To present the results of the first 2 years of a human papillomavirus (HPV) test-based screening programme outside the research context.
To present the results of the first 2 years of a human papillomavirus (HPV) test-based screening programme outside the research context.
Population-based cohort study.
A cervical service screening programme in Italy.
Women aged 25–64 years invited to screening from April 2009 to April 2011.
Eligible women were invited to undergo an HPV test: those with a negative HPV test went on to the next screening episode; those with a positive HPV went on to triage with a Pap smear. Women with positive cytology (i.e. positive for atypical squamous cells of undetermined significance or worse, ASC-US+) were referred to colposcopy, whereas those with negative cytology were referred to repeat HPV testing 1 year later.
Participation rate, positivity at HPV and at triage, referral rate to colposcopy, positive predictive value for cervical intraepithelial neoplasia grade 2+ (CIN2+) at colposcopy, and detection rate for CIN2+.
Participation increased compared with the previous Pap programme (60.6 versus 43.9%). The HPV positivity rate was 7.0; 39.6% of Pap smears were scored as positive, and therefore 2.8% of the women screened were referred for immediate colposcopy. The compliance of women who scored positive for HPV and negative for Pap for repeat HPV testing at 12 months was 78.6%, and the HPV positivity rate was 56.6%. The overall referral rate to colposcopy was 4.6%. The overall detection rate for CIN2+ was 4.5 versus 1.5% of the Pap programme (25–34 years, 8.2%; 35+ years, 3.6%).
Compared with the traditional Pap test, the HPV programme recorded a higher response to invitation and an increased DR for CIN2+. The most critical aspects were the reading of cytology in women that were positive for HPV and the increased workload at colposcopy.
There is clear evidence that testing for high-risk human papillomavirus (HPV) sequences (hereafter referred to as the HPV test) results in the earlier detection of cervical intraepithelial neoplasia (CIN) grade 3,[1-6] and in the prevention of cervical cancer.
Several randomised clinical trials using this strategy have been carried out, and many regions throughout the world are planning to implement it; however, very few data are available on the results of HPV-based screening programmes outside the research context. Therefore, data are still lacking about some important effects of the introduction of the HPV test in routine (‘real life’) service screening programmes. First, little is known about how women would respond to undergoing a new screening test, and abandoning the tried and tested Pap smear. Second, we do not know to what extent women would comply with the different management derived from the triage strategy, which is necessary to balance the lower specificity of the HPV test. For instance, women with a positive HPV and a negative triage test could be recommended to repeat the tests after 12 or 18 months. It is not known whether these women will be willing to wait without seeking other actions outside the screening programme, nor if they will return for retesting. Third, up until now cytology has represented the gold-standard test for triage, and it will now be analysed by personnel aware of the HPV test result (in contrast to the blind reading performed in most clinical trials). It is not clear whether, and to what extent, this may influence the interpretation of the Pap smears and, consequently, the colposcopy and histology workload. Many of these questions may be explored through pilot programmes that the European guidelines recommend be carried out in organised screening programmes, with careful monitoring of the quality and systematic evaluation of the outcomes.
The main objective of this descriptive study is to evaluate, outside of the research context, the outcomes and impact on workload of an HPV test-based strategy in a service screening programme.
The HPV test-based strategy was introduced in the cervical screening programme of the Local Health Unit of Este, Veneto Region, in the north-east of Italy in April 2009. Local Health Units (LHUs) are the public agencies that organise and administrate the health services of groups of municipalities, including screening programmes. Within the Veneto Region, data collection and quality monitoring of screening programmes are centrally coordinated by the Veneto Tumour Registry.
The organised, Pap-based screening programme of Este started applying the recommendations of the European guidelines in 1998.[8, 9] Since then the target population (women aged 25–64 years) has been invited to perform a Pap smear every 3 years. Compared with the other LHUs of the Veneto Region, the cervical screening programme of Este was characterised by lower levels of positivity in the Pap smear (2.6 versus 3.0% in the Veneto Region in 2008), and by lower levels of detection of high-grade cervical intraepithelial lesions (CIN2+; 1.3 versus 3.4%).
The screening algorithm, described in Figure 1, is in line with the GISCi (Italian Association of Cervical Screening Programmes) guidelines, and with the recommendation of a recent Health Technology Assessment Report on HPV DNA-based primary screening.
Three years after the previous screening episode, eligible women received a letter of invitation to the new round. The invitation letter included a leaflet with some information about HPV, about the HPV test, and about the new screening programme. These tools were specifically developed and pre-tested on the target population through focus groups.
The specimens were taken in the same four sampling units of the previous Pap-based programme by the same four midwives who underwent specific training before implementing the new strategy. A double sampling for conventional smear and HPV testing was performed on all participating women, except for women who had never been sexually active, who underwent the Pap smear alone. The HPV tests were performed at the Immunology and Molecular Oncology Unit of the Veneto Oncology Institute Istituto di Ricerca e Cura a Carattere Scientifico of Padua.
The women with a negative HPV test were informed of the result by letter and referred to the next screening episode.
The Pap smears of the women who tested HPV positive were processed and read by a cytologist, who was aware of the positivity at HPV. Diagnoses were performed by the pathology unit of the Este LHU, where the Pap smears of the previous Pap programme had been read.
Positive Pap smears (atypical squamous cells of undetermined significance, ASC-US, or worse) were reviewed by the same cytologist and two pathologists using a multi-view microscope (the same protocol was applied in the previous Pap programme). Diagnoses were reported according to the 2001 Bethesda System.[12, 13] Women with a diagnosis of ASC-US+ were referred to colposcopy; those with a diagnosis of less than ASC-US were requested to repeat the HPV testing 1 year later. Inadequate Pap smears were repeated.
Colposcopy-directed biopsies were read by two pathologists using a multi-view microscope and the histological findings were classified as negative, CIN1, CIN2, CIN3, or squamous cell carcinoma. Women with a diagnosis of CIN2+ were referred for excisional treatment. Women with a negative colposcopy and/or biopsy were referred to a follow-up colposcopy 6 months later.
Women who tested HPV positive and Pap negative (HPV+/Pap−) were informed of the result by letter. Twelve months later, they were sent a written invitation to repeat the HPV test and, in the case of non-compliance, were prompted by telephone. Again, two samples were taken (for the HPV test and for the Pap smear), and the HPV test was performed. Women who tested HPV negative returned to screening, whereas the women who remained HPV positive were referred to colposcopy, irrespective of the cytology. Their Pap smears were processed and the diagnoses were made available to the gynaecologists who carried out the colposcopies.
Cervical cells were collected for both conventional cytology and the HPV test. During the first year, cells were first collected by spatula and cytobrush for smear preparation, and subsequently by the cervical sampler brush, specific for the Hybrid Capture 2 test (HC2; Qiagen, Hilden, Germany). From the second year onwards, cells were collected by the Qiagen cervical sampler brush only; the brush was first used to prepare a cytological smear and then placed into a tube containing the HC2 transport medium.
Cervical cells were analysed by HC2 with the High Risk probe set, according to the manufacturer's instructions, using the Rapid Capture System (RCS; Qiagen). All samples with a relative light unit/positive control (RLU/PC) ratio ≥1 were considered positive. Samples with an RLU/PC ratio in the range 0.90–0.99 were re-tested. Besides the internal quality controls provided in the assay, random samples selected among those with RLU/PC values ranging between 0.50 and 1.10 were included as controls in subsequent assays. External quality controls were received by different sources and run in duplicate every few months.
To evaluate the workload generated by the HPV-based programme, its impact on the target population, and its diagnostic performance, we calculated the following parameters, both at baseline (including post-colposcopy follow-up) and at the 1-year re-testing in women who were HPV+/Pap−, stratified by age (25–29, 30–34, and over 35 years):
The results of the HPV programme were compared with those of the Pap smear programme in the three previous years by calculating the relative frequencies of the main indicators, with 95% confidence intervals.
We present data on women screened in the first 2 years of the programme up to April 2011, followed up to April 2012, and including the 1-year repeat for cases that were HPV+/Pap− at baseline.
In April 2009 the HPV test was introduced as the sole primary test.
In these first 2 years of the programme, 23 368 women were invited to attend the new screening by HPV testing (Figure 1), and 12 026 were screened (11 895 women underwent HPV testing and 131 women underwent Pap smear alone; crude response rate 51.5%). As 3511 women reported the result of a recent Pap smear performed outside the programme, the overall participation was 60.6%. Participation increased with age from 48.1% in women aged 25–29 years to 64.5% in those aged 60–64 years.
Of the 11 895 HPV-tested women, 9.3% were aged 25–29 years (n = 1102), 10.3% were aged 30–34 years (n = 1221), and 80.5% were aged 35+ years (n = 9572).
|Baseline||1-year repeat after HPV+/Pap− tests|
|Women invited||23 368||501|
|Total women examined (% = crude compliance)||12 026a||51.5||394||78.6|
|Proportion of positive HPV tests|
|Proportion of positive Pap tests among HPV+||328/829||39.6||48/223||21.5|
|Referral rate to immediate colposcopy||328/11 895||2.8||223/394||56.6|
|Compliance to immediate colposcopy||273/319b||85.6||178/223||79.8|
|CIN2+ DR (%)|
|PPV for CIN2+ at first colposcopy||32/273||11.7||11/178||6.2|
|Referral rate to 1-year recall||501/11 895||4.2|
Retesting the 14 samples with ratios in the 0.90–0.99 range (which represented 0.1% of the negative samples) confirmed a negative (<1.0) result in 11 samples (78.6%), whereas three samples (21.4%) gave a positive (>1.0) result (median 1.12; range 1.0–1.3).
External quality control procedures included the testing of eight synthetic and two clinical samples circulated among laboratories to assess reproducibility, as well as the testing of nine purchased samples. The results were always within the expected range.
The overall positivity rate at cytology among women who were HPV+ was 39.6% (328/829). Two women with an inadequate Pap smear repeated the test and the results were negative. Among the positive Pap smears, the most frequent diagnostic category was low-grade squamous intraepithelial lesion (LSIL; 89.3%), followed by high-grade squamous intraepithelial lesion (HSIL; 6.1%, including one case of malignant cells), atypical squamous cells-high-grade not excluded (ASC-H; 2.7%), and atypical glandular cells (AGC; 1.8%). No ASC-US cases were diagnosed.
The Pap smears from the 131 women who had never been sexually active were all negative.
Compliance with colposcopy was 85.6% (273/319). One or more biopsies were taken in 254 second-level colposcopies (93%). In the case of a negative colposcopy inspection, only one biopsy was usually taken.
Overall, one invasive carcinoma, 18 CIN3, and 22 CIN2 were diagnosed (seven and two of the latter, respectively, at post-colposcopy follow-up). Among the 32 CIN2+ lesions diagnosed at the first colposcopy, one had a negative colposcopy and was detected exclusively through blind biopsy.
The DR for CIN2+ was 3.4% (41/11 896) overall, and 3.6% (4/1102), 8.2% (10/1221), and 2.8% (27/9572) in the 25–29, 30–34, and 35+ age groups, respectively (25–34 years, 6.0%).
The PPV for CIN2+ at baseline colposcopy was 11.7% (32/273): 3.4% (2/58) in women aged 25–29 years, 17.9% (10/56) in those aged 30–34 years, and 12.6% (20/159) in those aged over 35 years, respectively.
Overall, 394/501 women underwent repeat HPV testing after 12 months, with a crude attendance rate of 78.6% (77% of women complied to the letter, and a further 1.6% responded to the telephone reminder). Of the 105 women who did not attend, 49 (47%) declared that they had already undergone a Pap smear spontaneously (diagnoses not available).
Overall, 56.6% (223/394) of the women tested HPV+ (25–29 years, 60.5% (52/86); 30–34 years, 59.0% (46/78); over 35 years, 54.3% (125/230) (chi-square test for the trend 1.117; P = 0.29).
We observed a direct relationship between the positivity at repeat HPV and the RLU/PC ratio in the sample taken at baseline, with the former being 37.5% (27/72) of cases with a baseline ratio of 1.00–1.99, 57% (58/102) of cases with a baseline ratio of 2.00–9.99, and 63% (138/220) of cases with a baseline ratio of ≥10 (chi-square test for trend 11.92; P < 0.001).
Among the 223 HPV+ cases, 78.5% had a negative Pap smear (n = 175), 19% had an LSIL (n = 43), 1.3% had an HSIL (n = 3), and 0.4% had an ASC-H (n = 1) or an AGC (n = 1).
The biopsy rate at colposcopy was 89% (158/178): six CIN2 and six CIN3 were found (one CIN3 at follow-up), with a DR for CIN2+ of 30.5% and a PPV at first colposcopy of 6.2% (11/178).
The overall DR of the HPV programme (baseline plus 1-year repeat) was 4.5% (53/11 895). The values observed in the 25–29, 30–34, and over 35-year cohorts were 4.5% (5/1102), 11.5% (14/1221), and 3.6% (34/9572), respectively (25–34 years: 8.2%) (Table 2).
|Pap test programme 2006–2008 (36 months)||HPV programme 2009–2011 (24 months + 1 year follow-up)||Relative frequenciesa (95% CI)|
|Invited women||43 913||23 368|
|Screened women||17 116||11 895 (+131 Pap tests)|
|Response rate (%)||39.0||51.5||1.32 (1.28–1.36)|
|Adjusted participation rate (%)||43.9||60.6||1.40 (1.34–1.42)|
|Proportion of positive tests (%)||(ASC-US+)||(HPV+/cytology+)|
|Overall||2.6 (454/17 611)||2.7 (328/11 895)|
|25–29 years||3.2 (47/1466)||6.0 (67/1102)|
|30–34 years||3.4 (62/1806)||5.3 (65/1221)|
|35–64 years||2.6 (345/14 339)||2.0 (196/9573)|
|Referral rate to colposcopy at baseline (%)||2.6 (454)||2.7 (328)||1.07 (0.92–1.24)|
|Cumulative referral rate to colposcopy (%)||2.6 (454)||4.6 (551)||1.80 (1.58–2.04)|
|Cumulative Detection rate for CIN2+ (%)|
|Overall||1.5 (27/17 611)||4.5 (53/11 895)||2.91 (1.79–4.81)|
|25–29 years||2.0 (3/1466)||4.5 (5/1102)||2.22 (0.43–14.3)|
|30–34 years||1.7 (3/1806)||11.5 (14/1221)||6.90 (1.92–37.5)|
|35–64 years||1.5 (21/14 339)||3.6 (34/9572)||2.43 (1.37–4.40)|
|Positive predictive value for CIN2+ at baseline (%)|
|Overall||7.7 (23/299)||11.7 (32/273)||1.52 (0.84–2.80)|
|25–29 years||9.7 (3/31)||3.4 (2/58)||0.36 (0.29–3.32)|
|30–34 years||7.3 (3/41)||17.9 (10/56)||2.44 (0.58–14.6)|
|35–64 years||7.5 (17/227)||12.6 (20/159)||1.68 (0.81–3.53)|
Overall, at baseline 14% (114/829) of HPV+ tests had an RLU/PC of 1.00–1.99, and 13% (106/829) had an RLU/PC of 2.00–4.99 (see Table 3). The cases with an RLU/PC of 1.00–1.99 accounted for 6.1% of colposcopies, and those with an RLU/PC of 2.00–4.99 accounted for a further 8.0%. No CIN2+ cases were diagnosed in the first group, four cases of CIN2 and two of CIN3 (both at the 1-year follow-up) were diagnosed in the second group, whereas the great majority (47/53, 89%) of CIN2+ lesions were diagnosed among women with an RLU/PC >10.
|RLU/PC||25–29 years, n (%)||30–34 years, n (%)||35–65 years, n (%)||All ages n (%)|
|1.00–1.99||17 (10%)||0||17 (10%)||0||80 (16%)||0||114 (14%)||0|
|2.00–4.99||19 (11%)||1 (17%)||22 (14%)||2 (15%)||65 (13%)||2 (6%)||106 (13%)||5 (9%)|
|5.00–9.99||13 (7%)||0||19 (12%)||0||46 (10%)||1 (3%)||78 (9%)||1 (2%)|
|≥10||127 (72%)||5 (83%)||103 (64%)||11 (85%)||301 (61%)||31 (91%)||531 (64%)||47 (89%)|
Table 2 compares the results of the HPV programme with the previous 3 years of the Pap programme. Participation increased by 40% from 43.9 to 60.6% (relative rate 1.40, 95% CI 1.34–1.42). Higher values were observed in all age classes.
The overall referral rate to immediate colposcopy at baseline was similar to the value recorded by the Pap programme (2.7 versus 2.6%), but increased by 80–4.6%, when also considering the colposcopies from the 1-year repeat.
The DR for CIN2+ of the HPV programme was almost three times higher than that of the Pap programme (relative DR 2.91, 95% CI 1.79–4.81). The largest gap was observed in the 30–34-year-old cohort, which showed a seven-fold increase, from 1.7% in the Pap programme to 11.5% in the HPV programme (25–34 years: relative DR 4.46, 95% CI 1.71–13.7).
The PPV for CIN2+ at baseline colposcopy increased from 7.7% of the Pap programme to 11.7% with HPV (relative rate 1.52, 95% CI 0.84–2.80).
This article describes an audit of the first 2 years of an organised HPV-based screening programme implemented into routine practice, outside of the research setting, compared with a previous Pap-based programme.
One of the most striking results is the increase in the uptake of this new strategy for cervical screening (60.6 versus 43.9%). The economic crisis could possibly explain this result: in an area where the uptake of private cervical smear tests is widespread, the economic shortage could be associated with an increased recourse to the free test offered by the organised screening programme, independently of the type of test. However, we did not notice a similar trend in the screening programmes of the other LHUs in the Veneto Region during the study period.
The triage by cytology of the HPV+ cases appears to be a major aspect. The positivity rates of the triage Pap tests were higher than those reported by the New Technologies for Cervical Cancer (NTCC) trial (38.2% in women younger than 35 years and 25.3% in women aged 35–60 years),[16, 17] where the Pap tests were read without knowledge of the HPV result. This is probably related to the cytologists' awareness of the HPV+ result. In our programme no ASC-US diagnoses were made among women who were HPV+, unlike the NTCC study, where ASC-US represented 14.2 and 11.2% of positive smears in women who were HPV+ aged 25–34 and over 35 years, respectively. This result arose from a local choice to shift from a morphological interpretation of the Pap smear towards a risk-based one: women with carcinogenic HPV+ ASC-US or with LSIL have the same 10% probability of being diagnosed with CIN3.
No CIN2+ cases were recorded in women with an RLU/PC ratio of 1.00–1.99, and an inverse relationship between the RLU/PC ratio and clearance after 1 year was observed. A higher specificity for CIN2+ with HC2 positivity cut-offs higher than an RLU/PC ratio of 1.00 has been reported by many authors, and our results confirm these findings.[16, 17, 19-22] Indeed, as positivity in the 1.00–1.99 range accounted for 14% of the further workup, a 2.00 cut-off value for RLU/PC would sensibly decrease referral to both immediate colposcopy and the 1-year repeat.
The 1-year repeat for HPV+/Pap− women contributed to 23% of the CIN2+ detected overall (12/53), confirming the importance of a high compliance with this step of the screening algorithm.
The overall DR of the HPV programme (baseline plus 1-year repeat) was 4.5%, a figure three times higher than that recorded by the Pap-based strategy. The trials that compared cytology and HPV as a primary test obtained similar cross-sectional results, and attributed them to diagnostic anticipation.[2, 4, 6, 16, 17] Our results could be related to the higher prevalence of disease in the screened population, e.g. if the increased response to invitation derived from the participation of a proportion of women who had been infrequently screened. However, it is unlikely that this could explain the threefold increase in the DR.
The assessment protocol of the previous Pap programme was modified, according to the findings of Pretorius,[23, 24] with the implementation of biopsies in negative colposcopies; however, this practice did not influence the overall detection rates of the HPV programme, because only a minimal (1/32) proportion of the lesions were detected through a blind biopsy alone.
The larger increase in relative detection rates versus the Pap programme took place in women aged 25–34 years. In particular we observed the highest diagnostic excess, which randomised trials partially attributed to over-diagnosis, in the cohort of 30–34 year olds. This is a matter of concern as a CIN treatment history is associated with an increased risk of pregnancy-related morbidity.[25-28]
This is one of the first reports on the implementation of primary HPV screening in a service programme, outside the research context.
The use of a historical non-randomised cohort as comparison implies some limitations (e.g. some of the results could be explained by confounding factors), but allows a better understanding of everyday activities. As the HPV programme was carried out by the same operative units and the same personnel in the same population as before, it is plausible that most of the observed differences can be attributable to the only change that was introduced, i.e. the screening algorithm.
The HPV programme was aimed at a population that had been frequently screened by Pap smear. On one hand, this implies that our results cannot be transferable to the whole population, even if according to Progressi delle Aziende Sanitarie per le Salute in Italia, a national survey about health-related lifestyle and behaviour, the proportion of women that declare that they have undergone a Pap smear in the previous 3 years is around 82% in the Este LHU. On the other hand, our results may somehow predict what could be expected in many European countries where Pap-based organised screening programmes continue to operate.
Our data suggest some strategies that could be used to deal with the over-diagnosis and the increased workload for colposcopists that represent the two major drawbacks of our HPV-based programme:
A recent pooled analysis on HPV testing in the secondary prevention of cervical cancer reported substantial evidence to support screening by HPV starting at 30 years of age. In this case, younger women would not incur pregnancy complications, but would not obtain any benefit either.
Our results confirm the evidence attained in research settings and give some insight into issues not addressed by clinical trials. As there are major clinical implications for millions of women worldwide, if screening programmes move from Pap smear to primary HPV testing, it is desirable that the results of other similar experiences are made available.
All authors declare that they have no conflicts of interest.
MZ, ADM, and CF drafted the article. CF, HFS, and LB assembled the data and performed the statistical analysis. AFa, LDB, AB, PC, MG, NM, and MGP ran the screening centres and collected the data. MZ, ADM, AFa, LDB, AB, CC, and AFe interpreted the data. All authors read and approved the final article. MZ is the guarantor.
A law of the Veneto Region declared that the Pap test could be replaced by HPV testing as the primary screening option within organised programmes using the nationally approved protocol; therefore, ethical approval was not necessary.
This project was supported by the Foundation Cassa di Risparmio di Padova e Rovigo.
We thank Rossana Trevisan for excellent technical assistance.
The authors report the impact of introducing HPV testing as a primary screening test within a northern region of Italy. Somewhat interestingly there was a dramatic increase in the proportion of women accessing cervical screening (by 72%), from 43.9% when screening was performed by cytology to 60.6%. The paper records the results of implementing HPV testing as routine practice, and not in the context of a research study. The dramatic rise in women attending screening after the introduction of HPV testing is an interesting finding. Cervical screening coverage rates are variable across western countries and have declined over the last 8 years, particularly in young women. In the UK, coverage rates have been on the order of 80%; however, in Italy coverage has never been above 50% in young women (Lancucki LJ Med Screen 2010;17:91–96).
A systematic review of studies evaluating women's views on HPV testing undertaken by Hendry and colleagues (Br J Cancer 2012;107:243–254) confirmed that women have generally negative views on HPV testing, and were particularly concerned about the issues of having a sexually transmitted disease. There have been a number of studies evaluating women's knowledge of HPV-related disease from all regions of Europe, USA, and beyond. In all studies, a significant number of women were not aware of the role of HPV in cervical cancer. Although cultural differences might play a role in the uptake of HPV primary cervical screening, it remains important to have effective health education programmes, not only to improve secondary prevention by screening but also primary prevention by vaccination.
Pooled data from five randomised controlled trials (RCTs) comparing primary HPV testing with conventional cytology demonstrate the significantly improved detection of high-grade disease in the first screening round, and a greater detection of disease was found in this implementation study, in keeping with research studies (Murphy J J Obstet Gynaecol Can 2012;34:443–452). The RCTs report a reduction of high-grade disease by the next round of screening, demonstrating the benefit of the improved detection of disease in the first round. A trade-off for the improved detection of disease is the increased colposcopy workload generated by a primary HPV test strategy, particularly in younger women. In the future, other biomarkers might be used after HPV testing to improve the specificity of screening. HPV testing alone appears to be more sensitive, but less specific, compared with cytology. Developing such biomarkers would reduce the number of women without disease being unnecessarily referred to secondary care for colposcopy.
Lancashire Teaching Hospital, Preston, PR2 9HT UK