Results of delayed triage by HPV testing and cytology in the Norwegian Cervical Cancer Screening Programme

Acta Oncologica ISSN: 0284-186X (Print) 1651-226X (Online) Journal homepage: http://www.tandfonline.com/loi/ionc20 Results of delayed triage by HPV testing and cytology in the Norwegian Cervical Cancer Screening Programme Tor Haldorsen, Gry Baadstrand Skare, Giske Ursin & Tone Bjørge To cite this article: Tor Haldorsen, Gry Baadstrand Skare, Giske Ursin & Tone Bjørge (2015) Results of delayed triage by HPV testing and cytology in the Norwegian Cervical Cancer Screening Programme, Acta Oncologica, 54:2, 200-209, DOI: 10.3109/0284186X.2014.932433 To link to this article: https://doi.org/10.3109/0284186x.2014.932433 Published online: 24 Jun 2014. Submit your article to this journal Article views: 393 View related articles View Crossmark data Citing articles: 1 View citing articles Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalinformation?journalcode=ionc20 Download by: [46.3.195.162] Date: 09 December 2017, At: 12:02

Acta Oncologica, 2015; 54: 200 209 ORIGINAL ARTICLE Results of delayed triage by HPV testing and cytology in the Norwegian Cervical Cancer Screening Programme TOR HALDORSEN 1, GRY BAADSTRAND SKARE 2, GISKE URSIN 3,4,5 & TONE BJ Ø RGE 3,6 Downloaded by [46.3.195.162] at 12:02 09 December 2017 1 Department of Research, Cancer Registry of Norway, Oslo, Norway, 2 Cervical Cancer Screening Programme, Cancer Registry of Norway, Oslo, Norway, 3 Cancer Registry of Norway, Oslo, Norway, 4 Institute of Basic Medical Sciences, University of Oslo, Oslo, Norway, 5 Department of Preventive Medicine, University of Southern California, Los Angeles, California, USA and 6 Department of Global Public Health and Primary Care, University of Bergen, Bergen, Norway ABSTRACT Background. High-risk human papilloma virus (hrhpv) testing was added to the cytology triage of women with equivocal screening smears in the Norwegian programme for cervical cancer screening in 2005. In this population-based observational before and after study we assessed the effect of changing the screening algorithm. Material and methods. In periods before and after the change 75 852 and 66 616 women, respectively, were eligible for triage, i.e. they had smear results of unsatisfactory, atypical squamous cells of undetermined significance (ASC-US), or low-grade squamous intraepithelial lesion (LSIL) at routine screening. The triage was delayed as supplementary testing started six months after the initial screening. The s were compared with respect to results of triage and later three-year cumulative incidence of cervical intraepithelial neoplasia grade 2 or worse (CIN2 ). Results. Before and after the change in the screening algorithm 5.2% (3964/75 852) and 8.1% (5417/66 616) of women, respectively, were referred to colposcopy. Among women referred to colposcopy cumulative incidence of CIN2 (positive predictive value of referral) increased from 42.0% [95% confidence interval (CI): 40.3 43.7%] in the period with cytology only to 48.0% (95% CI 46.6 49.4%) after the start of HPV testing. For women recalled to ordinary screening the three-year cumulative incidence decreased from 2.7% (95% CI 2.5 2.9%) to 1.0% (95% CI 0.9 1.2%) during the same period. Among women with LSIL at routine screening and HPV testing in triage, 52.5% (1976/3766) were HPV positive. Conclusion. The new algorithm with HPV testing implemented in 2005 resulted in an increased rate of referral to colposcopy, but in a better risk stratification with respect to precancerous disease. Cervical cancer screening is recommended by several international organisations [1,2]. Guidelines on screening policy and organisation of the programmes, as well as monitoring and evaluation, are provided both on international and national level [3 5]. Screening should take place within an organised programme and activities in all parts of the programme should be executed in accordance with given guidelines. The screening methods should be based on scientific evidence, and all aspects of the programme should undergo rigorous quality assurance. There should also be a plan for monitoring the screening process and for publication of the results [6]. Special care is needed if screening methods are altered over time. Changes will usually be based on international research results, but any such changes may require modification in order to be implanted in a local setting. When new modalities have been introduced, their effectiveness should be assessed [3]. Cervical cytology remains the standard test for cervical cancer screening in many countries. Women with negative tests are recommended to have a new test within 3 5 years [3,7]. Women with cytological Correspondence: T. Haldorsen, Department of Research, Cancer Registry of Norway, PO Box 5313 Majorstuen, 0304 Oslo, Norway. E-mail: tor.haldorsen@ kreftregisteret.no (Received 27 January 2014 ; accepted 2 June 2014 ) ISSN 0284-186X print/issn 1651-226X online 2014 Informa Healthcare DOI: 10.3109/0284186X.2014.932433

Downloaded by [46.3.195.162] at 12:02 09 December 2017 results indicating high-grade lesions are referred to diagnostic work-up, such as colposcopy. For a long time there has been a discussion on the best management of women with lower grade cytology results, such as atypical squamous cells of undetermined significance (ASC-US) and low-grade squamous intraepithelial lesion (LSIL) [3,8]. In previous decades, the strategy in Norway and elsewhere for handling these women has been repeat cytology (Figure 1). After the introduction of commercial high-risk human papilloma virus (hrhpv) tests, several studies have investigated the performance of hrhpv tests in the management of these women [9 12] and recommendations on their use have been given [3]. So far, there seems to be a lack of reports with results from the introduction of HPV test for this purpose in routine screening. However, preparatory work has been reported [13,14]. In the Norwegian Cervical Cancer Screening Programme (NCCSP) the use of hrhpv tests in triage of abnormal cytological results has been advocated since July 2005 [15]. Before that time repeat cytology was the only option. The national recommendations from 2005 include certain options not often found in other studies and programmes for HPV testing in triage 201 cervical cancer screening; the secondary testing with hrhpv tests occurs six months after the initial Pap smear (delayed triage), and includes a new cytology as well (Figure 2). Another special feature was the triage of women with an unsatisfactory screening smear that took place until 2009. The aim of this observational study was to analyse the effects of introducing HPV testing in triage in the NCCSP in 2005. Specifically, we determined the effect of the changes in guidelines on the proportion of women referred to colposcopy, as well as cancer precursor rates in s defined by the screening process. Material and methods Cervical cancer screening in Norway Norwegian health authorities recommend women 25 69 years old to have a screening smear every three years. A national screening programme against cervical cancer was established in 1995 and relies on centralised registration of cytological smears from the cervix starting in 1991. Pap smears are obtained by general practitioners (GP) or gynaecologists. By Figure 1. Flowchart for triage with repeat cytology.

202 T. Haldorsen et al. Downloaded by [46.3.195.162] at 12:02 09 December 2017 linkage to the national population registry the Secretariat of the NCCSP at the Cancer Registry of Norway (CRN) can identify women who are not registered with a smear for the last three years. These women are sent a reminder (personal letter) recommending that they contact their GP or gynaecologist for having a smear. A second reminder is sent after an additional year if no test result is registered in the meantime. The registries of individual screening data are linked to the national cancer registry, which includes information on incident cancers and precursors lesions since 1953. More details on cervical cancer screening in Norway have been reported elsewhere [16,17]. Data sources Figure 2. Flowchart for triage with hrhpv test and cytology. NEG, negative; POS, positive. All private as well as public pathology and microbiology laboratories are required by law to report cervical test results to the CRN. The reports include personal identifiable information (including the unique personal identification number assigned to all residents of Norway) on all individuals with Pap smears as well as histology specimens and HPV tests. Since 2005 cytological results are coded according to the Bethesda 2001 classification [18]. For this analysis results recorded before 2005 were recoded according to the same classification to make comparable s. Cervical intraepithelial neoplasia (CIN) is classified according to international recommendations [2]. Triage protocols Women with the screening cytology results, unsatisfactory, ASC-US and LSIL, undergo additional testing (triage). Before 2005, triage was based on repeated testing with cytological smears. Guidelines at that time advised a new cytology six months after the equivocal initial smear (delayed triage) (Figure 1). If the first triage cytology did not give a clear result (i.e. unsatisfactory, ASC-US or LSIL), a second triage cytology was performed 12 months after the first triage cytology (18 months after the initial smear). Women with a normal result on the first or second triage cytology were advised to have a new screening cytology in three years. The others were referred to colposcopy. Since July 2005 onwards, the guidelines [15] specify the use of hrhpv tests in triage (Figure 2). According to the guidelines, a HPV test and a new cytology should be obtained six months after the initial cytology. Women with high-grade cytology or ASC-US/LSIL cytology and positive HPV test at six months are referred to colposcopy. Women with other cytology results and negative HPV test at this point

Downloaded by [46.3.195.162] at 12:02 09 December 2017 are returned to routine screening. If the HPV test is positive but the second cytology is normal or unsatisfactory, then a new HPV test is recommended after another six-month period (12 months after the initial smear) (Figure 2). If the second HPV test is negative, the women are returned to routine screening, otherwise they are referred to colposcopy. Liquid-based cytology was rare during the study period and HPV tests were taken on separate samples. Study population In this observational study we compared the results from triage in the national programme during two time periods. The first period represented the historical comparison, where we included all women who underwent triage resulting from an initial cytology obtained between 1 January 1999 and 31 December 2001 (Period 1). In the second period we included all women who underwent triage with an initial cytology between 1 July 2004 and 30 June 2008 (Period 2), the will cover all triage procedures the first three years of HPV testing. In addition, Period 2 was divided into those who underwent triage with HPV test and cytology (HPV ) and those who had no HPV test, i.e. where the physician presumably had continued using the old guidelines (Cytology ). Cytological smears and HPV tests may be taken for multiple purposes and the registry does not contain information on the indication of each test. Our main interest was triage following an initial cytology obtained as part of screening, or as part of primary prevention of cervical cancer. Therefore, we excluded women with a previous diagnosis of gynaecological cancer and women who within two years before the potential initial cytology had either had an abnormal cytology or had been diagnosed with CIN2 or worse (CIN2 ). The aim of the triage is to separate the women into two s, one for diagnostic work-up (referral ) and the other for routine screening (screening ). The separation should take place within the time limits of the guidelines. We applied a liberal view when setting these time limits for this analysis. For the triage tests with a recommended interval of six months, we accepted intervals from 1 to 12 months. For the second triage cytology in Period 1 with a recommended interval of 12 months after the first triage cytology, we accepted intervals of 1 18 months. If a HPV test was used for triage, there should be a simultaneous cytology. A smear registered the same month or 1 2 months after the HPV test was considered as part of the secondary testing. Even with this liberal interpretation of the guidelines there were HPV testing in triage 203 some deviations in testing practice. This was usually because the recommended triage test took place after our liberal upper time limit. For these cases we assigned the date of the upper limit as the end of triage and coded the triage as incomplete. Some events interrupted the triage process; women were censored at death, emigration, or if she had a CIN2 or gynaecological cancer registered before the appropriate triage tests were obtained. These cases were coded as censored in the presentations of the results. Five different hrhpv tests were used during the first three-year period of cotesting in triage and included in this study. The three most commonly used tests were the Hybrid Capture II assay, [(HC II, Qiagen, Gaitersburg, MD USA), PreTect HPV- Proofer (NorChip, Klokkarstua, Norway) and the Amplicor HPV Test (Roche Diagnostics, Basel, Switzerland)]. A comparative effectiveness study of the three tests has been performed [19]. Statistical analyses For evaluating the introduction of HPV tests in triage two types of comparisons were made. First, the results in Period 1 were used as a historical comparison (baseline) for results in Period 2. Secondly, the results from the HPV were compared to those in the Cytology (both within Period 2). These s were compared with respect to the results of the triage process. Specifically, the s and subs defined by the triage were followed for incidence of CIN2. Maximum follow-up was to the end of 2003 for women in Period 1 and to the end of 2009 for women in Period 2. Accordingly, the maximal length of follow-up was between 2.0 and 5.0 years for women in Period 1 and between 1.5 and 5.5 years for women in Period 2. Each woman was followed until the end of follow-up, CIN2 or a censoring event (death, emigration or a gynaecological cancer), whichever came first. Incidence of disease was measured by the cumulative hazard rate and estimated by the Nelson-Aalen estimator [20]. For presentation, we chose to use cumulative incidence computed by 1-exp(-H), where H is the cumulative hazard rate. Only data on women in the referral and the screening were included for the summary presentation of the follow-up after triage (Table II). Outcome measures included percentage referred to colposcopy among those with completed triage, and cumulative incidence of CIN2 at three years of follow-up for the referral and for the screening. The end point at three years was chosen since this is the recommended screening interval in Norway. The cumulative incidence of the referral

204 T. Haldorsen et al. Downloaded by [46.3.195.162] at 12:02 09 December 2017 might also be called the positive predictive value (PPV) of referral and the cumulative incidence of the screening equals one minus the negative predictive value (1-NPV) of referral. We also computed supplementary measures that could be called algorithm sensitivity and specificity, i.e. estimates for the proportion of women with CIN2 that were selected to the referral and the proportion of women without CIN2 that were selected to the screening (Table II). Analyses were made for all women in triage and separately for each with respect to results of the initial cytology. Frequencies were compared by test for proportions in two independent s. Differences between cumulative incidences at three years were tested by estimates from Stata of cumulative hazards. Differences between sensitivities were tested using standard errors for a linear approximation to this parameter. The statistical packages Stata 11 and Stata 12 were used for the analyses [21]. We used two-sided tests for statistical significance and a significance level of 0.05. Results There were 75 852 women eligible for triage in Period 1 and 66 616 women in Period 2. In both periods, more than 50% of the women had unsatisfactory as initial cytology result. The ratio of women Table I. Results of triage (%) by initial cytology. Result of triage with LSIL versus ASC-US cytology was higher in Period 2 (8812/22 224) than in Period 1 (6709/19 987). For the two s in Period 2, initial cytologies of ASC-US and LSIL were more common in the HPV than in the Cytology (Table I). From Period 1 to 2 the percentage of women referred to colposcopy increased from 5.2% to 8.1% (Table I) (p 0.001). For those with initial cytology ASC-US the percentage increased from 8.5% to 11.8% (p 0.001) and from 17.3% to 23.8% (p 0.001) for those with LSIL. In both periods the lower percentage for referral was observed among those with an unsatisfactory initial cytology (Table I). The percentage of incomplete triage was 31.0% in Period 1 and 33.2% in Period 2. In both periods the higher percentage was observed in the with unsatisfactory initial smears. The percentage of being censored before end of triage was 1.0% and 1.6% in Period 1 and Period 2, respectively (Table I). A higher percentage of incomplete triage was observed in the Cytology than in the HPV, 40.1% versus 9.8%. For all categories of initial cytology the percentage of referral was higher for the HPV than for the Cytology. During follow-up, 2513 CIN2 were observed among triaged women in Period 1 (Table II), of these 647 CIN2 and 63 cancers. The corresponding numbers for Period 2 were 2763 CIN2, 998 CIN2 and 38 cancers. Among women in the screening s, Initial cytology Unsatisfactory ASC-US LSIL Period 1 * Screening 31 491 (64.1) 12 397 (62.0) 3692 (55.0) 47 580 (62.7) Referral 1108 (2.3) 1696 (8.5) 1160 (17.3) 3964 (5.2) Incomplete triage 16 322 (33.2) 5601 (28.0) 1606 (23.9) 23 529 (31.0) Censored 235 (0.5) 293 (1.5) 251 (3.7) 779 (1.0) Total 49 156 (100.0) 19 987 (100.0) 6709 (100.0) 75 852 (100.0) Period 2 Screening 20 441 (57.5) 13 587 (61.1) 3980 (45.2) 38 008 (57.1) Referral 700 (2.0) 2617 (11.8) 2100 (23.8) 5417 (8.1) Incomplete triage 14 242 (40.0) 5537 (24.9) 2364 (26.8) 22 143 (33.2) Censored 197 (0.6) 483 (2.2) 368 (4.2) 1048 (1.6) Total 35 580 (100.0) 22 224 (100.0) 8812 (100.0) 66 616 ( 100.0) HPV Screening 2288 (85.0) 5892 (67.9) 1759 (46.7) 9939 (65.7) Referral 195 (7.2) 1866 (21.5) 1526 (40.5) 3587 (23.7) Incomplete triage 202 (7.5) 840 (9.7) 435 (11.6) 1477 (9.8) Censored 8 (0.3) 76 (0.9) 46 (1.2) 130 (0.9) Total 2693 (100.0) 8674 (100.0) 3766 (100.0) 15 133 (100.0) Cytology Screening 18 153 (55.2) 7695 (56.8) 2221 (44.0) 28 069 (54.5) Referral 505 (1.5) 751 (5.5) 574 (11.4) 1830 (3.6) Incomplete triage 14 040 (42.7) 4697 (34.7) 1929 (38.2) 20 666 (40.1) Censored 189 (0.6) 407 (3.0) 322 (6.4) 918 (1.8) Total 32 887 (100.0) 13 550 (100.0) 5046 (100.0) 51 483 (100.0) Total * Initial cytology 1 January 1999 to 31 December 2001; Initial cytology 1 July 2004 to 30 June 2008; HPV test for triage (Period 2); No HPV test for triage (Period 2); Incorrect tests or timing of tests; Death, emigration, gynaecological cancer or CIN2 before completed triage testing.

HPV testing in triage 205 Table II. Follow-up on CIN 2 for 3 years after triage for women returned to routine screening and referred to colposcopy. Initial cytology Group Observed events during follow-up Screening Cumulative incidence (%) (95% confidence interval) Observed events during follow-up Referral Cumulative incidence (%) (95% confidence interval) Percentage of triaged women in referral Estimated percentage of all diseased * in referral Estimated percentage of all nondiseased in screening Downloaded by [46.3.195.162] at 12:02 09 December 2017 All Period 1 1029 2.7 (2.5 2.9) 1484 42.0 (40.3 43.7) 7.7 56.5 95.3 Period 2 337 1.0 (0.9 1.2) 2426 48.0 (46.6 49.4) 12.5 86.9 93.0 HPV 114 1.3 (1.1 1.5) 1803 51.9 (50.2 53.6) 26.5 93.6 85.0 Cytology 223 0.9 (0.8 1.1) 623 38.9 (36.5 41.5) 6.1 72.9 96.1 Unsatis Period 1 225 0.9 (0.8 1.1) 135 13.7 (11.6 16.1) 3.4 34.2 97.0 factory Period 2 61 0.4 (0.3 0.5) 140 21.5 (18.5 24.9) 3.3 66.6 97.4 HPV 6 0.3 (0.1 0.6) 75 39.4 (32.8 46.8) 7.9 92.3 95.1 Cytology 55 0.4 (0.3 0.5) 65 13.8 (10.9 17.2) 2.7 50.2 97.6 ASC-US Period 1 551 5.6 (5.2 6.1) 820 53.9 (51.2 56.6) 12.0 56.8 93.7 Period 2 169 1.5 (1.3 1.7) 1260 51.6 (49.6 53.7) 16.2 87.2 91.4 HPV 61 1.2 (0.9 1.5) 950 52.9 (50.5 55.3) 24.1 93.6 86.9 Cytology 108 1.7 (1.4 2.1) 310 47.4 (43.4 51.6) 8.9 73.1 95.0 LSIL Period 1 253 8.7 (7.7 9.8) 529 51.7 (48.3 54.8) 23.9 65.2 85.8 Period 2 107 3.1 (2.6 3.8) 1026 51.9 (49.6 54.2) 34.5 89.7 79.2 HPV 47 3.0 (2.3 4.1) 778 52.2 (49.6 54.8) 46.5 93.7 70.0 Cytology 60 3.3 (2.5 4.2) 248 51.3 (46.6 56.3) 20.5 80.3 88.5 * CIN2 within 3 years; No CIN2 within 3 years; Initial cytology 1 January 1999 to 31 December 2001; Initial cytology 1 July 2004 to 30 June 2008; HPV test for triage (Period 2); No HPV test for triage (Period 2). the cumulative incidence of CIN2 after three years decreased from 2.7% to 1.0% from Period 1 to Period 2 (p 0.001) (Table II). For the referral s there was an increase in PPV from 42.0% to 48.0% (p 0.001). A higher proportion of those diagnosed with CIN2 within three years was triaged into the referral in Period 2 than in Period 1. This was reflected in the algorithm sensitivity which increased from 56.5% in Period 1 to 86.9% in Period 2 (p 0.001). The proportion of non-diseased in the screening was 95.3% in Period 1 and 93.0% in Period 2. Accordingly, there was a decrease in the algorithm specificity from Period 1 to Period 2. In the last period a higher cumulative incidence of CIN2 was observed in the HPV both for those referred to colposcopy and to screening, 51.9% and 1.3% versus 38.9% and 0.9% for the Cytology (Table II). A higher algorithm sensitivity was found in the HPV than in the Cytology, 93.6%, 72.9% (p 0.001). For the screening s the cumulative incidence of CIN2 in Period 1 was 5.6% and 8.7% for ASC-US and LSIL screening cytology, respectively (Table II). In Period 2 the cumulative incidence was 1.5% and 3.1% for these s (p 0.001 and p 0.001). For all categories of initial cytology the higher algorithm sensitivity was found in the HPV. Follow-up for the estimates in Table II started at the date of referral to colposcopy or return to routine screening. For all women eligible for triage, follow-up from date of the initial cytology in Period 2 showed that the cumulative incidence of CIN2 at three years was 1.0% [95% confidence interval (CI) 0.9 1.2%] for the with initial unsatisfactory cytology, 9.8% (95% CI 9.4 10.2%) for the ASC-US and 19.6% (95% CI 18.8 20.5%) for the LSIL. The CIN2 rate was 1.85 10 2 (95% CI 1.80 1.91 10 2 ) in Period 1 and 1.96 10 2 (95% CI 1.91 2.02 10 2 ) in Period 2. The mean numbers of person years were 3.48 and 3.42, respectively. In the HPV the positivity rates of the HPV tests by the results of the initial cytology were as follows: 14.0% (378/2693) (unsatisfactory), 30.0% (2603/8674) (ASC-US) and 52.5% (1976/3766) (LSIL) (Table III). For all three s there was a decreasing trend in positivity rate by age. Discussion A successful triage algorithm is indicated by a high rate of CIN2 among women in the referral (PPV), i.e. those referred to colposcopy, and a very low rate in the screening, i.e. the women

206 T. Haldorsen et al. Table III. hrhpv positive (%) by age and initial cytology. Initial cytology Unsatisfactory ASC-US LSIL Total Age (years) Number of women hrhpv-positive (%) Number of women hrhpv-positive (%) Number of women hrhpv-positive (%) Number of women hrhpv-positive (%) 25 34 896 193 (21.5) 2487 1211 (48.7) 1687 1031 (61.1) 5070 2435 (48.0) 35 44 913 112 (12.3) 2718 821 (30.2) 1225 614 (50.1) 4856 1547 (31.9) 45 54 591 54 (9.1) 2189 380 (17.4) 643 258 (40.1) 3423 692 (20.2) 55 69 293 19 (6.5) 1280 191 (14.9) 211 73 (34.6) 1784 283 (15.9) Total 2693 378 (14.0) 8674 2603 (30.0) 3766 1976 (52.5) 15133 4957 (32.8) Downloaded by [46.3.195.162] at 12:02 09 December 2017 recalled to ordinary screening as a result of triage. The introduction of hrhpv tests in triage in the Norwegian screening programme resulted in better separation of the women with respect to the risk of CIN2. The new algorithm with HPV testing and cytology had higher sensitivity than the former algorithm based on cytology only. However, the new algorithm resulted in a modest decrease in specificity. It also resulted in a higher percentage of women referred to colposcopy, and thus to a higher workload. These conclusions are suggested by the observed changes from Period 1, when repeat cytology was the only available triage, to Period 2 where cotesting for hrhpv was an option. The results from the comparison of the HPV and the Cytology in Period 2 corroborate that use of HPV testing in triage has caused at least part of the observed changes. The decrease in three-year cumulative incidence of CIN2 from Period 1 to Period 2 in the screening contributed substantially to the better risk stratification in triage in the second period. This meant that risk of disease in women triaged back to routine screening were at more acceptable levels in Period 2. In the first period the risks were above 2.0% which has been suggested as an upper limit [22]. Comparison of results in Period 1 and 2 is the main part of the analysis. Since not all triage in the second period was performed with HPV cotesting, we computed linear extrapolations to see if these could be used as an estimate for the situation with complete adherence. Results for the separate HPV and Cytology in Period 2 were used for supplementary analysis. We have to be aware of possible sources of bias in the comparison of Period 1 and 2, and between the HPV and the Cytology. Cytological results from Period 1 (1999 2001) were recoded in accordance with the Bethesda classification adapted from 2005 in cervical cancer screening in Norway. However, differences remain in the definition of unsatisfactory, and this definition had a broader content in the first period. Relatively more women had unsatisfactory as initial cytology in Period 1. We also observed that the ratio of LSIL cytology to ASC-US cytology is higher in Period 2 than in Period 1. Since the risk of precursors of cervical cancer is highly different in the three initial s, these observations indicate a higher average risk in Period 2. Several of the measures we present are dependent on the risk level and comparison between the periods should be undertaken with analysis stratified on categories of initial cytology. For some of the women triage was not attempted at all. As we classified women with no HPV test in Period 2 as belonging to the Cytology, the number of incomplete screening was high in this. Even when we restricted our analysis to women with complete triage, it is possible that systematic differences between the s could have influenced our results. For each of initial cytology the percentage of women triaged to the referral were smaller in the cytology in Period 2 than in Period 1. Within Period 2 there may have been systematic differences between those who underwent HPV testing and those who underwent cytology only. Specifically, it is possible that women with higher risk more often were selected for HPV triage in Period 2. Such a selection might have influenced the observed differences for measures that are dependent on the risk level (Table II). Follow-up of later disease was performed by linkage of national registries of screening and disease and no adjustment for the higher diagnostic activity in the referral was attempted. Since unsatisfactory is seldom used as indication for triage in other programmes and international guidelines are given separately for ASC-US and LSIL, results for each initial cytology are discussed separately below. Triage of initial LSIL cytology From Period 1 to Period 2 the referral rate increased from 23.9% to 34.5%. Extrapolation of these results

Downloaded by [46.3.195.162] at 12:02 09 December 2017 yields 45.1% with complete adherence since HPV cotesting was used for about half the women with complete triage. This is below the observed result 46.5% in the HPV which we find reasonable since it was indicated that this was a selective with risk of CIN2 above the average. Cumulative incidence in the screening decreased from 8.7% to 3.1%. Extrapolation of this result implies a decrease of 8.7 3.1 5.6% points from 3.1%, a negative, impossible value. It is likely that some change from Period 1 to Period 2 is not caused by introduction of HPV tests in triage. Cumulative incidence in the screening is a parameter increasing with the risk level and we propose the observed value 3.0% in the HPV as an upper limit for this parameter if complete adherence to the recommendations. Current European guidelines for cervical cancer screening do not recommend HPV testing in triage for all women with LSIL as initial cytology result, because most women in the are HPV test positive and the test will offer low discriminatory power [3]. This may, however, depend on whether HPV testing is conducted at the same time as the initial cytology (reflex HPV testing) or delayed, as in our study. Liquid-based cytology was very rare in Norway in 2005. Therefore, the national guidelines recommend delayed triage with a new cytology and hrhpv test six months after the initial cytology, also for the LSIL. The hrhpv positivity rate for the LSIL in Norway was 52.5% (Table III). This is lower than observed in most international studies with reflex testing of HPV [14,23], and as explained above, is possibly due to the delayed HPV testing. We observed favourable changes in cumulative incidence of CIN2 for the screening and improved sensitivity from Period 1 to Period 2. The difference in these measures between the HPV and Cytology (Table II) supports that introduction of HPV in triage at least in part caused these changes. If the new guidelines with HPV testing were applied for all eligible women, we estimate that less than 50% of the LSIL would be referred to colposcopy. Higher referral rates are indicated with reflex HPV testing [14,24]. For the LSIL the three-year risk of CIN2 was approximately 20%. A risk for which referral to colposcopy should be considered [22]. If one chooses to triage this, high sensitivity is paramount and it may be defendable to refer more than 40% to colposcopy for achieving optimal sensitivity. In our case, delayed triage of the LSIL was not optimal; there was a 3% risk of CIN2 among those who were sent back to screening. Some will argue that a with such a risk needs prolonged HPV testing in triage 207 surveillance [22] so our results illustrate the wellknown problems of finding a perfect management for the LSIL in all respects. Triage of initial ASC-US cytology From Period 1 to Period 2 the referral rate increased from 12.0% to 16.2%. Extrapolation of these results yields 20.4% with complete adherence since HPV cotesting was used for about half the women with complete triage. This is below the observed result 24.1% in the HPV which probably is a result influenced by selective recruitment to the. The cumulative incidence of CIN2 in the screening decreased from 5.6% to 1.5% from Period 1 to Period 2 and extrapolation results in negative, impossible values as for women with LSIL as initial cytology. It is likely that some of the improvement from the first to the second period is not caused by the introduction of HPV test in triage. In analogy with the result for the LSIL we think that the observed value of 1.2% in the HPV may serve as an upper limit for possible values if complete adherence. HPV triage is recommended for women with an initial ASC-US cytology [3]. The positivity rate of HPV tests in NCCSP with delayed triage was about 30%. This is below levels reported by others [14,23]. Reflex HPV testing would be expected to result in a higher percentage of HPV-positive women. Use of hrhpv testing in triage of the ASC-US may result in more than 20% of women being referred to colposcopy. However, this triage with delayed HPV testing gave a satisfying separation in s with respect to later disease. At least 75% were returned to routine screening without further examinations, and the three-year cumulative incidence of CIN2 for this was estimated to be as low as 1.2%. Further, the incidence of CIN2 in the referral (PPV) was above 50%. Thus, the low risk in the screening and the high risk in the referral meant that a satisfactory separation was achieved with the new algorithm. The hrhpv positivity rate for the younger age 25 34 years was 48.7%, but even with this high positivity rate, one could argue that triage with delayed HPV test seems tenable. Triage of initial unsatisfactory cytology Unsatisfactory cytology has been used as an indication for triage in Norway both before and the first four years after the introduction of HPV tests in triage. A recommendation for such treatment of those women cannot be found in the European guidelines [3]. Former analysis of Norwegian data

Downloaded by [46.3.195.162] at 12:02 09 December 2017 208 T. Haldorsen et al. has demonstrated an increased risk of CIN2 for this of screened women [25]. Our results suggest that adjunct HPV testing was used in less than 10% of cases with unsatisfactory screening cytology. These results provide no good clues to predictions of trends in the situation with complete adherence to the recommendations. In our analysis triage with cotesting had more favourable predictive values than triage with repeated smears only (Table II). The sensitivity of HPV triage was equally high for this and for women with ASC-US and LSIL as initial cytology. There are, however, some doubts if this is at elevated risk. The three-year risk of CIN2 in Period 2 was 1% for the unsatisfactory, a risk some authors find acceptable for advising the women to return to routine screening [22,23].These results support the Norwegian guideline from 2009 for repeat cytology alone after unsatisfactory cytology. The women are just recommended to have a new cytology within six months. Adherence to recommendations After HPV tests were recommended for triage in 2005, not all triage were performed in accordance with the new guidelines. Some medical doctors simply continued the use of repeat cytology. Among those with ASC-US or LSIL at initial cytology, HPV tests were used in about 60% of the cases at the end of the study period. Given the promising results with HPV triage, its use should be encouraged among doctors and laboratories. A challenge with the delayed triage was that it likely reduced adherence to recommendations, i.e. some women did not make an appointment for the subsequent smear/hpv test. The analysis revealed that for more than 30% of the triage episodes the surveillance with new tests was not executed properly. This may reduce the benefits of triaging and lower the effectiveness of the screening programme. Better information to the women and their medical doctors might improve adherence to the recommended guidelines. The result, however, necessitates a discussion of pros and cons of delayed triage testing versus reflex testing in the programme. Conclusion Introduction of HPV testing in triage in the NCCSP in 2005 has resulted in a better risk stratification of women. The increase in referral to colposcopy seems acceptable. It also appears as if additional benefit could be gained if all physicians adhered to the current guidelines. Acknowledgements We thank Bente Kristin Johansen, previous leader of the NCCSP, for participating in planning of this study, Jan Nyg å rd, CRN for contributing to the data preparation for the study, and Bjarte Aagnes, CRN for assistance with the data analysis. Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper. References [1] Council of the European Union. Council recommendation of 2 December 2003 on cancer screening (2003/878/EC). Off J Eur Union 2003 ; L327 : 34 8. [2] IARC Handbooks of Cancer Prevention Volume 10. Cervix cancer screening. Lyon, France: IARC Press ; 2005. [3] European Commision. European Guidelines for Quality Assurance in Cervical Cancer Screening, 2nd edition. Luxembourg: Office for Official Publications of the European Communities ; 2008. [4] Jordan J, Arbyn M, Martin-Hirsch P, Schenck U, Baldauf JJ, Da SD, et al. European guidelines for quality assurance in cervical cancer screening: Recommendations for clinical management of abnormal cervical cytology, part 1. Cytopathology 2008 ; 19 : 342 54. [5] Saslow D, Solomon D, Lawson HW, Killackey M, Kulasingam SL, Cain J, et al. American Cancer Society, American Society for Colposcopy and Cervical Pathology, and American Society for Clinical Pathology screening guidelines for the prevention and early detection of cervical cancer. CA Cancer J Clin 2012 ; 62 : 147 72. [6] Ronco G, van Ballegooijen M, Becker N, Chil A, Fender M, Giubilato P, et al. Process performance of cervical screening programmes in Europe. Eur J Cancer 2009 ; 45 : 2659 70. [7] Anttila A, von Karsa L, Aasmaa A, Fender M, Patnick J, Rebolj M, et al. Cervical cancer screening policies and coverage in Europe. Eur J Cancer 2009 ; 45 : 2649 58. [8] Coleman D, Day N, Douglas G, Farmery E, Lynge E, Philip J, et al. European Guidelines for Quality Assurance in Cervical Cancer Screening. Europe against cancer programme. Eur J Cancer 1993 ; 29A(Suppl 4) : S1 38. [9] ASCUS-LSIL Triage Study (ALTS) Group. A randomized trial on the management of low-grade squamous intraepithelial lesion cytology interpretations. Am J Obstet Gynecol 2003 ; 188 : 1393 400. [10] ASCUS-LSIL Triage Study (ALTS) Group. Results of a randomized trial on the management of cytology interpretations of atypical squamous cells of undetermined significance. Am J Obstet Gynecol 2003 ; 188 : 1383 92. [11] Arbyn M, Buntinx F, Van Ranst M, Paraskevaidis E, Martin-Hirsch P, Dillner J. Virologic versus cytologic triage of women with equivocal Pap smears: A meta-analysis of the accuracy to detect high-grade intraepithelial neoplasia. J Natl Cancer Inst 2004 ; 96 : 280 93. [12] Arbyn M, Ronco G, Anttila A, Meijer CJ, Poljak M, Ogilvie G, et al. Evidence regarding human papillomavirus testing in secondary prevention of cervical cancer. Vaccine 2012 ; 30(Suppl 5) : F88 99. [13] Moss S, Gray A, Legood R, Vessey M, Patnick J, Kitchener H. Effect of testing for human papillomavirus as a triage during screening for cervical cancer: Observational before and after study. Br Med J 2006 ; 332 : 83 5.

Downloaded by [46.3.195.162] at 12:02 09 December 2017 [14] Kelly RS, Patnick J, Kitchener HC, Moss SM. HPV testing as a triage for borderline or mild dyskaryosis on cervical cytology: Results from the Sentinel Sites study. Br J Cancer 2011 ; 105 : 983 8. [15] Faglig r å dgivningsgruppe for Masseunders ø kelsen mot livmorhalskreft. Kvalitetsmanual. Masseunders ø kelsen mot livmorhalskreft. Oslo: Kreftregisteret ; 2005. [16] Nygard JF, Skare GB, Thoresen SO. The cervical cancer screening programme in Norway, 1992 2000 : Changes in Pap smear coverage and incidence of cervical cancer. J Med Screen 2002; 9 : 86 91. [17] Johansen BK, Bj ø rge T. Cervical cancer screening in Norway. In: Haldorsen T, editor. Cancer in Norway 2009. Special issue: Cancer screening in Norway. Oslo: Cancer Registry of Norway ; 2011. p. 118 29. [18] Solomon D, Davey D, Kurman R, Moriarty A, O Connor D, Prey M, et al. The 2001 Bethesda System: Terminology for reporting results of cervical cytology. JAMA 2002 ; 287 : 2114 9. [19] Nyg å rd M, R ø ysland K, Cambell S, Dillner J. Comparative effectiveness study on human papillomavirus detection methods used in the cervical cancer screening programme. HPV testing in triage 209 Br Med J Open 2013 ; 4 : e003460. doi:10.1136/bmjopen- 2013-003460. [20] Cleves MA, Gould WW, Gutierrez RG. An introduction to survival analysis using stata. Revised edition. College Station, Texas: Stata Pres s; 2004. [21] Stata: Release 11. Statistical Software. [computer program]. College Station, Texas: StataCorp LP; 2009. [22] Castle PE, Sideri M, Jeronimo J, Solomon D, Schiffman M. Risk assessment to guide the prevention of cervical cancer. J Low Genit Tract Dis 2008 ; 12 : 1 7. [23] Arbyn M, Roelens J, Martin-Hirsch P, Leeson S, Wentzensen N. Use of HC2 to triage women with borderline and mild dyskaryosis in the UK. Br J Cancer 2011 ; 105 : 877 80. [24] Cuzick J, Thomas CJ, Zhang G, Einstein MH, Stoler M, Trupin S, et al. Human papillomavirus testing for triage of women with low-grade squamous intraepithelial lesions. Int J Cancer 2013 ; 132 : 959 66. [25] Nygard JF, Sauer T, Nygard M, Skare GB, Thoresen SO. CIN 2/3 and cervical cancer in an organised screening programme after an unsatisfactory or a normal Pap smear: A seven-year prospective study of the Norwegian populationbased screening programme. J Med Screen 2004 ; 11 : 70 6.