 17 this had a very small effect on the overall number of colposcopies and detected high-grade cervical intraepithelial lesions at any age In the absence of reliable information on the reason for taking a sample, and to exclude women whose first pilot samples were most likely made for follow-up of a recent abnormality, we excluded women if their first pilot sample was preceded by any NCRAS cervical cancer record or by an NCRAS CIN3 record in the previous 3 years, any LBC or HR-HPV test in the past 2 years or the baseline sample's management code reported that it was taken as part of follow-up or at colposcopy. All records for each woman were ranked by their dates to determine the sequence of primary, early recall and diagnostic events within each episode.
We determined the age-specific detection of squamous or glandular cervical cancer and CIN2/3. Throughout this analysis, information on CIN diagnoses made during the pilot was based on the laboratory data, while the source of data on cervical cancers was NCRAS. Cancers were stratified into FIGO stages IA vs IB+ vs unknown. 22 The detection was observed for LBC and HR-HPV testing separately after a positive test in the prevalence round, and, for all women with a negative test in the prevalence round, either before the incidence round (where cervical cancers are usually diagnosed following symptoms or other clinical concerns) or after a positive test at incidence-round screening 3 years later.
We could not estimate the cumulative detection of lesions across the two screening rounds using traditional approaches such as, for example, by summing the counts from each round. 14 This was because 29% of the women attending the incidence round after a prevalence-round LBC screen were tested for HR-HPV owing to the early conversion to HR-HPV testing in some LBC pilot areas (see above). Including incidence-round lesions from these women in the counts would overestimate the cumulative detection for LBC and make HR-HPV testing T A B L E 1 Observed numbers of CIN2, CIN3 and cervical cancer diagnoses in the prevalence and incidence screening rounds, by age in the prevalence round and the type of the screening test Combining cancer cases in women with a negative screening test in the prevalence round who were diagnosed before the incidence round and women with a negative screening test in the prevalence round who were diagnosed after a positive screening test in the incidence round, after accounting for attendance at the incidence round.
is a standard English postcode-based measure of deprivation and laboratory site. Analyses were undertaken using R v3.6.1.

| RESULTS
HR-HPV testing was used in the prevalence round for 56 544 women aged 24 to 29, and LBC for 116 858 (  (Table 3). About one in four of these HR-HPV positive cancers initially presented with negative triage cytology and was diagnosed at early recall with persistently positive HR-HPV tests, the majority at 12 months (Table 1). After a negative HR-HPV test, 0.07/1000 women were diagnosed with cancer before (n = 2) or at the incidence round (n = 1); of these three cases, two were diagnosed at stage IB and none were diagnosed at stage II+. Across both rounds combined, 1.43 cancers were detected per 1000 women, with 0.35/1000 diagnosed at stage IB+ ( Table 2).
The detection of cancer across two rounds decreased with age both after positive and negative screening tests (Tables 1 and 2 Tables 1-3), (b) a lower risk of cancer after a negative HR-HPV test than after negative LBC (Tables 1-3) and (c) a similar cumulative detection across two screening rounds for both screening tests ( Table 2). The proportion of women who had a cancer diagnosed at stage IB+ after negative LBC was approximately the same below and above age 30 (range: from 0.14 to 0.19/1000, Table 2).

| Detection of CIN2/3
As expected, the detection of CIN2/3 was higher in the prevalence than in the incidence round and was higher in younger than in older women in both rounds (Table 1). In all age groups, the cumulative detection of CIN2/3 across two screening rounds 3 years apart, remained higher with HR-HPV testing (Table 4). At 24 to 29, HR-HPV testing detected 9.9/1000 or 50% more CIN2, and 6.9/1000 or 19% more CIN3 than LBC screening (30% more when CIN2 and CIN3 were combined). The increases in cumulative CIN2/3 detection achieved with HR-HPV testing were similar at older ages, 40% at 30 to 39 and 24% at 40 to 49. Adjustment for deprivation and laboratory site hardly changed any of the age-specific ORs comparing HR-HPV testing with LBC (Table 3).

| Two LBC screening rounds vs one HR-HPV round in women younger than 30
Among 60 996 women screened with LBC at 25, 106 374 primary screening tests were made across two screening rounds (Table 5).
These two rounds resulted in 9247 positive tests and colposcopy

| Sensitivity analyses
With the 3-year routine recall interval, women aged 24 to 26 were typically invited for an incidence round before they turned 30, while those aged 27 to 29 were reinvited after they turned 30. Both age groups, however, showed similar patterns in terms of the excess detection of high-grade CIN and cervical cancer with HR-HPV testing compared to LBC (Tables S1-S3). Likewise, the results remained virtually the same when only laboratories with HPV DNA testing were included in the analysis (Tables S4-S6; the two mRNA laboratories were too small to provide reliable age-specific results).
Finally, we repeated the analyses to determine the cumulative frequencies of women with positive screening tests and colposcopy referrals (Tables S7-S9). As with all other outcomes, these were highest at younger ages, but the proportional increases after HR-HPV testing compared to LBC were larger at older ages. This pattern did not change when we included only women with at most CIN1 diagnosed at colposcopy. The PPV for either CIN2+ or CIN3+ were higher at younger ages regardless of the screening test.

| DISCUSSION
This large observational study undertaken within the context of the quality-assured English CSP, has shown that the balance between the clinical benefits and the harms of HR-HPV testing in women younger than 30 is more favourable than has been considered so far. Women aged 24 to 29 were more likely diagnosed with cervical cancer after negative LBC than older women, while HR-HPV testing resulted in fewer delayed diagnoses which emerged in the years after negative LBC. A single round of HR-HPV testing roughly replacing two rounds of LBC under the age of 30, as planned within the English CSP, would increase colposcopy referrals from 15.1% to 20.3% of screened women (+34%), and the detection of CIN2/3 from 6.6% to 7.7% (+18%). These relative increases are substantially lower than estimated in the literature (eg, more than doubling in the cumulative detection of CIN2/3 diagnoses 14 ), and are comparable to those in older women.
Our data demonstrate that HR-HPV testing at 24 to 29 provides a clinical benefit with respect to an earlier diagnosis of cancer. In LBCnegative women, 40% of cancers were diagnosed at stage IB+. These require more aggressive treatment than stage IA cancers and are also more likely to be fatal. [25][26][27][28] After two rounds of screening the cumulative frequency of IB+ cancer was similar for the two tests, however, the diagnoses after a positive HR-HPV test were made before women developed symptoms and with a lead time of up to 3 years, allows for an earlier start of treatment. Additionally, we observed a few cases of stage II+ cancers diagnosed in LBC negative women but none in HR-HPV negative women. Although based on small numbers, this may be suggestive of interval progression to more advanced FIGO stages in LBC negative cancers before age 30.
Furthermore, much like at older ages a negative HR-HPV test at a young age was associated with a very low 3-year risk of CIN3 and cervical cancer. With HR-HPV testing, these lesions were diagnosed predominantly in the prevalence round, whereas with LBC they were more uniformly distributed across the two rounds. This confirms the safety of extended HR-HPV screening intervals for all participants including those who undergo screening irregularly. In the pilot, about a quarter of women screened at 25 did not attend the next round for 5 years or longer.
The increases in colposcopy referrals and high-grade CIN diagnoses at a young age are frequently discussed harms of HR-HPV testing. 5,11 Although our data showed that HR-HPV testing increases the relative detection of CIN2 more than it increases the relative detection of CIN3, these harms are overestimated when the opportunity to extend screening intervals is not taken into account. Conservative criteria for a referral of HR-HPV positive women with negative cytology 29 and careful selection for treatment after a diagnosis of CIN2 3 would further diminish the harms of HR-HPV screening under the age of 30. In the pilot and within the English CSP, colposcopy referral required evidence of viral persistence for at least 12 months after screening; histologically confirmed CIN2/3 are usually treated, 30 although CIN2 is sometimes also managed conservatively. 31 Nevertheless, even after accounting for longer screening intervals, our data still suggested a doubling of the numbers of women with a positive screening test under the age of 30. This may carry a risk of adverse psychological consequences. 32 The strength of our data is their size, incorporating more than twice as many women screened with HR-HPV testing as all randomised trials combined, 1 with detailed screening information and a linkage to the national cancer register where diagnostic FIGO stages were known in more than 90% of the cases. The pilot was embedded within the national CSP and was subject to the same quality assurance as routine cervical screening. A limitation was that the allocation of screening tests was not randomised. Small 16 differences in age and area deprivation between women undergoing HR-HPV testing and those screened with LBC had no practical consequences for the comparison of screening outcomes (Table 3), although some degree of residual confounding could not be excluded. Finally, the analysis included data from the first two rounds of screening with HR-HPV testing. This is a short period compared to the time it takes for CIN to progress to cancer. 33,34 It is therefore not surprising that we could not yet detect a reduction in the frequency of cancer diagnoses derived from detecting more HR-HPV-positive but cytology-negative CIN2/3 in any of the studied age groups. As in randomised trials, 1 it is likely that this reduction will only be measurable after several more years.
The national HPV vaccination programme, which was initiated in England in 2008 with greater than 80% coverage, 35 will exert a highly significant impact on the risk of cervical cancer, which will affect the efficiency of screening and will require changes to the programme if its cost-effectiveness is to be maintained. Although most screening programmes have not yet revised screening recommendations for vaccinated women, cost-effective measures would include longer screening intervals and a later starting age (potentially at age 30 or later). 36,37 In conclusion, data from the English pilot provide evidence that the increased sensitivity of HR-HPV testing in women younger than 30 promotes earlier diagnosis of cervical cancer without creating a large strain on colposcopy services. This new evidence calls for a reassessment of screening guidelines that recommend withholding HR-HPV testing from young women in favour of continued screening with LBC.