National experience in the first two years of primary human papillomavirus (HPV) cervical screening in an HPV vaccinated population in Australia: observational study.

OBJECTIVE: To review the first two years of the primary human papillomavirus (HPV) cervical screening programme in an HPV vaccinated population.
DESIGN: Observational study.
SETTING: Australia. PARTICIPANTS: 3 745 318 women with a primary HPV test between 1 December 2017 and 31 December 2019; most women aged <40 years had previously been offered vaccination against HPV16 and HPV18.
INTERVENTIONS: Primary HPV screening with referral if HPV16 or HPV18 (HPV16/18) positive and triage with liquid based cytology testing (threshold atypical squamous cells-cannot exclude high grade squamous intraepithelial lesion) for women who were positive for high risk HPV types other than 16/18. A 12 month follow-up HPV test was recommended in triaged women with a negative or low grade cytology result, with referral if they tested positive for any high risk HPV type at follow-up. MAIN OUTCOME MEASURES: Proportion of women who had attended for their first HPV screening test, tested positive, and were referred for colposcopy; and short term risk of detecting cervical intraepithelial neoplasia (CIN) grade 2 or worse, CIN grade 3 or worse, or cancer.
RESULTS: 54.6% (n=3 507 281) of an estimated 6 428 677 eligible women aged 25-69 had undergone their first HPV test by the end of 2019. Among those attending for routine screening, positivity for HPV16/18 and for HPV types not 16/18 was, respectively, 2.0% and 6.6% in women aged 25-69 (n=3 045 844) and 2.2% and 13.3% in highly vaccinated cohorts of women aged 25-34 (n=768 362). Colposcopy referral (ages 25-69 years) was 3.5%, increasing to an estimated 6.2% after accounting for women who had not yet had a 12 month repeat test. Cervical cancer was detected in 0.98% (456/46 330) of women positive for HPV16/18 at baseline, including 0.32% (89/28 003) of women with HPV16/18 and negative cytology. Women with HPV types not 16/18 and negative or low grade cytology at both baseline and 12 months were at low risk of serious disease (3.4% CIN grade 3 or worse; 0.02% cancer; n=20 019) but estimated to account for 62.0% of referrals for this screening algorithm.
CONCLUSIONS: Colposcopy referral thresholds need to consider underlying cancer risk; on this basis, women with HPV16/18 in the first round of HPV screening were found to be at higher risk regardless of cytology result, even in a previously well screened population. Women with HPV types not 16/18 and negative or low grade cytology showed a low risk of serious abnormalities but constitute most referrals and could be managed safely with two rounds of repeat HPV testing rather than one. HPV16/18 driven referrals were low in HPV vaccinated cohorts. © Author(s) (or their employer(s)) 2019. Re-use permitted under CC BY-NC. No commercial re-use. See rights and permissions. Published by BMJ.

Introduction

Strong evidence supports the use of testing for oncogenic human papillomavirus (HPV) types as a primary cervical screening test (primary HPV screening).1 2 3 4 5 6 Consequently, many countries are undergoing or planning a transition to primary HPV screening. Australia was one of the first countries to completely transition its national cervical screening programme to primary HPV testing, and the first country where this included cohorts of young women who had received HPV vaccination as part of the national vaccination programme. HPV vaccination was introduced to Australia in 2007, with a catch-up programme for women aged up to 26 until the end of 2009, and inclusion of boys from 2013. On 1 December 2017, the entire country switched from cytology (cervical or Pap smear) tests every two years for women aged 18-69, to primary HPV testing every five years with HPV16/18 genotyping and liquid based cytology for women with HPV positive screening results, starting at age 25, with an exit test between the ages of 70 and 74. Previous modelling studies have shown that improving screening is key to achieving reductions in cervical cancer in the short term and expediting the elimination of cervical cancer.7 8 9 10 HPV screening is a key part of Australia’s strategy to achieve the World Health Organization threshold for elimination of cervical cancer. Women attending for routine screening are referred for colposcopy (classified by the programme as higher risk) if HPV16/18 is detected or both an HPV type other than 16 or 18 (not 16/18) is detected and the triage liquid based cytology result shows a possible high grade lesion (equivalent to atypical squamous cells-cannot exclude high grade squamous intraepithelial lesion; ASC-H) or worse or glandular abnormalities. For women aged 70-74 attending for an exit test, colposcopy is recommended for any who are HPV positive. Thus, at the time the national cervical screening programme transitioned, women aged up to around 36 had been offered HPV vaccination, and women aged 25-29 had been offered the vaccine as teenagers (detailed coverage estimates at end of 2017 have been published and range from 57.8% in women aged 25 to 23.2% in women aged 3611). Another novel aspect of the renewed national cervical screening programme was the option of HPV testing on a self-collected sample for women aged 30 or older who had never undergone screening or whose screening was overdue by two or more years.

Changes to the national cervical screening programme were based on an extensive evidence review and modelled effectiveness and economic evaluation of many potential screening strategies to identify an optimal approach, including for both unvaccinated cohorts and those offered HPV vaccination,12 13 14 and supported by the findings of Compass, the first phase of a randomised trial of primary HPV screening in Australia.15 Nevertheless, it was recognised that evidence on the optimal management of HPV positive women was evolving. One of the specific areas of uncertainty was appropriate management of women with low grade cytological abnormalities and HPV types not 16/18.16 The consideration at the time was whether these women should be immediately referred to colposcopy; but based on available evidence, women with HPV types not 16/18 and negative or low grade liquid based cytology results on their baseline test were referred for repeat HPV testing at 12 months (classified by the programme as intermediate risk).17 18 Initial management guidelines recommended that if HPV infection was detected at this 12 month follow-up test, the women would then be classified as higher risk and therefore referred for colposcopy.16

As women in routine screening were due a first HPV test two years after their last cytology test, all eligible women were due to attend by 30 November 2019. Therefore, the first two years after the programme transitioned is a useful time point to review and document Australia’s experience, and to estimate how many women had attended for HPV screening (and conversely, how many women were overdue for screening). Modelling had predicted a substantial transient increase in demand for colposcopy, but larger increases have been reported in some jurisdictions,19 20 and data were not yet available to investigate the proportion of referrals coming from different clinical pathways (primary screening, triage, surveillance tests, or investigation of symptoms). Sentinel data from the Compass trial and programme indicated a low prevalence of HPV16/18 positive results at screening, regardless of age, due to the impact of first generation HPV16/18 vaccines, but high positivity for HPV types not 16/18 in young women.15 21 22. Preliminary programme data suggested that a large proportion (65.6%) of women with HPV types not 16/18 detected at baseline had infections due to non-16/18 types detected again at their 12 month follow-up test. Consequently, they were referred for colposcopy under the management recommendations current at the time, but most (92.6%) of these women still had negative or low grade liquid based cytology results.21 Data from the first phase of Compass indicated that the risk of cervical intraepithelial neoplasia (CIN) grade 2 or more severe diagnoses was lower (5.7%) in the women who were referred at 12 months, but sample sizes precluded stratification by liquid based cytology result at 12 months.15 Limited data are available on the utilisation of self-collection in the new programme, although it appears that self-collection was under-utilised for a range of reasons, including difficulties in ascertaining whether or not women were eligible at the point of care, unexpected regulatory hurdles leading to delays, and a lack of confidence among providers.20 23 24

We reviewed the real world data and outcomes from the first round of screening in the first two years of the primary HPV screening programme in Australia, and calculate the risk of detecting CIN grade 2 or worse, CIN grade 3 or worse, and cervical cancer among different groups of women referred for colposcopy in the first round of HPV screening in a national programme. These results have informed post-implementation adjustments to clinical management guidelines for HPV positive women.

Methods

Data source, attendance, and test results

Data were sourced from Australia’s National Cancer Screening Register. Laboratories are required to report all test results to the national register, including HPV tests, liquid based cytology, and histology, and are required to use tests that separately identify HPV16 and HPV18. Data were extracted from the register for all women with at least one HPV test, conducted between 1 December 2017 and 31 December 2019 inclusive, including all subsequent HPV tests, cytology, and colposcopy or histology in these women. Colposcopy and histology tests were included if they were conducted up until 16 April 2020 and recorded in the register on 22 September 2020. The proportion of women eligible for screening who attended for their first HPV test took into account women enrolled in Compass (as the register excludes data on women participating in Compass; about 75 944 were enrolled as of 31 December 2019)25 26 and the estimated proportion of women who had undergone hysterectomy, by excluding both groups from the denominator in line with standard methods.21 Age specific results for the proportion of women eligible for screening who had attended for their first HPV test by the end of 2019 are based on a woman’s age on 31 December 2019; all other age specific results are reported based on a woman’s age at the time of testing. The National Cancer Screening Register does not yet contain reliable data on vaccination status, so this could not be considered at an individual level, but some results are reported stratified based on whether women had or had not previously been age eligible for HPV vaccination in Australia (age <40 v ≥40 years).

Colposcopy referrals and risk in HPV positive women

Figure 1 shows how HPV positive women were managed. Reflex cytology was done for all HPV positive women, at baseline screening and at 12 month testing. Cytology results did not influence management for women with HPV16/18 detected at baseline or with any HPV type detected at 12 month follow-up, but they are used to inform colposcopy, and in this study were used as stratification information to define groups for estimating the risk of detecting high grade abnormalities and cancer (CIN grade 2 or worse, CIN grade 3 or worse, cancer).

This analysis included women with a positive HPV primary screening result. Women with HPV types not 16/18 detected at their baseline screening test but without a corresponding satisfactory liquid based cytology result were excluded from the analysis, as their risk level could not be determined. We examined histological outcomes and short term risks of detecting CIN grade 2 or worse, CIN grade 3 or worse, and cancer among women who attended for colposcopy, stratified by their baseline test results (for women referred on the basis of their baseline test result—that is, classified as higher risk at baseline) or by their 12 month follow-up test result (for women referred on the basis of their 12 month follow-up test—that is, classified as intermediate risk at baseline but higher risk at 12 months). These were summarised into specific pathways of women referred based on the results of their primary test (positive for HPV16/18), triage test (positive for HPV types not 16/18 with ASC-H or worse or glandular abnormalities detected by liquid based cytology), or 12 month test (stratified into HPV16/18 positive or ASC-H or worse or glandular abnormalities detected by liquid based cytology versus HPV types not 16/18 positive with low grade squamous intraepithelial lesions or less serious liquid based cytology result). Women were excluded from the analyses of histological outcomes and risk if there was an indication that a biopsy sample was taken at colposcopy, but no histology result was available. In sensitivity analyses this was further restricted to women with histology within six or 12 months of their referral. Women who attended colposcopy but with no indication that a histology sample was taken were assumed to have had a normal colposcopy result and no biopsy taken, and therefore were assumed to not have CIN grade 2 or worse. Risk was also calculated by age group. Pre-test and post-test probabilities of disease were calculated and presented based on previous methods.27 28 Findings for women aged 70-74 are reported separately, as clinical management guidelines differ for women attending for exit testing (women with any oncogenic HPV type detected are referred for colposcopy).

Patient and public involvement

This analysis was undertaken in part to explore drivers of colposcopy referrals and risk in different groups of people referred for colposcopy, because of observed increases in colposcopy wait times that directly affect patients, and to inform a review of the clinical management guidelines. The guidelines development process includes a period of public consultation on drafted updates. Patients and members of the public were not directly involved in the design of this research but had the opportunity to provide feedback on the analysis, conclusions, and proposed changes to the guidelines during the public consultation process for the draft updates to the guidelines in late 2020.

Results

Attendance, reason for test, test results

Overall, 3 745 318 women had a cervical screening test for any reason between 1 December 2017 and 31 December 2019 inclusive (fig 1 and supplementary fig A2). This number with a test for any reason included 3 507 281 women aged 25-69 years, and 134 660 women in the age range for an exit test (70-74 years; see supplementary table A1). These numbers represent 54.6% of the estimated 6 428 677 women aged 25-69 eligible for screening (ranging from 39.4% of eligible women aged 25-29 to 61.3% of eligible women aged 55-59), and 35.1% of the eligible women in the age range for an exit test. In total, 4522 women had HPV tests on a self-collected sample (0.1% of women attending primary screening; see supplementary table A2). In this initial phase of the new programme, among women age eligible for screening (25-74 years), about 13.3% (479 442/3 614 175) of cervical screening tests were non-screening tests, most commonly for symptoms, compared with 65.8% of cervical screening tests in women younger than 25 and 54.6% in women older than 75 (table 1, and supplementary table A3).

Fig 1

Cohort description (results and management for women aged 25-69). Low grade=cytology results of atypical squamous cells of undetermined significance or low grade squamous intraepithelial lesion. High grade=cytology results of atypical squamous cells-cannot exclude high grade squamous intraepithelial lesion, high grade intraepithelial lesion, or more severe abnormality or any glandular abnormality. Non-screening reasons include investigation of signs and symptoms or follow-up after a previous abnormality (including test of cure after treatment). Follow-up not completed = women who had not yet attended for their 12 month follow-up test (including that 12 months had not yet elapsed), or were followed up outside guidelines. Assessable=women with a record that colposcopy occurred after the referring human papillomavirus (HPV) test but excluding women with an indication that histology was collected but the histology result was not available. Women with HPV16/18 detected and unsatisfactory liquid based cytology result are classified as higher risk and recommended to have a sample for liquid based cytology collected at colposcopy. Risk level cannot be determined for women with HPV types not 16/18 detected and unsatisfactory liquid based cytology result; guidelines recommend that these women reattend for a repeat liquid based cytology test in six weeks. In the current analysis, women with HPV types not 16/18 detected are only categorised as having an unsatisfactory liquid based cytology result if no subsequent satisfactory result was available on the National Cervical Screening Register and had occurred by 31 December 2019 (restricted to tests where the reason was reflex liquid based cytology after detection of oncogenic HPV in primary screening)

Table 1

Number of women with at least one cervical screening test for human papillomavirus (HPV) between 1 December 2017 and 31 December 2019, by age and reason for test recorded on Australia’s National Cancer Screening Register

Women with an HPV test (any reason)*	Age (years) at first HPV test
	<25†	25-29‡	30-34	35-39	40-44	45-49	50-54	55-59	60-64	65-69	70-74	≥75	Total
Primary screening HPV test	40 136	367 006	401 356	389 396	359 002	373 665	330 981	323 100	279 059	222 279	88 889	6194	3 181 063
Non-screening	77 370	69 362	72 186	69 488	63 468	62 763	48 494	37 841	26 304	18 839	10 697	7443	564 255
Follow-up HPV test§	12 881	15 843	11 652	8592	7099	6643	4871	3508	2359	1679	625	196	75 948
Co-test:
Test of cure	2369	11 952	22 721	25 315	23 078	21 188	14 558	10 957	7189	4327	1729	442	145 825
Investigation of symptoms	41 600	33 755	28 796	26 741	24 837	25 730	21 201	15 994	10 312	7506	5377	4989	246 838
Other¶	2183	3439	3809	3603	3302	3620	3057	2925	2573	2197	1216	641	32 565
Other**	18 337	4365	5194	5222	5144	5570	4795	4449	3864	3123	1748	1175	62 986
Not reported	-	8	14	15	8	12	12	8	7	7	2	-	93
Total	117 506	436 368	473 542	458 884	422 470	436 428	379 475	360 941	305 363	241 118	99 586	13 637	3 745 318

Reason for test recorded on register as specified by healthcare provider.

Based on reason for first test in women with more than one test in time period.

Aged <24 years and nine months at first HPV test.

Aged ≥24 years and nine months at first HPV test.

After previous abnormality.

Performed for reason indicated in guidelines that was not included in earlier categories.

Tests that do not fit into any other category. See also supplementary table A3.

Among 3 045 844 women aged 25-69 attending for primary screening, 2.0% were positive for HPV16/18 and 6.6% were positive for HPV types not 16/18. In 88 889 women aged 70-74 attending for primary screening (exit test), 1.6% were positive for HPV16/18 and 2.7% were positive for HPV types not 16/18. HPV16/18 positivity was low in all age groups, ranging from 1.6% in women aged 65-69 and 70-74 to 2.5% in women aged 30-34 (see supplementary tables A4 and A5). Positivity for HPV types not 16/18 was much higher, particularly in younger women, and generally declined with increasing age. In non-screening tests, HPV positivity patterns by age were like those for screening tests, but HPV positivity was much higher (see supplementary tables A6 and A7 and figure A3).

Colposcopy referrals and timeliness

Figure 2 presents the proportion of women in the routine screening pathway classified as higher risk (colposcopy referral recommended) over the first two years, stratified by the specific pathway by which they were referred, and age. Overall, 3.5% of the 3 045 844 women aged 25-69 attending for routine HPV screening were classified as higher risk by the end of 2019; 2.4% based on their baseline test (ranging from 1.7% in women aged 65-69 to 3.2% in women aged 30-34) and 1.2% at 12 months (see supplementary tables A8 and A9). An additional 3812 women aged 70-74 were classified as higher risk after an exit test (any HPV detected; 4.3% of screened women in that age group; see supplementary tables A4 and A5). Final referral rates for women referred at 12 months were not available at the time of the analysis, as follow-up testing is ongoing for the 187 623 women aged 25-69 who were classified as intermediate risk at baseline and recommended to reattend in 12 months (6.2% of those screened aged 25-69, ranging from 2.6% in women aged 65-69 to 16.3% in women aged 25-29). At the time of data extraction, 29.7% of the women classified as intermediate risk at baseline had a record of a 12 month follow-up test (for many women, 12 months had not yet elapsed). Among the 55 739 women who had returned, 35 079 (62.9%) were HPV positive and referred for colposcopy, mostly (90.0%) because of detection of an HPV type not 16/18 in conjunction with a negative or low grade liquid based cytology result. When test results and referrals among women initially classified as intermediate risk who had returned were extrapolated to women who had not yet returned for their follow-up test, referrals at 12 months would be expected to increase from 1.2% to 3.9%; overall referral rates would be expected to increase from 3.5% to 6.2%; and the group of women classified as intermediate risk at baseline would ultimately be expected to generate 62.0% of all referrals in women aged 25-69. Referral rates were higher in younger women (25-39 years) who were age eligible for HPV vaccination than in older women (40-69 years); however, the difference was driven by higher positivity for HPV types not 16/18 (not targeted by the first generation HPV vaccines) (fig 2). Among all women recommended to attend for colposcopy (n=156 939, considering HPV tests performed for screening and those for other reasons), 70.4% were referred on the basis of a screening test (including 12 month follow-up of some women initially classified as intermediate risk, but not scaled to take into account women who have not yet had their 12 month test) versus 29.6% on the basis of a non-screening test (see supplementary tables A12 and A13).

Fig 2

Proportion of women classified as higher risk (colposcopy referral recommended), with and without scaling to account for women who had not yet had their 12 month repeat test, by age group and reason for referral. Solid bars represent referrals observed so far; transparent bars and values in parentheses include scaling to account for women who had not yet had their 12 month repeat test. Results are restricted to women aged 25-69, as different management is recommended for women aged 70-74 undergoing exit testing. HPV16/18+=positive for human papillomavirus (HPV) types 16/18 at baseline screening test. Triage+=positive for HPV types not 16/18 with atypical squamous cells-cannot exclude high grade squamous intraepithelial lesion (ASC-H) or worse or glandular abnormalities on liquid based cytology. 12 month follow-up (HPV16/18+ or HPV types not 16/18 and ASC-H+)=12 month test result is HPV positive, and either positive for HPV16/18, or liquid based cytology result is ASC-H or worse or any glandular abnormality, or both. 12 month follow-up (HPV types not 16/18 and ≤LSIL (low grade squamous intraepithelial lesion))=12 month test result is HPV types not 16/18 positive and liquid based cytology result is negative or ASC-US (atypical squamous cells of undetermined significance) or LSIL. ASC-H+=ASC-H, high grade squamous intraepithelial lesion, or more severe abnormality or any glandular abnormality, or both. See supplementary tables A10 and A11 for data supporting this figure

Overall (considering women of any age), 86.9% of women had a record of colposcopy within six months of their referring test (when restricting to women with at least six months of follow-up data available; see supplementary tables A12 and A13). The proportion who attended colposcopy within six months was higher among those referred at baseline (91.4%; just over 95% in women with ASC-H or worse or with glandular abnormalities on liquid based cytology) compared with those referred at follow-up (84.6%). The proportion who attended colposcopy within six months was lowest for those with the reason for the test recorded as signs or symptoms (72.1%), although some women in this group might have been directly referred to an oncology specialist.

Histological outcomes

Overall, 78 223 women aged 25-69 had a record of colposcopy, including complete histology information when a biopsy sample was taken (table 2). Among these women, the overall detection rates for CIN grade 2 or worse, CIN grade 3 or worse, and cancer were 22.1% (n=17 324), 14.7% (n=11 499), and 0.7% (n=538), respectively. Risk of detecting serious disease was higher in women referred at baseline compared with those referred on the basis of a 12 month follow-up test result: more than twice as high for CIN grade 2 or worse, more than three times higher for CIN grade 3 or worse, and nearly 20 times higher for cancer. Risks for CIN grade 2 or worse and CIN grade 3 or worse were highest among women referred because of their triage test (ie, those with ASC-H or worse or glandular abnormalities on liquid based cytology; CIN grade 2 or worse: 59.3% (n=5510) for ages 25-69; CIN grade 3 or worse 39.3% (n=3650) for ages 25-69; fig 3 and supplementary tables A17 and A18). In contrast, cancer detection was highest in women referred because of their primary test (HPV16/18 positive at baseline). Corresponding risks for women of all ages were broadly similar (see supplementary tables A14 and A15). In total (including women of all ages), 546 cancers were identified through routine screening (354 squamous cell carcinomas, 188 other (non-squamous cell) cancers, and four mixed squamous and non-squamous cell carcinomas; see supplementary table A16). Of these, 535 (98.0%) were in women referred at baseline; after accounting for women who had not yet returned for their 12 month test, this would be expected to reduce to 93.5% of an estimated 572 cancers in total.

Table 2

Histological outcomes in women aged 25-69 years attending for routine human papillomavirus (HPV) screening by baseline and follow-up result. Values are numbers (percentages) unless stated otherwise

	No of women	Women with completed follow-up data*	Follow-up histology result
			CIN grade 2 or worse	CIN grade 3 or worse	Cancer
Higher risk (referred) at baseline	72 364	55 628 (76.9)	14 686 (26.4)	10 246 (18.4)	527 (0.95)
HPV16/18 detected:	60 786	46 330 (76.2)	9176 (19.8)	6 596 (14.2)	456 (0.98)
LBC ASC-H or worse or glandular abnormalities	9274	7583 (81.8)	5275 (69.6)	4232 (55.8)	330 (4.35)
LBC ASC-US or LSIL	13 121	9821 (74.8)	1691 (17.2)	957 (9.7)	26 (0.26)
LBC negative	37 173	28 003 (75.3)	2117 (7.6)	1345 (4.8)	89 (0.32)
LBC unsatisfactory	1218	923 (75.8)	93 (10.1)	62 (6.7)	11 (1.19)
HPV not 16/18 detected:
LBC ASC-H or worse or glandular abnormalities	11 578	9298 (80.3)	5510 (59.3)	3650 (39.3)	71 (0.76)
Higher risk (referred) at 12 months (intermediate risk at baseline)	35 079	22 595 (64.4)	2638 (11.7)	1253 (5.5)	11 (0.05)
HPV16/18 detected at follow-up test:	1122	776 (69.2)	77 (9.9)	37 (4.8)	1 (0.13)
LBC ASC-H or worse or glandular abnormalities	95	73 (76.8)	37 (50.7)	25 (34.2)	1 (1.37)
LBC ASC-US or LSIL	376	261 (69.4)	27 (10.3)	9 (3.4)	0 (0.00)
LBC negative	637	435 (68.3)	13 (3.0)	3 (0.7)	0 (0.00)
LBC unsatisfactory	14	7 (50.0)	(0.0)	(0.0)	0 (0.00)
HPV not 16/18 detected at follow-up test:	33 957	21 819 (64.3)	2561 (11.7)	1216 (5.6)	10 (0.05)
LBC ASC-H or worse or glandular abnormalities	2009	1554 (77.4)	873 (56.2)	531 (34.2)	6 (0.39)
LBC ASC-US or LSIL	10 826	6991 (64.6)	883 (12.6)	320 (4.6)	1 (0.01)
LBC negative	20 736	13 028 (62.8)	791 (6.1)	355 (2.7)	3 (0.02)
LBC unsatisfactory	386	246 (63.7)	14 (5.7)	10 (4.1)	0 (0.00)
Total	107 443	78 223 (72.8)	17 324 (22.1)	11 499 (14.7)	538 (0.69)

CIN=cervical intraepithelial neoplasia; LBC=liquid based cytology; ASC-H or worse or glandular abnormalities =atypical squamous cells-cannot exclude high grade squamous intraepithelial lesion (ASC-H), high grade intraepithelial lesion (HSIL), or more severe abnormality or any glandular abnormality. ASC-US=atypical squamous cells of undetermined significance; LSIL=low grade squamous intraepithelial lesion.

Excludes women with HPV test performed for reason unrelated to routine screening (eg, investigation of symptoms, post-treatment follow-up, at a colposcopy visit) and those with tests performed on self-collected sample. Histological outcomes up to 16 April 2020. Excludes women aged 70-74, as different management applies to this group (any HPV positive is referred at baseline screening).

Restricted to women with record of colposcopy or histology, or both; excludes women with record of a histology sample being collected where histology results are not available. See also supplementary tables A14 and A15 for results across all ages and supplementary table A16 for more detailed histological outcomes.

Fig 3

Risk of detecting histological cervical intraepithelial neoplasia (CIN) grade 2 or worse, CIN grade 3 or worse, cancer, and frankly invasive cancer among women attending for colposcopy, by reason for referral and age group. Results are restricted to women aged 25-69, as different management is recommended for women aged 70-74 undergoing exit testing. HPV16/18+=positive for human papillomavirus (HPV) types 16/18 at baseline screening test. Triage+=positive for HPV types not 16/18 with atypical squamous cells-cannot exclude high grade squamous intraepithelial lesion (ASC-H) or worse or glandular liquid based cytology result. 12 month follow-up (HPV16/18+ or HPV not 16/18 and ASC-H+)=12 month test result is HPV positive, and either positive for HPV16/18 or ASC-H or worse or glandular abnormality on liquid based cytology, or both. 12 month follow-up (HPV not 16/18 and ≤LSIL (low grade squamous intraepithelial lesion))=12 month test result is HPV types not 16/18 positive and liquid based cytology result is negative or ASC-US (atypical squamous cells of undetermined significance) or LSIL. ASC-H+=liquid based cytology result of ASC-H, high grade squamous intraepithelial lesion or more severe abnormality or any glandular abnormality. See supplementary tables A17 and A18 for data supporting this figure

Among women with HPV16/18 detected at baseline, risk of detecting CIN grade 2 or worse and CIN grade 3 or worse increased with increasing severity of the cytological abnormality, especially for high grade cytology; however, this pattern differed for cancer (table 2). Although women with high grade cytology still had the highest risk (4.4%; 330/7583), women with negative cytology had a relatively high risk of underlying invasive cancer (0.32%; 89/28 003), and this was similar to the risk in women with low grade cytology (0.26%; 26/9821; P=0.41). Most of the cancers identified in women with HPV16/18 detected and negative cytology were non-squamous cell carcinomas (68.5% ages 26-69; 67.8% all ages; 61 women) and almost a third of the 188 non-squamous cell carcinomas detected in the period of this analysis were in women with HPV16/18 detected and negative cytology. A substantial proportion of the adenocarcinoma in situ detected was in women with HPV16/18 detected and negative cytology (29.6% ages 25-69; 29.3% all ages; 227 women; see supplementary table A16). Among women referred at 12 months, the risk of detecting serious disease was high in those who had high grade cytological abnormalities (33.5% for CIN grade 3 or worse and 0.4% for cancer), but low in women with other test results. Among women with HPV types not 16/18 detected at both baseline and 12 months, and with negative or low grade cytology on both occasions (n=20 019), the risk was 3.4% for CIN grade 3 or worse and 0.02% for cancer. Overall, the risks for CIN grade 3 or worse were comparable for women referred with HPV16/18 (n=47 106) and women referred with HPV types not 16/18 (n=31 117; 14.1% and 15.6%, respectively); however, cancer risk was higher in those referred with HPV16/18 (1.0% v 0.3% for the HPV types not 16/18 pathway; P<0.001). Most cancers (84.9% ages 25-69; 84.6% all ages) and virtually all non-squamous cell carcinomas (97.3% ages 25-69; 97.3% all ages) were in women who tested positive for HPV16/18. However, these results were biased by having more follow-up data available for the HPV16/18 referral pathway than the HPV types not 16/18 referral pathway (as not all women had attended for a 12 month follow-up test). If results are extrapolated to account for women initially testing positive for HPV types not 16/18 who had not yet returned for their 12 month follow-up HPV test, an estimated 37 total (26 additional) cancers would be expected at follow-up (three in women testing positive for HPV16/18 and 34 in women testing positive for HPV types not 16/18). This extrapolation reduces the proportion of cancers detected in women who test positive for HPV16/18, from 84.9% to an estimated 81.6%. Risks associated with specific combinations of test results remained similar if the analysis was restricted to women with a follow-up histology or colposcopy result within six or 12 months of their referring tests (data not shown).

Risk by age

Figure 3 shows the risks of detecting CIN grade 2 or worse, CIN grade 3 or worse, and cancer by age group and referral pathway (see supplementary tables A17-A19 for more detailed histological outcomes and risks associated with combinations of test results by age group). Risks of CIN grade 2 or worse and CIN grade 3 or worse tended to decrease with increasing age, regardless of the referral pathway, whereas cancer risk was highest in women aged 30-39 and 40-49 and lower in other age groups. In all age groups, the risk of CIN grade 2 or worse and CIN grade 3 or worse was highest for women referred because of their triage test result (that is, with high grade or worse or glandular abnormalities on liquid based cytology), and relatively little variation in risk was found by age among women younger than 50. In contrast, cancer risk was highest in women referred because of their primary screening test result (that is, HPV16/18 positive at baseline) in some age groups (25-29, 30-39, and 60-69 years). Risks of detecting CIN grade 3 or worse and cancer were low in all age groups for women who were positive for HPV types not 16/18 at both baseline and 12 months, but who had no high grade or glandular abnormalities. Among women referred based on their primary test result (HPV16/18 positive), risk of detecting cancer tended to be higher in women aged 30-49, although 95% confidence intervals overlapped for all age groups.

We undertook an additional analysis of women initially classified as intermediate risk who continued to be positive for HPV types not 16/18 and had either negative or low grade liquid based cytology results at 12 months, stratified on age (<50 v ≥50 years) to guide an update to the national cervical screening programme guidelines (see supplementary table A19). Risks for detecting CIN grade 2 or worse, CIN grade 3 or worse, and cancer were 10.0%, 4.0%, and 0.02%, respectively, among women younger than 50 (n=16 236), compared with 3.4%, 1.5%, and 0.00%, respectively, for women aged 50 or older (n=4968; P<0.001, P<0.001, P=0.27, respectively).

Utility of tests for risk stratification

Figure 4 shows how each test (HPV positivity, HPV16/18 genotyping, liquid based cytology) contributes to stratifying risk of detecting CIN grade 3 or worse and cancer at each point in the screening pathway (and supplementary figure A4 shows the pre-test and post-test probability plots stratified for women younger than 50 versus 50 or older). In this first round of HPV screening, ASC-H or worse results or glandular abnormalities on liquid based cytology in HPV positive women were consistently a strong predictor of risk for CIN grade 3 or worse and cancer, at baseline and 12 months. Among women with ASC-H or worse/glandular liquid based cytology results, CIN grade 3 or worse, and cancer were more likely to be detected in women positive for HPV16/18 than those with only HPV types not HPV16/18, although risk of CIN grade 3 or worse (but not cancer) was comparable regardless of the HPV type in women with ASC-H or worse or with glandular abnormalities on liquid based cytology at their 12 month follow-up test. Positivity for HPV16/18 at baseline had strong utility for identifying risk of underlying invasive cancer (0.98% in women aged 25-69, 0.96% all ages; table 2, supplementary table A15), and in those positive for HPV16/18, relatively little difference in cancer risk was found between low grade versus negative liquid based cytology. Based on these results, some groups of women classified as higher risk in the original guidelines (women who are HPV positive at 12 months with negative or low grade liquid based cytology) had risks for both CIN grade 3 or worse and cancer that were more comparable with women classified at baseline as intermediate risk than those classified as higher risk, suggesting another round of repeat testing may be appropriate, rather than colposcopy referral.

Fig 4

Pre-test and post-test probability (%) of cervical intraepithelial neoplasia grade 3 or worse, cancer, and frankly invasive cancer. Results are for women of all ages. High grade+=atypical squamous cells-cannot exclude high grade squamous intraepithelial lesion (ASC-H), high grade squamous intraepithelial lesion (HSIL), or more severe abnormality or any glandular abnormality. Low grade=atypical squamous cells of undetermined significance (ASC-US) or low grade squamous intraepithelial lesion (LSIL); LBC=liquid based cytology. See supplementary tables A20-A22 for data supporting this figure

Discussion

Because the previous cytology based national cervical screening programme in Australia recommended screening every two years, the first round of screening for HPV involved women having their first HPV test within two years of the transition date. By the end of 2019, all women eligible for cervical screening would have been recommended to attend, but only slightly more than half had done so. Although this finding is broadly consistent with two year participation in the pre-renewed programme,29 it highlights substantial room for improvement, especially in women aged 25-29, where slightly fewer than 40% had attended for their first HPV test. It is anticipated that the proportion of women who are up to date with screening (had an HPV test in the previous five years) will naturally increase over the first five years of the screening programme, as the proportion of women who attended for screening in a five year period before the transition has been reported at around 80%.30 New pro-active invitations for women to reattend for screening (rather than only reminders for overdue appointments) had also not been fully implemented in 2018-19, and so the mix of screening frequencies among women attending in this period is likely to be similar to what it was in the cytology programme. These results for attendance include those who were screened on a self-collected sample (only available during this period to women aged 30 and older who were overdue for screening by two or more years), and so self-collection has not yet achieved its full potential in increasing participation, although there are a range of reasons why utilisation was initially low.20 31 The universal option to use self-collection from mid-2022 is anticipated to remove several current barriers and further increase participation.32 33 These results for attendance include women with an HPV test for any purpose and are not restricted to those whose test was for routine primary screening. This was done to capture all women engaged in the programme in some way (including active follow-up as well as primary screening), and for comparability with how participation had been measured before transitioning to HPV screening29 (a recent monitoring report for Australia changed to a different methodology restricted to primary screening only and reported 46.3% of women having a primary screening test in 2018-1934). No directly comparable data exist from the previous cytology programme for test positivity, colposcopy referrals, or risk or positive predictive value associated with routine screening or surveillance, as neither the reason for the test nor colposcopy data were recorded on the separate jurisdictional registers that preceded the National Cancer Screening Register.

During the first two years of the new programme, a moderate proportion of women aged 25-39 in Australia in this period had previously been vaccinated against HPV types 6, 11, 16, and 18, and consistent with earlier data, HPV16/18 infections were relatively low (and consistent across all age groups). Importantly, however, among young women who were age eligible for vaccination but nonetheless had HPV16/18 detected (possibly from an infection acquired before HPV vaccination), risk for serious disease was at least as high as, and sometimes higher than, in older, likely unvaccinated, women who were positive for HPV16/18. Across all ages, those with HPV16/18 detected at baseline screening had an important risk of cancer, at least in this prevalent round of screening, including in the absence of cytological abnormalities.

An increase in colposcopy referrals was anticipated in the first few years of the primary HPV programme.19 Our results indicate a substantial minority of these (31%) are due to non-screening tests. Among those due to screening, referrals are mostly from two groups: women with HPV16/18 and negative liquid based cytology results, and women with HPV types not 16/18 and negative or low grade liquid based cytology results at both baseline and 12 months, with the latter category predicted to eventually constitute more than half of all referrals from screening. These two groups of women differ in their risk of serious disease, however, particularly for cancer (0.32% in women with HPV16/18 but negative liquid based cytology v 0.02% for HPV types not 16/18 and negative or low grade liquid based cytology results).

Strengths and limitations of this study

Our analysis has several strengths. The study made use of comprehensive national data and therefore reports on a large number of women. Australia was one of the first countries to transition its national screening programme to HPV screening (after the Netherlands and Turkey), and it is the first to do so in the context of a partially HPV vaccinated screening population. Australia uses a different clinical management algorithm from these countries and others where pilots are underway (which generally immediately refer HPV positive women with a cytological abnormality, regardless of HPV type or level of cytological abnormality).35 36 37 38 The Australian programme makes use of HPV16/18 genotyping and deliberately uses different colposcopy referral thresholds for women positive for HPV16/18 (referral regardless of liquid based cytology result) and HPV types not 16/18 (referral at ASC-H or worse or, originally, if still HPV positive at a 12 month follow-up test). This combination of HPV vaccination and screening algorithm is a strength of this analysis that has provided important insights, including that while HPV16/18 positivity rates are lower in young women as a result of vaccination, the risk of disease in young women who are HPV16/18 positive is comparable to or higher than that in older women.

Our analysis has some limitations. Firstly, as in any snapshot of data, women will be at different points in the follow-up continuum, and not all women in our initial cohort had yet attended for their recommended 12 month follow-up HPV test or colposcopy. We took this into account in our estimates for referral rates by presenting both results to date and estimates that scaled-up initial results to account for women who have not yet returned for their 12 month test (generally because 12 months had not yet elapsed). Our estimates of histological outcomes will be partially affected by women who have been referred but not yet attended for colposcopy. Women referred after their 12 month follow-up HPV test who did not have high grade or glandular cytological abnormalities were less likely to have attended than women who did, so aggregated estimates of risk across all women with HPV types not 16/18 will potentially be overestimated, since the group of women with follow-up has some bias towards women at higher risk. Conversely, it is also possible that women at higher risk of underlying disease (eg, under-screened women) might be less likely to have attended for colposcopy, and this may mean our aggregated estimates of risk are potentially underestimated. In both cases, however, this is unlikely to affect our estimates of risk within a group of women with the same test results, but the absolute number of women with serious disease outcomes may eventually be higher (noting that the absolute number is primarily of interest within Australia, whereas the risk estimates are likely of broader interest). Secondly, it was not possible to determine if women who were positive for an HPV type not 16/18 at both their initial screening test and their 12 month repeat test had the same genotype detected on both occasions. This is because in Australia’s national cervical screening programme, HPV testing is most commonly performed with tests that do not provide information on the specific genotype detected unless it is HPV16 or HPV18, and additionally because the screening programme does not routinely receive or record detailed genotyping information in the minority of women where this is available. Therefore, some women with non-16/18 HPV types detected at both initial screening and 12 month follow-up may have been infected with different HPV types at each visit, rather than being persistently infected with the same non-16/18 type. Consequently, our estimates of risk in women with HPV types not 16/18 at both baseline and 12 months do not directly reflect and potentially underestimate the risk in women known to be infected with the same non-16/18 HPV type.39 Additionally, the risk associated with persistent non-16/18 HPV infection is known to vary by specific genotype.40 41 Nevertheless, our findings reflect the current reality that in many cases where HPV screening is undertaken, detailed genotyping information may not be available to directly inform clinical management, especially for HPV types not 16/18. Thirdly, there may have been under-reporting of histology results to the National Cancer Screening Register owing to the newly established reporting process. This was dealt with by excluding women from the analysis of disease risk if a biopsy sample had been taken (which can be detected through colposcopy visit forms and histology reimbursement claims) but no histology result was available, to avoid systematically underestimating risk.

Policy implications for HPV screening and screening an HPV vaccinated population

By using partial genotyping for HPV16/18 and differential management of HPV16/18 versus non-16/18 oncogenic HPV types, the renewed screening programme in Australia was designed to operate effectively in the population with mixed and dynamically changing vaccination status and concomitant cancers risks, as those cohorts offered vaccination became eligible for, and progress through, the cervical screening programme. First generation vaccination with quadrivalent vaccine was associated with a reduction in HPV16/18 infections in younger women; however, some women in the catch-up vaccination cohorts were HPV16/18 positive either because of exposure before vaccination or because catch-up coverage was lower than that achieved through school based vaccination of younger cohorts.42 Conversely, there remained a relatively high prevalence of infections with non-16/18 oncogenic HPV types in young women, but in this study these infections were associated with lower risk.35 43 44 The screening programme was designed to deal with these two issues by differentially managing HPV positive women based on the HPV type detected at the time of screening. This includes categorising women with HPV16/18 infections as higher risk and managing them appropriately according to their risk (regardless of whether or not they had been offered, or had, vaccination, and without the need to have information on individual women’s vaccination status) and managing non-16/18 oncogenic HPV types less aggressively than HPV16/18. The findings of our study support this differential management by genotype group, including a cytology referral threshold that is relatively lower for HPV16/18 and relatively higher for HPV types not 16/18 compared with that used in many other settings.35 36 37 38 Our findings for the risk of serious disease, and in particular cancer, in women who are HPV16/18 positive could also help to inform possible prioritisation strategies in the context of self-collection, where no liquid based cytology result will be available to assist in prioritisation.

This combination of vaccination and screening algorithm also highlighted that in populations vaccinated against HPV16/18, referral rates are largely driven by women with non-16/18 HPV types but no cytological evidence of high grade abnormalities, even when a repeat HPV test at 12 months is used to try and reduce over-referral. Our findings indicate that this group of women is relatively large and that their risk of serious disease is low, and so Australia’s approach of deferring referral of women with non-16/18 HPV types but no evidence of high grade changes is not only safe, it is conservative to refer all of those who remain HPV positive at 12 months. Although Australia does not have an explicit risk threshold for colposcopy referral versus 12 month follow-up, we have previously estimated that some women referred for a 12 month repeat test in the cytology programme (women with low grade squamous intraepithelial lesion cytology and negative cytology in the previous two years) had a 24 month risk of CIN grade 3 or worse of around 5-8%, so this had implicitly been seen as acceptably low.14 Based on the analysis reported here, the national clinical management guidelines for the Australia’s cervical screening programme have now been modified such that women with non-16/18 HPV infections who do not have high grade cytological abnormalities at initial screening or 12 months are now referred for repeat surveillance at 24 months; this change came into effect on 1 February 2021. Based on the precautionary principle, there are some exceptions to this change, including women who are overdue for screening by two or more years, and also women aged 50 or older, owing to potentially higher risk of harbouring a covert abnormality in the endocervical canal compared with younger women who are HPV positive.

These findings reflect a first round of HPV screening, and it is acknowledged that risk of detecting serious disease will potentially be lower in in all groups attending for a second round of testing, after prevalent disease has been detected and treated in the first round.36 45 Management recommendations for women who test HPV positive will need to continue to be reviewed and evolve over time, not only because of changing disease risk in the population but also due to new technology and new information becoming available, and to changing workforce issues such as availability of trained cytologists.

We found that 6.6% of all HPV tests and 8.0% of referrals for colposcopy were recorded as being due to symptoms (and the reason for referral was unclear for a further 1.7% of tests and 1.1% of colposcopy referrals). Investigating symptoms is one of the few reasons in Australia where co-testing or any testing for women younger than 25 is recommended and reimbursed; potentially a low threshold was used for symptoms because of a lack of trust in changes to the screening programme or a lack of understanding of the symptoms that are indicative of cervical cancer.20 24 46 Strengthened communications were provided to clarify this, and the need for clearer information and support for providers around management of symptoms (and more broadly) has been identified as a lesson from Australia’s experience.20 47

Conclusions

Australia was the first country to introduce a publicly funded HPV vaccination programme and to reconfigure its screening programme to explicitly consider both vaccinated and unvaccinated cohorts. Consequently, the Australian experience provides insight that is 5-10 years ahead of other countries with vaccination programmes that began later, had more limited catch-up, or lower coverage. We have shown that primary HPV screening with partial genotyping for HPV16/18 from age 25 years is viable in a vaccinated population, but attention should be paid to how HPV types not 16/18 are managed. The detection of prevalent invasive cervical cancer in the first round of primary HPV screening is relatively high. These findings are of crucial importance to other countries as the screen eligible population increasingly overlaps with HPV vaccinated cohorts in many settings.

Primary HPV screening is more effective than cytology for the prevention of cervical cancer

Australia was one of the first countries to make the transition from cytology (cervical smear or Pap testing) to primary HPV screening, and the first to observe the effects of HPV screening in an HPV vaccinated population

Debate continues about the optimal triage for HPV positive women

The findings of this study suggest that colposcopy should be considered for women whose first primary test is HPV16/18 positive, regardless of cytology result, owing to a high risk of detecting cancer in this group (1%), even in a previously well screened population

In a population with substantial uptake of HPV vaccination in young women, referral rates of HPV16/18 positive women are relatively low (2% of those screened), even when screening starts at age 25 (rather than 30)

It also suggests that in the absence of high grade or glandular cytological abnormalities, women with only HPV types not 16/18 detected can safely be referred for surveillance beyond 12 months (eg, by repeat HPV testing at 12 months and again at 24 months)