POBASCAM, a population-based randomized controlled trial for implementation of high-risk HPV testing in cervical screening: Design, methods and baseline data of 44,102 women
Version of Record online: 9 FEB 2004
Copyright © 2004 Wiley-Liss, Inc.
International Journal of Cancer
Volume 110, Issue 1, pages 94–101, 20 May 2004
How to Cite
Bulkmans, N. W.J., Rozendaal, L., Snijders, P. J.F., Voorhorst, F. J., Boeke, A. J. P., Zandwijken, G. R.J., van Kemenade, F. J., Verheijen, R. H.M., v Groningen, K., Boon, M. E., Keuning, H. J.F., van Ballegooijen, M., van den Brule, A. J.C. and Meijer, C. J.L.M. (2004), POBASCAM, a population-based randomized controlled trial for implementation of high-risk HPV testing in cervical screening: Design, methods and baseline data of 44,102 women. Int. J. Cancer, 110: 94–101. doi: 10.1002/ijc.20076
- Issue online: 25 MAR 2004
- Version of Record online: 9 FEB 2004
- Manuscript Accepted: 7 OCT 2003
- Manuscript Revised: 11 AUG 2003
- Manuscript Received: 18 MAR 2003
- ZON. Grant Number: 30-05220
- human papillomavirus;
- cervical intraepithelial neoplasia;
- randomized controlled trial;
Cytological cervical screening is rather inefficient because of relatively high proportions of false negative and false positive smears. To evaluate the efficiency of high-risk human papillomavirus (hrHPV) testing, by GP5+/6+ PCR-enzyme immunoassay (EIA), in conjunction with cytology (Intervention Group) to that of the classical cytology (Control Group), we initiated the Population Based Screening Study Amsterdam (POBASCAM). POBASCAM is a population-based randomized controlled trial for implementation of hrHPV testing in cervical screening. The outcome measure is the proportion of histologically confirmed ≥CIN3 lesions in each study arm up to and including the next screening round after 5 years. We present the design, methods and baseline data of POBASCAM. When, in the next 5 years, the follow-up will be completed, the data obtained will be used in model studies, including a cost-effectiveness study, to advise the Dutch Ministry of Public Health in deciding whether cervical screening should be based on combined hrHPV and cytology testing instead of cytology alone. Between January 1999 and September 2002, 44,102 women (mean age = 42.8 years; range = 29–61) that participated in the regular Dutch screening program were included in our study. In the Intervention Group the distribution of cytology and hrHPV by cytology class was as follows: normal cytology 96.6% (3.6% hrHPV positive); borderline and mild dyskaryosis (BMD) 2.5% (34.6% hrHPV positive); and moderate dyskaryosis or worse (>BMD) 0.8% (88.3% hrHPV positive), i.e., 0.4% moderate dyskaryosis (82.9% hrHPV positive), 0.3% severe dyskaryosis (92.5% hrHPV positive), 0.1% carcinoma in situ (95.2% hrHPV positive), <0.1% suspected for invasive cancer (hrHPV positive 100.0%). In the Control Group 96.5% of the women had normal cytology, 2.4% BMD and 0.8% >BMD, i.e., 0.4% moderate dyskaryosis, 0.3% severe dyskaryosis, 0.1% carcinoma in situ, <0.1% suspected for invasive cancer. The presence of hrHPV was age-dependent, decreasing from 12.0% at 29–33 years to 2.4% at 59–61 years. Among women with a positive hrHPV test, the prevalence of BMD was age-dependent ranging from 20.2% at 29–33 years to 7.8% at 54–58 years. In contrast, the risk of >BMD of 13.7% among women with a positive hrHPV test was not age-dependent. Our study indicates that large-scale hrHPV testing by GP5+/6+ PCR-EIA in the setting of population-based cervical screening is practically feasible, is accepted by both participating women and general practitioners and yields highly reproducible results. © 2004 Wiley-Liss, Inc.
Cervical screening by cytology in population-based screening programs has contributed substantially to the decrease in the incidence and mortality of cervical cancer.1, 2, 3, 4 Even in countries with well-organized, quality controlled population-based screening programs, however, the proportion of false positive and false negative smears is at least 10%.5, 6 In most screening programs a 3-way triage is used on the basis of the cytological classification. The large majority of women with a normal smear are advised to have the smear repeated at the next screening round. The interval between screening rounds varies from 1 year in Germany and the United States, 3 years in the United Kingdom and Sweden or 5 years in Finland and The Netherlands.7 The small group of women with serious cytomorphological abnormalities, i.e., moderate dyskaryosis or worse (abbreviated as >BMD, meaning “worse than borderline or mild dyskaryosis”; comparable to HSIL; Fig. 1), are referred for colposcopically directed biopsies (ColpoBx). A relatively large group of women, however, have minor abnormalities, i.e., borderline and mild dyskaryosis (BMD, comparable to ASCUS/LSIL; Fig. 1). The policy to repeat the smear for women with minor abnormalities is quite a burden because only 10% of them have or will develop clinically relevant lesions, i.e., cervical intraepithelial neoplasia Grade 3 (CIN 3), the most advanced precursor lesion for cervical carcinoma or worse.8, 9 In the screening programs ongoing currently in Europe, women with BMD are only referred for ColpoBx when a repeated smear after 6 or 18 months is read as BMD or worse. If women with BMD who are at risk for CIN 3 or worse (≥CIN 3) could be identified with a new test, a more efficient referral strategy could be devised.
Such a test could involve high-risk human papillomavirus (hrHPV) detection because the presence of hrHPV is necessary for the development and maintenance of cervical intraepithelial neoplasia (CIN) lesions and progression to cervical cancer.10, 11, 12, 13, 14, 15 Absence or clearance of hrHPV is associated with regression of CIN lesions.8, 9, 11, 16, 17 In addition, women with hrHPV positive, cytomorphologically normal smears have a 210× (95% CI = 27–1,600) increased relative risk to develop CIN 3 lesions in the next 6 years compared to women with hrHPV negative normal smears.18 Thus, hrHPV testing can be used to identify women at risk for lesions ≥CIN 3.
Recent studies have suggested that adding hrHPV testing to cytology might improve the efficiency of cervical screening by increasing the sensitivity and the negative predictive value for lesions ≥CIN 3. This suggestion opens possibilities for prolonged screening intervals and less referrals for ColpoBx.8, 9, 15, 17, 19, 20, 21, 22, 23, 24, 25, 26 Such an improvement in efficiency would result in better cost-effectiveness of cervical cancer screening. Longitudinal long-term studies in large-scale primary screening settings, however, are missing. To evaluate prospectively in a large-scale population-based primary screening setting whether adding hrHPV testing to classical cytology might result in more efficient screening, we initiated the Population Based Screening Study Amsterdam (POBASCAM). POBASCAM is a population-based randomized controlled trial for implementation of hrHPV testing in cervical screening, in which the efficiency of detecting histologically confirmed cervical lesions ≥CIN 3 by combined hrHPV and cytological testing (Intervention Group) is compared to testing by classical cytology alone (Control Group).
The parameters obtained will be used in modeling studies aimed at improving the efficiency of cervical screening. In that context, issues will include prolonging the screening interval and reducing the number of referrals for ColpoBx, under the condition that not less histologically confirmed lesions ≥CIN 3 are detected compared to the current screening method, i.e., classical cytology. We present the trial design and the cross-sectional baseline data of the 44,102 women enrolled.
MATERIAL AND METHODS
POBASCAM was initiated in the setting of the regular screening program, as a population-based, randomized controlled trial for implementation of hrHPV testing in cervical screening to obtain parameters in 2 screening policies that can be used in modelling studies to optimize cervical screening. The 2 screening policies to be compared use hrHPV testing by GP5+/6+ PCR-enzyme immunoassay (EIA) combined with cytology (Intervention Group) vs. cytology alone (Control Group, with a blinded hrHPV test result). The design of the trial is given in Figure 2.
Our study began in January 1999 and the recruitment of women was completed in September 2002. The study was carried out within the setting of the regular Dutch nationwide screening program. The follow-up will be completed when the women have participated in the next screening round, i.e., in 5 years after enrollment in our study. In this way, in each study arm all histologically confirmed lesions ≥CIN 3 up to and including those detected at the next screening round will be included in the final analyses.
The objective of this trial is to evaluate whether the efficiency of screening is improved by the addition of hrHPV testing to cytology by testing 2 major hypotheses. The first hypothesis concerns the possibility of a prolonged screening interval for women with normal cytology and a negative hrHPV test. A risk of lesions ≥CIN3 of 0.5–0.8% in the time period of 5 years (until the next screening round in the Dutch cervical screening program) for women with normal cytology is accepted currently by health authorities and professionals. We hypothesize that for women with normal cytology and a negative hrHPV test at baseline, this risk is much lower than the accepted risk. The screening interval for them, therefore, might be prolonged. This first hypothesis will be evaluated by testing whether the proportion of histologically confirmed ≥CIN 3 lesions among women with normal cytology and a negative hrHPV test at baseline (Group A.1 in Fig. 2), is less than that among women with normal cytology not hrHPV tested (Group D in Fig. 2).
The second hypothesis concerns the possibility of a reduced number of referrals for colposcopic examination for women with BMD without missing histologically confirmed ≥CIN 3 lesions. This will be evaluated by testing whether, during follow-up of up to 5 years, the proportion of lesions ≥CIN 3 is less in women with BMD of group B.1 B.2.1 compared to those of group E.1 (Fig. 2). The proportion of ≥CIN3 in women with BMD and a negative hrHPV test at baseline (group B.1) and women with BMD at baseline and hrHPV clearance at 6 months (group B.2.1), should be less than that among women with BMD at baseline and normal cytology at 6 months as screened by classical cytology (group E.1).
Triage of women with normal cytology
Women assigned to the Control Group were advised according to the current guidelines for cervical screening in The Netherlands (Fig. 2). The advice was based on cytology results alone (hrHPV test result blinded). According to the guidelines, women with normal cytology will be recalled at the next screening round after 5 years. In the Intervention Group the triage was based both on cytology and hrHPV test result. Women with normal cytology and a negative hrHPV test will be recalled at the next screening round. Women with normal cytology at baseline and a positive hrHPV test were advised to repeat the tests at 6 and 18 months. These intervals were chosen in relation to the mean clearance time of hrHPV of about 1 year.16 In case of BMD/hrHPV+ at 18 months, or normal cytology/hrHPV+ at 18 months, women were referred for ColpoBx. In case of BMD/hrHPV− at 18 months, or normal cytology/hrHPV− at 18 months, women were not recalled until the next screening round.
Triage of women with BMD
Women with BMD were advised to repeat the tests after 6 and 18 months. In the Control Group, those who still had BMD or worse after either 6 or 18 months were referred to a gynecologist for ColpoBx. Women with regression to normal cytology after 18 months will not be recalled until the next screening round. In the Intervention Group, women who had BMD/hrHPV+ at 6 months, or BMD/hrHPV+ or normal cytology/hrHPV+ at 18 months, were referred for ColpoBx. In case of BMD/hrHPV− or normal cytology/hrHPV−, at 18 months, women will not be recalled until the next screening round.
Triage of women with >BMD
In the Intervention and Control Groups, women with >BMD detected at any time (i.e., at baseline or after retesting because of abnormal cytology in both groups or retesting because of a positive hrHPV test in the Intervention Group) were referred directly for ColpoBx. Referrals because of >BMD were always independent of the hrHPV test result.
Randomization, referral and follow-up
Women were randomly 1:1 assigned using the computer's random number generator to the Intervention or Control Group at the time of receiving their administrative data in the central study database. Randomization was independent of cytology or hrHPV test result. Women could be referred for ColpoBx to one of about 40 gynecologists in 4 regional and 1 academic hospital. The gynecologists, who had been informed about POBASCAM, carried out colposcopy and colposcopically-directed biopsy for histological examination according to standardized guidelines of the Dutch Society of Obstetrics and Gynaecology. Histological biopsies were only taken when cervical abnormalities were seen, regardless of the hrHPV status. Eventually, women were treated according to standard protocols. Follow-up data are supplied by the 4 pathology laboratories participating in this trial via the Dutch Pathologisch Anatomisch Landelijk Geautomatiseerd Archief (PALGA, the Dutch nationwide network and registry of histo- and cytopathology).
The primary outcome measure of POBASCAM is the proportion of histologically confirmed ≥CIN 3 lesions found during the time span from intake up to and including the next screening round (i.e., in 5 years). Because women with normal cytology at the next screening round will not be referred for ColpoBx and therefore will not have a histological endpoint, it will be assumed that no CIN lesions are present. This policy complies with standard cervical screening. Parameters obtained will include progression and regression of cytology diagnosis, clearance and acquisition of hrHPV and the number of referrals for ColpoBx to a gynecologist.
Recruitment and enrollment procedures
According to the Dutch cervical cancer-screening program, women between 30–60 years of age are invited every 5 years to have a Pap smear. The invitations for the regular screening program are coordinated by the District Health Authority (DHA). Together with the invitation for the regular screening program, women were informed about POBASCAM. Principally, women were invited by their general practitioner. Women not registered at any general practitioner were invited directly by the DHA. Women were eligible when they lived in a defined semi-urbanized region demarcated according to the DHA Southwest of Amsterdam, were between 30–60 years of age, were invited for the regular population-based screening program having a uterus in situ and were able and willing to give written informed consent for our study. Women were excluded from analysis if they had abnormal cytology or a CIN lesion within a period of 2 years preceding enrollment. The 2-year period was chosen because according to Dutch guidelines during that time, smears could be repeat smears of an indicative smear in the previous 2 years. Women were also excluded for analysis when the sample taken for hrHPV testing at baseline was lost. When inadequate smears were repeated (from 6 weeks–6 months) the adequate smear was chosen as intake smear. The study was approved by the Medical Ethics Committee of the VU University Medical Center (nr 96/103A), and the Ministry of Public Health (VWS nr 328 650).
Education of general practitioners
A total of 242 general practitioners participated in our study. They were invited to attend postgraduate medical education courses and received written information (including a 22-page booklet) pertaining to the issue of answering frequently asked questions of participating women. An information help desk was also available for general practitioners and participating women.
Cervical smears were taken by the general practitioner or their assistant using a Cervex-Brush® or a cytobrush. After making a conventional smear for cytological examination, the brush was placed in a vial containing collection medium (i.e., 5 ml PBS and 0.5% thiomersal) for hrHPV testing. Cervical smears were classified according to the CISOE-A classification by cytotechnologists and abnormal smears were reviewed by experienced cytopathologists (standard way of cervical screening in the nationwide screening program). For cervical screening, a detailed quality control program has been defined by the Netherlands Society for Pathology and implemented at the national level.27 The CISOE-A classification (in Dutch KOPAC-B) is the standard classification used in the Netherlands (Fig. 1).28 Briefly, smears were classified as normal (Pap 1), borderline dyskaryosis (Pap 2), mild dyskaryosis (Pap 3a1), moderate dyskaryosis (Pap 3a2), severe dyskaryosis (Pap 3b), suspected for carcinoma in situ (Pap 4) or suspected for invasive cancer (Pap 5). The comparison of this classification with the Bethesda 2001 classification is given in Figure 1. For our study, the smears were classified in either 1 of 4 participating regional pathology laboratories (Department of Pathology, Spaarne Ziekenhuis, Heemstede; Leiden Cytology and Pathology Laboratory, Leiden; Stichting PA Laboratorium Kennemerland, Haarlem; Department of Pathology, VU University Medical Center, Amsterdam).
Human papillomavirus testing
All women, including the women in the Control Group, have been hrHPV tested at the time of receipt of the HPV sample and all hrHPV testing has been carried out on a daily basis on freshly received material. All hrHPV tests were carried out independent of Pap smear reading in the Department of Pathology (Unit Molecular Pathology, VU University Medical Center). The vials containing the brush in collection medium were sent daily to the molecular pathology laboratory by courier or by mail. A duplicate GP5+/6+ PCR-EIA test was done as described previously.29 A mixture of probes for the detection of 14 high-risk HPV types (HPV 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66 and 68) was used. When the duplicates gave inconsistent results (1%), a repeat duplicate test was carried out and only samples with at least 2 positive tests were ultimately scored hrHPV positive. All the cases with these inconsistent duplicates involved smears with optical density values around the cut-off point, which is routinely used for this test (3 times the optical density of a series of blanks), indicating a relatively low viral load. The final test result was hrHPV positive or hrHPV negative.
Data processing involved a central study database, 4 administrative databases (one in each participating pathology laboratory), and a database holding the hrHPV PCR-EIA test results. Between the central study database and the administrative databases, data were exchanged every night automatically by E-mail in a closed and secure computer network. Authorized hrHPV test results, as registered in the database holding the hrHPV PCR-EIA test results, were automatically copied to the central study database. Patient data and hrHPV test results were matched in the central study database. Patient data, cytology test results, as well as final hrHPV test results (either positive or negative for the Intervention Group or blinded for the Control Group) were manually entered in the administrative databases. Each laboratory had 2 coordinating cytotechnologists who generated a final trial advice according to the study protocol.
A total sample size of 44,000 women was chosen to detect significant differences at a level of p = 0.05, 2-sided, with a power of 0.80, according to assumptions at the time the trial was designed (1996) of a prevalence for BMD of 14% and a prevalence for normal cytology of 84.5%. Women are invited for the regular screening program according to years of birth, every 5 years in the year they become 30, 35, up to 60 years of age. The first screening could be at 29 or 30 years of age. Therefore, the chosen age categories are 29–33, 34–38, etc. To assess the relation between hrHPV status and age categories and between cytology subclasses and age categories, logistic regression and multinominal regression methods were used. Odds ratios (OR) and predictive values were estimated from two-by-two tables. All analyses were carried out using SPSS software version 9.
Enrollment and age
Between January 1999 and September 2002, 49,220 women participated in cervical screening and were eligible for the POBASCAM study. A total of 44,938 women (91.3%) were enrolled. The major reason for women not to be enrolled in our study was largely of logistic origin at the general practitioner's. The proportion of women who refused to participate in POBASCAM was very low (0.3%). After randomization, 22,420 women were allocated to the Intervention Group and 22,518 to the Control Group. In the Intervention Group, 216 women were excluded from analysis because of abnormal cytology or a CIN lesion within 2 years preceding enrollment. Another 7 women were excluded because of a previous hysterectomy and 201 women because of a missing hrHPV test result at baseline. For the same reasons, 217, 9 and 186 women, respectively, from the Control Group were excluded from analysis. Ultimately, the Intervention and Control Groups consisted of 21,996 and 22,106 women, respectively (Fig. 2). The Intervention and Control Groups were comparable for absolute numbers at enrollment, baseline cytology results, age and age distribution. The mean age in both groups was 42.8 years (range = 29–61). Table I shows the baseline cytology and baseline hrHPV test results for the Intervention and Control Group.
|Intervention group||Control group2|
|hrHPV +||hrHPV −||Total|
|Inadequate||3 (6.5)||43 (93.5)||46 (0.2)||43 (0.2)|
|Normal||763 (3.6)||20,482 (96.4)||21,245 (96.6)||21,338 (96.5)|
|BMD||185 (34.6)||349 (65.4)||534 (2.5)||546 (2.4)|
|Borderline||118 (27.4)||312 (72.6)||430 (2.0)||450 (2.0)|
|Mild dyskaryosis||67 (64.4)||37 (35.6)||104 (0.5)||96 (0.4)|
|>BMD||151 (88.3)||20 (11.7)||171 (0.8)||179 (0.8)|
|Moderate dyskaryosis||68 (82.9)||14 (17.1)||82 (0.4)||85 (0.4)|
|Severe dyskaryosis||62 (92.5)||5 (7.5)||67 (0.3)||69 (0.3)|
|Carcinoma in situ||20 (95.2)||1 (4.8)||21 (0.1)||24 (0.1)|
|Invasive cancer||1 (100.0)||0 (0.0)||1 (<0.1)||1 (<0.1)|
|Total||1,102 (5.0)||20,894 (95.0)||21,996 (100.0)||22,106 (100.0)|
Cytology in both arms showed comparable results (Table I). The Intervention Group showed 0.2% inadequate smears, 96.6% normal, 2.0% borderline dyskaryotic, 0.5% mildly dyskaryotic, 0.4% moderately dyskaryotic, 0.3% severely dyskaryotic, 0.1% carcinoma in situ and <0.1% invasive carcinoma smears. The Control Group showed 0.2% inadequate smears, 96.5% normal, 2.0% borderline dyskaryotic, 0.4% mildly dyskaryotic, 0.4% moderately dyskaryotic, 0.3% severely dyskaryotic, 0.1% carcinoma in situ and <0.1% invasive carcinoma smears.
HrHPV testing by GP5+/6+ PCR-EIA in this large-scale screening setting turned out to be practically feasible, and with hrHPV testing being carried out in duplicate, we observed good reproducibility. As analyzed in the Intervention Group, a more severe cytology diagnosis was associated with an increased rate of hrHPV positivity (p-trend < 0.01) (Table I). Among the women with normal cytology 3.6% had a positive hrHPV test. This percentage was 27.4% for women with borderline dyskaryosis and 64.4% for women with mild dyskaryosis. Combining these women into one group of women with BMD showed a positive hrHPV test of 34.6%. In the combined group of women with >BMD, 88.3% had a positive hrHPV test, i.e., an hrHPV positivity rate in moderate dyskaryosis of 82.9%, in severe dyskaryosis of 92.5%, in suspected for carcinoma in situ of 95.2% and in suspected for invasive cancer of 100%.
Histological diagnoses at baseline for women with >BMD
Because women with >BMD at baseline are directly referred for ColpoBx irrespective of the hrHPV test result, cytology and hrHPV results could be related to histologically confirmed diagnoses (Fig. 3). The yield of histologically confirmed lesions ≥CIN 3 was comparable for the intervention and the Control Group, 92/171 and 92/179 respectively. Among the 171 women with >BMD in the Intervention Group, 151 women had a positive hrHPV test and for 23 women histologically follow-up was not yet available, leaving 128 analyzable women in the Intervention Group with >BMD, a positive hrHPV test and histology. Eighty-nine had a histologically confirmed lesion ≥CIN 3 including 5 squamous cell carcinomas (SCC). Likewise, for 5 of 20 women in the Intervention Group with >BMD and a negative hrHPV test, histological follow-up was not yet available, leaving 15 women in the Intervention Group with >BMD, a negative hrHPV test and histology. Three of the 15 had a histologically confirmed CIN 3 lesion (no SCCs). Thus, among women with >BMD, the proportion of histologically confirmed lesions ≥CIN 3 was associated with a positive hrHPV test (OR = 9.1, 95% CI = 2.4–34.2; Fig. 3). The sensitivity and negative predictive value for histologically confirmed lesions ≥CIN 3 of hrHPV testing under the condition of having >BMD were 96.7% and 80.0%, respectively.
hrHPV prevalence in relation to age and cytology class
For all women in the Intervention Group taken together, the prevalence of hrHPV decreased significantly with increasing age from 12.0% in the age category of 29–33 years, 6.8% at 34–38 years, 4.4% at 39–43 years, 3.1% at 44–48 years, 2.5% at 49–53 years, 2.6% at 54–58 years, to 2.4% at 59–61 years (p-trend < 0.01). In women with normal cytology, the hrHPV prevalence decreased from 8.3% in younger women (29–33 years of age) to 2.0% in older women (59–61 years of age; p-trend < 0.01). In Table II, classification of cytology is stratified according to age and hrHPV test result. Among women with a negative hrHPV test, small but age-independent proportions of about 0.1% had >BMD and about 1.7% had BMD. For women with a positive hrHPV test, the risk of >BMD was 13.7% (151/1102), which was independent of age (p-trend = 0.26, range = 13.7% in women 29–33 years, 11.6% in women 59–61 years). Among hrHPV positive women, however, the risk of BMD of 16.8% decreased with increasing age, (p-trend < 0.01, range = 20.2% in women 29–33 years, 7.8% in women 54–58 years).
|29–33||246 (66.1)||75 (20.2)||51 (13.7)||2,713 (98.2)||48 (1.7)||2 (0.1)|
|34–38||208 (66.9)||61 (19.6)||42 (13.5)||4,180 (98.3)||67 (1.6)||4 (0.1)|
|39–43||95 (69.9)||18 (13.2)||23 (16.8)||3,307 (98.0)||64 (1.9)||4 (0.1)|
|44–48||67 (69.8)||15 (15.6)||14 (14.6)||3,324 (98.2)||58 (1.7)||2 (0.1)|
|49–53||57 (74.0)||11 (14.3)||9 (11.7)||2,947 (97.8)||64 (2.1)||3 (0.1)|
|54–58||52 (81.3)||5 (7.8)||7 (10.9)||2,136 (98.4)||31 (1.4)||4 (0.2)|
|59–61||38 (88.4)||0 (0.0)||5 (11.6)||1,875 (99.0)||17 (0.9)||1 (0.1)|
|Total||763 (69.4)||185 (16.8)||151 (13.7)||20,482 (98.2)||349 (1.7)||20 (0.1)|
|p-value4||Reference||p < 0.01||p = 0.26||Reference||p = 0.25||p = 0.75|
The POBASCAM trial (n = 44,102) shows that hrHPV testing by GP5+/6+ PCR EIA is practical in a large-scale, population-based routine primary screening setting. With hrHPV testing being carried out in duplicate we observed good reproducibility. The participation rate among women attending the nationwide cervical screening at one of the general practitioners participating in the study was very high. Of the eligible women, 91.3% was enrolled in the study. In most cases, women were not enrolled in our study due to logistic reasons at the general practitioner's. The proportion of women not consenting to participate in POBASCAM was low (0.3%). This participation rate is in accordance with other studies.19, 30 Because the general practitioners had to answer questions of women about hrHPV testing and the POBASCAM study, it was of essential value to provide them with expert advice. This was done by concise written information (a 22-page booklet), by postgraduate medical education courses including the issue of answering frequently asked questions of participating women and by providing the possibility of asking immediately additional information at a help-desk.
The POBASCAM study will yield parameters to be used in modelling studies, including a cost-effectiveness analysis. These studies will be done to estimate the most efficient algorithm for the detection of ≥CIN 3 lesions in well-organized population-based cervical screening with less inconvenience for the women concerned. Parameters to be estimated include the age-dependent prevalence of hrHPV, hrHPV clearance and acquisition, the risk of having or developing lesions ≥CIN 3 stratified to cytology class among hrHPV positive and hrHPV negative women, the number of regressive CIN lesions among women more closely followed than the standard 5-years interval of regular screening and the sensitivity and specificity of hrHPV testing. The sensitivity and specificity of cytology can be estimated closely using the number of lesions ≥CIN 3 detected among women in the Intervention Group with normal cytology and a (persistently) positive hrHPV test because the vast majority (about 90%) of women with lesions ≥CIN 3 and a false-negative Pap-smear will have persistently positive hrHPV tests. Hence, in the Intervention Group it can be expected that about 90% of the women with false negative cytology will be detected (Groups A.2.3, A.2.2.2 and A.2.2.3 in Fig. 1). In the POBASCAM study women with normal cytology and a positive hrHPV test are retested after 6 and 18 months. These intervals were chosen in relation to the mean clearance time of hrHPV of about 1 year.16
The sample size of 44,000 women was chosen according to assumptions at the time when the trial was designed. In 1996, however, the guidelines of the Dutch regular screening program have changed, by introducing the CISOE-A classification. As a consequence, the proportion of women categorized as BMD decreased from 11.4–2.6%, whereas the subcategory of women with normal cytology increased from 84–96%.31, 33 In the new CISOE-A classification the diagnosis of “borderline changes” is more strictly defined. Similar changes were made in the new guidelines of Bethesda 2001 Classification.32 Before the introduction of the new CISOE-A Classification, a total number of about 5,000 women with BMD would have been diagnosed among 44,000 women screened. To obtain a similar number of women with BMD after the introduction of CISOE-A, the study should have been based on a total of about 190,000 women screened. Although this could mean that testing of the hypothesis concerning referral of women with BMD smears might not reach statistical significance, the parameters obtained with the POBASCAM study will have good accuracy for modelling studies.
From the outset of the POBASCAM study some unavoidable detection bias can be expected when comparing the number of ≥CIN 3 lesions in Group A.1 vs. Group D (Fig. 2). In the current Dutch primary cervical screening program (women age 30–60 years, screening interval = 5 years), a very low risk for interval cervical carcinoma is accepted by health authorities and professionals, which is related to an incidence of lesions ≥CIN 3 of 0.5–0.8% per screening interval of 5 years. The Medical Ethics Committee and the Ministry of Public Health did not gave permission for additional cytology and hrHPV testing or ColpoBx for women with normal cytology with that accepted low risk (Groups A.1 and D in Fig. 2). They further reasoned that the number of ColpoBx and the results to be expected do not warrant the strain for the women involved. A similar type of bias can be expected when comparing the number of ≥CIN 3 lesions in Group B.1 + B.2.1 vs. Group E.1 (Fig. 2). According to our study design, these women were retested at 6 and 18 months of follow-up. They were referred for ColpoBx if retesting showed either >BMD or a positive hrHPV test. However, women with BMD and a negative hrHPV test are not referred for ColpoBx but have to wait to the next screening round, which is after 3.5 years. This follows from studies from Manos et al. and Clavel et al. and from our own experience (follow-up up to 4.3 years). From these studies it can be calculated that the risk of ≥CIN 3 of women with a negative hrHPV test, either in conjunction with normal cytology or BMD is very low when tested additionally after 6 and 18 months (Group B.1 + B.2.1 in Figure 2).8, 9, 15, 17 This risk is expected to be at least as low as that for women with normal cytology in the standard Dutch nationwide screening program with an interval of 5 years. The magnitude of these biases in detecting histologically confirmed lesions ≥CIN 3 up to and including those found at the next screening round can still be estimated when the results of hrHPV testing in the Control Group can be taken into account and data of the Control Group can be compared to those of the Intervention Group.
The cytology baseline results of our study are comparable with the cytology results of the Dutch nationwide screening program in the time period of recruitment31 and with the results reported in one region of the Dutch nationwide screening program.33 Moreover, our baseline hrHPV prevalence of 5.0% is in accordance with international literature. Two studies in the setting of population-based screening among women with a comparable mean age showed similar hrHPV prevalences.20, 25 In other studies, higher prevalences of hrHPV were found, which is probably related to the study population, age or the hrHPV test used. Using HCII in a hospital-based population, Clavel et al.8 found an hrHPV prevalence of 15%. Among younger women (mean age = 25 years) Kulasingam et al.19 found an hrHPV of 18%, and among women with a mean age of 30 years, Ratnam et al.22 found a prevalence of 10%. Because hrHPV is considered to be the predominant agent for the development of cervical carcinoma and its precursor lesions, we analyzed the prevalence of BMD and >BMD among women stratified according to hrHPV status.
In women with a negative hrHPV test the prevalences of BMD or >BMD were low and independent of age. In women with a positive hrHPV test, the prevalence of >BMD of 13.7% (or about 1 in 7) was also independent of age, i.e., women with an hrHPV infection have a flat risk of having severe cervical lesions. In contrast, for women with a positive hrHPV test the prevalence of BMD was age-dependent, which suggests that most BMD lesions reflect a cytopathological effect of primary hrHPV infection. Naturally, with increasing age, the chance of contracting a primary hrHPV infection decreases that could explain the relative decrease in BMD prevalence.
Of special interest are the 3 cases of histologically confirmed ≥CIN 3 lesions among women with >BMD and a negative hrHPV test. After additional analyses, one sample shown to be β-globin negative and after DNA-isolation this sample turned out to contain hrHPV types 33 and 51. Using hrHPV E7 type-specific PCR, another sample shown to be positive for type 66, indicating viral integration with disruption of the L1 region that is recognized by the GP5+/6+ primers.10 The third sample remained hrHPV negative after additional analyses with E7 type specific PCRs. Because women with >BMD are directly referred for ColpoBx, irrespective of hrHPV status, hrHPV prevalence among these women does not have clinical consequences.
Our large-scale study indicates that hrHPV testing by GP5+/6+ PCR EIA in the setting of population-based cervical screening is practically feasible, is accepted by both participating women and general practitioners and yields highly reproducible results.
We gratefully acknowledge the work of the 242 general practitioners and their assistants, the cytotechnologists and administrators from the 4 regional laboratories, the cytotechnologists and the molecular technicians of the pathology department of VU University Medical Center, Medial Haarlem, the District Health Authority Amstelveen, DHV Kennemerland-Haarlemmermeer e.o., and PALGA, for their work and support in our study. We especially thank, in alphabetical order, R. van Andel, B. Bakker, I. Barendse, R. van Beek, J. Beliën, A. Bieze, J. ter Borg, S. Bulk, M. van Casteren, P. van Diemen, N. Fransen-Daalmeijer, A. van der Geest, A. Groothuismink, F. Hendrikse, H. van Keep, M. van der Laan, M. Lettink, M. Meurs, J. Pleisters, R. Pol, S. van Schaick, D. Schumacher, F. Stuart-Vogelsang and M. Verkuyten, for their technical assistance. We thank H. Berkhof for his statistical advice. The study is funded by ZON, Zorg Onderzoek Nederland (Netherlands Organisation for Health Research and Development; grant 30-05220).