IJC International Journal of Cancer
Specimen self-collection and HPV DNA screening in a pilot study of 100,242 women }rincz2, Leticia Torres1, Jorge Salmero n1,3, Aurelio Cruz1, Rosalba Rojas1, Eduardo Lazcano-Ponce1, Attila T. Lo 1 1 ndez and Mauricio Herna ndez Pilar Herna 1
blica de Mexico. Cuernavaca Morelos, Mexico Centro de Investigacion en Salud Poblacional, Instituto Nacional de Salud Pu Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine, Queen Mary University of London, London, United Kingdom 3 Unidad de Investigacion Epidemiologica y en Servicios de Salud, Instituto Mexicano del Seguro Social, Cuernavaca Morelos, Mexico 2
Key words: high-risk HPV testing, vaginal self-collected samples, cervical cancer, Mexico Abbreviations: ASe: adjusted sensitivity; ASp: adjusted specificity; CCSP: Mexican cervical cancer screening program; CI: confidence interval; CIN21: cervical intraepithelial neoplasia grade 2 or higher; CIN31: cervical intraepithelial neoplasia grade 3 or higher; HC2: Hybrid Capture 2; HPV: Human Papillomavirus; hrHPV: high-risk HPV; LR1: positive likelihood ratio; LR2: negative likelihood ratio; NPV: negative predictive value; PASS: Papillomavirus assessment in self-collected specimens; PPV: positive predictive value This article was published online on 7 December 2013. An error was subsequently identified. This notice is included in the online and print versions to indicate that both have been corrected 24 February 2014. Grant sponsor: Cancer Research UK; Grant number: C569/A10404; Grant sponsors: National Institute of Public Health of Mexico, the Health Ministry of Mexico, and Digene Corp. (now QIAGEN Corp.) DOI: 10.1002/ijc.28639 History: Received 13 May 2013; Accepted 11 Nov 2013; Online 28 Nov 2013 Correspondence to: Attila T. L}orincz, Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine, Queen Mary University of London, London, United Kingdom, Tel.: 1[44 (0)20 7882 3540], E-mail: [email protected]
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
High-income nations are increasingly adopting High-Risk HPV DNA (hrHPV) testing as an adjunct to primary cytology screening, since epidemiologic studies have shown that this strategy may enhance screening effectiveness.1–8 Cervical cancer remains an important problem in Mexico despite a decades-old cytology program that in 2010 screened approximately eight million women.9 Recognizing the poor effectiveness of the Mexican cytology program, the government health insurance system for low-income citizens “Seguro Popular” recently introduced an alternative cervical cancer screening program (CCSP) based on hrHPV testing for affiliated women aged 35–64 years. The goal of the new CCSP is to achieve 85% coverage of six million women over a period of 5 years.10 To date, according to the Mexican Women’s Cancer Information System (SICAM) almost three million eligible women have been tested for hrHPV nationwide. Cytological evaluation is used subsequently as a triage test in hrHPV positive women to identify those at higher risk of cervical intraepithelial neoplasia who must be referred for colposcopy.10 Prior research suggests that hrHPV screening will be effective in Mexico, including findings from large randomized controlled trials (RCT) and multiple local demonstration studies,11,12 including the MARCH RCT that validated vaginal self-sampling at home as clinically useful among participants in a state-sponsored poverty reduction program living in
Early Detection and Diagnosis
Since cervical cancer remains common in Mexico despite an established cytology screening program, the Ministry of Health recently introduced pilot front-line HPV testing into the Mexican cervical cancer screening program (CCSP). Here, we present the key field performance metrics of this population-based study. High-risk HPV DNA (hrHPV) testing was conducted on selfcollected vaginal specimens from 100,242 women aged 25–75 years residing in Morelos State. All hrHPV positive women and a random sample of 3.2% (n 5 2,864) of hrHPV negative participants were referred for colposcopic examination. The main disease endpoint of interest was cervical intraepithelial neoplasia grade 2 or higher (CIN21). We calculated relative risk, positive predictive value and negative predictive value adjusted for screening test verification bias. The overall prevalence of hrHPV was 10.8% (95%CI 10.6–11.0). Women positive for hrHPV had a relative risk of 15.7 for histologically detectable CIN21. The adjusted positive predictive value of the hrHPV test was 2.4% (95%CI 2.1–2.7); whereas the adjusted negative predictive value was 99.8% (95%CI 99.8–99.9). These findings suggest that large-scale vaginal hrHPV testing in a middle-income country can identify women at greater risk of advanced cervical abnormalities in a programmatically meaningful way but care is warranted to ensure that disease not detectable at colposcopy is kept to a minimum. PASS shows areas that need improvement and sets the stage for wider use of hrHPV screening of self-collected vaginal specimens in Mexico.
110
HPV DNA screening in self-collected specimens
Early Detection and Diagnosis
What’s new? DNA testing for high-risk human papillomavirus (hrHPV) in self-collected vaginal specimens may be useful for the routine detection of cervical precancers. In this investigation of 100,242 women in Mexico, DNA testing of self-collected vaginal specimens was found to have an adjusted positive predictive value of 2.4% and an adjusted negative predictive value of 99.8%. The study sheds light on potential scale-up problems associated with HPV DNA testing and reveals that a triage test is needed to prevent excessive referrals of women to limited colposcopy services.
underdeveloped areas of the country.13 The hrHPV testing of self-collected vaginal specimens is desirable because it is an efficient screening procedure that poses fewer barriers to access for women living in areas with little access to health services. Home-based collection of self-collected vaginal specimens for hrHPV screening has been shown to significantly increase participation in screening in comparison with clinic-based selfcollection in previous studies.14 At the same time, it has the advantage of reducing the need for screening clinics and can free-up trained personnel to concentrate on more effective follow-up of women who test positive.15 Women generally had good acceptance of self-sampling, although concerns around obtaining a proper sample collection by women was a common concern and seems to be the reason why women preferred self-sampling at a clinic than at home.16 Even clinicbased vaginal self-sampling is less demanding on professional staff and eliminates the need for insertion of a speculum; many studies have shown that vaginal self-sampling is more acceptable to women than speculum-based screening.17–21 We carried out the PASS (Papillomavirus assessment in self-collected specimens) study to evaluate the performance, feasibility and logistics of hrHPV DNA testing on selfcollected vaginal samples while referring all positive women to colposcopy, at the population level under real “Seguro Popular” screening program conditions in the State of Morelos, Mexico. The PASS study employed hrHPV DNA screening of vaginal specimens among women attending clinics participating in the “Seguro Popular” CCSP program. The study’s aim was to collect evidence on the performance of hrHPV testing in a large regional pilot program to identify and refine the most effective strategy for employing hrHPV testing nationwide. Our long-term goal is to create a screening program that will minimize inequities in the quality of screening tests offered throughout Mexico.
Material and Methods Patients and eligibility
The study began in February 2007 and patient follow-up ended in July 2010. The target population for screening using self-collected samples was 150,535 women between 25 and 75 years of age who had “Seguro Popular” insurance coverage in the State of Morelos.22 “Seguro Popular” was created in 2003 to address healthcare needs of Mexicans without formal employment. The program seeks to provide health service coverage through voluntary public insurance for low-income persons who are not covered by other public insurance
schemes, regardless of employment status.23,24 A total of 121,650 women from the target population were invited to participate using the recruitment strategies described below. A total of 9,869 women were excluded because they had a prior cervical cancer screening in the previous year, an additional 893 women were excluded because they had a hysterectomy or were currently pregnant. Of the 10,646 (9.6%) women who refused to participate in the study, some declined because they simply were not interested whereas others preferred a cervical cancer screening service where a clinician obtained a cervical specimen for hrHPV testing. Thus, the final sample size for the study was 100,242 women who underwent hrHPV testing by vaginal self-sampling at local screening clinics. There were 10,863 women who tested positive for hrHPV and they were contacted by up to three visits to their homes, 8,267 (76.1%) subsequently attended colposcopy services. In addition, a randomly selected control group of 2,864 women from the hrHPV negative group (89,379) were also referred to colposcopy, 2,399 of whom were successfully evaluated, offering the necessary information for partial verification-bias adjustment.25 Settings and data collection sites
Over an eighteen-month period, the PASS study employed three simultaneous recruitment strategies. The first took place at homes and used a list of the potential population affiliated with “Seguro Popular.” Twenty community brigades (groups of at least two nurses, who also provided information on the benefits and procedure of specimen self-collection) offered vaginal self-sampling at home in 754 communities. In the second recruitment strategy, female medical service users of all 288 “Seguro Popular” primary health care centers were invited to participate in vaginal self-sampling at the health center with the same sampling method used for the first group at home. A third group of women were invited while they were attending health meetings during local government-organized health weeks. These national health weeks implement multiple disease prevention activities, including vaccination of children and adults. During these weeks, radio and television advertisements described the project, explained the usefulness of HPV testing and promoted community participation. Self collection of vaginal specimens
Collection of specimens followed the standard procedures for self-collection of vaginal samples that have been described in C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
previous studies.13 The standard Digene conical collection brush (Qiagen Inc., Gaithersburg, MD) was used to obtain the vaginal specimens. Women were instructed to grasp the brush midway along the shaft, insert the brush head as far as possible into the vagina and rotate it to the left and right. They were then instructed to place the brush immediately into the Digene cervical sampler tube that contains specimen transport medium (STM) and seal the tube with the provided cap. Specimens were collected daily and sent to the Cancer Prevention and Control Program’s coordination office in Cuernavaca, Morelos, transported weekly to the HPV laboratory at the Instituto Nacional de Salud P ublica (INSP) in Cuernavaca, Morelos, Mexico. hrHPV DNA detection by hybrid capture 2 (HC2)
The hrHPV DNA was detected with a standard commercial kit employing in vitro nucleic acid hybridization and chemiluminescent microplate-based signal amplification. This HC2 test has been described in detail by Lorincz,26 and was conducted in a validated and accredited routine HPV testing laboratory at the INSP, which is formally recognized as a regional center of expertise in CCSP HPV testing in central Mexico. HC2 RLU/CO cutoff of 1 was considered a positive test. The hrHPV positive women and a randomly selected subset of 3.2% of hrHPV negative participants serving as controls were referred to colposcopic examination for diagnosis confirmation. Colposcopy and histological confirmation
Six colposcopy clinics from the Morelos CCSP performed all colposcopy procedures. They were blinded to hrHPV status, although they knew that most of the women evaluated were hrHPV positive. Lesions were graded using the Reid Index,27 and colposcopy-directed biopsies were collected from areas of suspected cervical abnormalities. Biopsies were reviewed and diagnosed independently by two pathologists who were blinded to the colposcopy data and hrHPV testing results. If there was disagreement, a third pathologist did a reading and the final diagnosis was rendered by majority agreement. CIN21 cases were treated according to Mexico’s Cervical Cancer Screening Program guidelines.28
jects verified by histology, after which the values of the adjusted variables are calculated in the standard way. However, the method of Begg and Greenes cannot correct for an inability of the reference standard (colposcopy) to detect all disease; hence, we regard it as a partial verification bias adjustment. An approach to full verification bias correction is realistic only with long follow-up, multiple applications of the best reference methods, and the ethical clarity and permission to apply them. Crude and adjusted positive and negative predictive values (PPV and NPV) were calculated for hrHPV as a function of the probability of presence for each of CIN21 and CIN31 lesions among positive or negative hrHPV groups respectively. It is obvious that these estimates are also biased by the observed prevalence of disease and cannot be fully corrected because they are directly dependent on both the true prevalence of disease and the ability of downstream clinical process to detect and classify disease correctly. We also calculated adjusted positive and negative Likelihood Ratios (LR): 1LR 5 ASe/ (1-ASp); 2LR 5 (1-ASe/ ASp). A value of LR 5 1 indicates that the test result is equally likely in patients with and without the disease, values of 1LR > 1 and 2LR < 1 indicate the magnitude to which test results are likely to occur in patients with disease or without disease respectively, whereas LR1/LR2 gives the fold improvement for detection of the endpoint (CIN21 or CIN31) in women who are hrHPV1 versus hrHPV2. Unadjusted relative risk was calculated directly by dividing the proportion of disease in hrHPV positive women by the proportion of disease among the hrHPV negative control women. Power calculations were not conducted because the study was designed as a pilot implementation to be much larger than other adequately powered studies with similar endpoints.31 Calculations were conducted using Stata version 11.0 (Stata Corp. College Station). Ethics
The study was approved by the Research, Biosafety, and Ethics Committees at the National Institute of Public Health of Mexico, and was assigned registry number 657. All patients in the study gave informed consent.
Statistical analysis
Results
Our primary goal was to determine whether hrHPV testing, using the standard cutoff of 1 pg/ml for HC2 in routine implementation, would perform similarly to previously successful, adequately powered, smaller studies.11 We calculated the partial verification-bias of the hrHPV screening test, due to nonreferral of hrHPV negative women, using the method of Begg and Greenes; this method corrects for differential incomplete assessment of CIN21 and allowed us to derive an estimate of the adjusted sensitivity (ASe) and adjusted specificity) (ASp) for CIN21 or CIN31.29,30 The method initially adjusts the observed test counts in each contingency cell by dividing the subject count by the proportion of sub-
The study included 100,242 “Seguro Popular” affiliated women in the State of Morelos (Table 1). The overall prevalence of hrHPV DNA was 10.8% (95% CI 10.6–11.0). Among the 8,267 hrHPV positive women who attended the colposcopy service, 162 were histologically diagnosed with CIN21 (1.96%), which included 31 cases of CIN2, 76 of CIN3, and 55 invasive cancers. Among the control group of 2,399 hrHPV negative women, only three cases of CIN21 (0.125%; 2 cases of CIN2 and 1 case of CIN3) were identified (Fig. 1). Owing to the design of our study we were unable to calculate an unbiased estimate of clinical sensitivity; however, we provide estimates of ASe and ASp in Table 2. Comparing
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
Early Detection and Diagnosis
111
Lazcano-Ponce et al.
112
HPV DNA screening in self-collected specimens
Table 1. Prevalence of hrHPV and CIN21 among 100,242 Mexican women aged 25–75 years, stratified by participants’ characteristics 1
Total N 5 100,242
hrHPV prevalence2 positive (n 5 10,863)
CIN21 prevalence2 N 5 165
n
%
%
95% CI
%
95% CI
25–29
1,106
1.1
18.7
16.4–21.0
0.36
0.01–0.72
30–34
13,905
13.9
12.5
11.9–13.0
0.14
0.08–0.21
35–39
21,406
21.4
10.3
9.9–10.7
0.14
0.09–0.18
40–44
16,474
16.4
9.7
9.2–10.1
0.10
0.05–0.15
45–49
13,794
13.8
9.4
8.9–9.9
0.19
0.12–0.26
50–54
11,016
11.0
9.8
9.2–10.3
0.22
0.13–0.30
55–59
8,069
8.1
10.8
10.1–11.4
0.35
0.22–0.48
60–64
5,519
5.5
12.3
11.4–13.2
0.09
0.01–0.17
65–69
3,897
3.9
14.0
12.9–15.0
0.21
0.06–0.35
70–75
5,056
5.0
13.1
12.1–14.0
0.08
0.00–0.16
Homemaker/informal employee
65,404
65.3
9.5
9.2–9.7
0.14
0.11–0.17
Formal employee
34,240
34.2
12.4
12.0–12.7
0.19
0.14–0.24
Never
10,707
10.7
12.5
11.8–13.1
0.21
0.13–0.30
Ever
89,513
89.3
10.6
10.4–10.8
0.16
0.13–0.18
1 sexual partner
82,736
82.5
7.1
7.0–7.3
0.08
0.06–0.10
2 or more sexual partners
17,486
17.4
28.3
27.7–29.0
0.55
0.44–0.66
Urban
65,006
64.9
10.9
10.7–11.2
0.16
0.13–0.19
Suburban
19,504
19.5
9.8
9.6–10.6
0.20
0.14–0.26
Rural
15,134
15.1
9.6
9.5–10.4
0.11
0.05–0.16
Cervical clinician collected
13,714
13.7
10.1
9.6–10.6
0.13
0.07–0.19
Self sampling
85,930
85.7
10.5
10.3–10.7
0.16
0.13–0.19
Variable Age groups (years)
Occupation
Pap test history
Sexual partner history
Early Detection and Diagnosis
Place of Residence
Screening test preference for use in the future
Cells that do not add up to 100% are due to missing values. The prevalence of CIN21 shown in each row category was computed from the number of cases of CIN21 in each row category divided by the total of participants in each row category. 1 Percentages on the total are based on the column. 2 Percentages for hrHPV and CIN21 are based on the rows. Abbreviations: CIN21: cervical intraepithelial neoplasia grade 2 or worse; hrHPV: high-risk human papillomavirus.
hrHPV positive versus hrHPV negative women, the relative risk of CIN21 was 15.67 (0.01959/0.00125). Figure 2 shows the age-specific hrHPV DNA prevalence for the PASS study when compared with the age stratified distribution of hrHPV in vaginal self-sample specimens obtained from one of our previous studies in Morelos.32 In both groups we observed two peaks of hrHPV prevalence, the first among women aged 30–34 years, and a second among women over the age of 55 (Fig. 2). Participants with formal employment (12.4% vs. 9.5%), as well as women having a history of 2 or more lifetime sexual partners (28.3% vs. 7.1%) had a significantly higher rate of hrHPV detection. The hrHPV prevalence was also significantly higher in the group of urban participants (10.9%) in
comparison with rural women at 9.6% (Table 1). Verification bias corrected estimates as well as the crude estimates for sensitivity and specificity, PPV and NPV are shown in Table 2. The partial verification-bias-corrected estimate of adjusted sensitivity for CIN21 was 64.4% (95%CI 59.5–69.1). The corrected estimate for adjusted specificity in women 25 years was 89.4% (95%CI 89.2–89.6) for CIN21 (Table 2). ). The estimates were similar across different age groups [25 and 35 years (data not shown)]. High adjusted negative predictive values were observed for both CIN21 (99.8%) and CIN31 (99.9%). The positive likelihood ratios of hrHPV testing for CIN21 and CIN31 detection were 6.1 and 7.7, respectively (Table 2). Negative likelihood ratios for CIN21 and CIN31 detection were 0.40 and 0.21, respectively. C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
113
Early Detection and Diagnosis
Lazcano-Ponce et al.
Figure 1. Flowchart of the screening profiles of women in the PASS study. The target population was all women residing in the state of Morelos who were affiliated with the Seguro Popular insurance program, directed to address the healthcare needs of Mexicans without formal employment. The study protocol specified that all formally enrolled high-risk human papillomavirus (hrHPV) DNA–positive women would be referred to colposcopy. In addition, 3.2% of the hrHPV negative women were also sent to colposcopy to act as controls. In all, 1.96% (n 5 162) of the 8,267 hrHPV positive women who completed colposcopy had a histopathological diagnosis of CIN21; in comparison 0.125% (3) of women who completed colposcopy with negative hrHPV were found to have CIN21.
The proportion of biopsies collected from hrHPV positive women referred to different colposcopy clinics varied from 41.3% (111/269) to 9.1% (228/2496) (Table 3).
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
In addition, there was a clear positive correlation between the proportion of biopsies collected in the hrHPV positive women by each colposcopy clinic and the proportion of
114
HPV DNA screening in self-collected specimens
Table 2. Crude and verification bias adjusted estimates of test performance characteristics for hrHPV testing on self-collected vaginal specimens as primary screening of CIN21 and CIN31 among Mexican women aged 25–75 years Crude estimates CIN
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
LR1
95% CI
LR2
95% CI
21
98.2%
94.8–99.6
89.3%
89.1–89.5
1.5%
1.3–1.7
100%
100–100
9.2
8.9–9.4
0.02
0.006–0.06
31
99.2%
95.9–100
89.3%
89.1–89.5
1.2%
1.0–1.4
100%
100–100
9.2
9.0–9.5
0.008
0.001–0.06
Verification bias adjusted estimates CIN
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
LR1
95% CI
LR2
95% CI
21
64.4%
59.5–69.1
89.4%
89.2–89.6
2.4%
2.1–2.7
99.8%
99.8–99.9
6.1
5.6–6.5
0.40
0.35–0.45
31
81.5%
76.2–86.1
89.3%
89.1–89.5
1.9%
1.7–2.2
99.9
99.9–100
7.7
7.2–8.1
0.21
0.16–0.27
Early Detection and Diagnosis
The adjusted estimates are based on histological results of a large set of women randomly selected from among the hrHPV negative participants. This allowed adjustment for verification bias caused by preferential colposcopy referral of hrHPV positive cases. Corrected estimates were computed according to Begg CB, Greenes RA30
Figure 2. Comparison of hrHPV prevalence in women stratified by age in the years 2000 and 2007, in two studies from the state of Morelos. The patterns of hrHPV prevalence by age were quite similar in these two studies from the same geographic area conducted 7 years apart. The hrHPV prevalence was high, reaching almost 13% in women aged 30–34 years. The prevalence decreased significantly in women past the childbearing age until a nadir of 9% in women aged 40–49 years. The hrHPV prevalence increased again in women aged 65–69 to 15%.
CIN21 cases detected. For example, Table 3 compares 41.3% of hrHPV positive women biopsied and 4.1% of CIN21 cases detected, versus 9.1% women biopsied and 1.2% CIN21 detected.
Discussion Our study describes the first large-scale introduction of hrHPV testing into a middle-income country’s routine CCSP. The pattern of hrHPV prevalence by age was similar to that found in a previous study in the same geographic area conducted almost 7 years earlier; hrHPV prevalence was high in younger women, decreased significantly in women past the childbearing age until 49 years, and increased again in older women12 (Fig. 2). The hrHPV testing of self-collected vaginal samples demonstrates a high NPV for relevant cervical disease, highlighting this test’s benefit as a primary screening procedure in
developing countries, as women who had a negative hrHPV result have virtually no risk of CIN31 lesions. Self-collected hrHPV testing may be particularly attractive for increasing screening intervals, and it appears to be an acceptable screening option in large-scale routine settings. In addition, given the magnitude of the relative risk and positive likelihood ratio, our findings demonstrate that hrHPV testing of self-collected samples is a satisfactory screening test for identification of women with target CIN21 lesions. Increasing screening coverage with hrHPV testing of selfcollected vaginal samples is appealing because this system is easier to implement in settings lacking infrastructure and has also been shown to be more sensitive than routine cervical cytology.13 However, if routine testing of self-collected vaginal specimens is to be used for primary screening, we need to account for potential problems with that system. For example, the logistical difficulties posed by Mexico’s current system are exemplified by the fact that 23.9% of hrHPV positive participants did not receive colposcopy. The relatively low follow-up rate is due to a number of factors, primarily difficulties regarding transportation, clinic hours, costs, and child care needs that impede women’s access to health services. Other factors include lack of clinician recommendations and information, a culture of reluctance, and high levels of migration or change of residence in deprived Mexican populations. Thus, an effective screening program must include a system for comprehensive follow-up of hrHPV positive women regardless of whether the specimens are self or clinician-collected. The variation we found in follow-up colposcopy also suggests that directly referring all hrHPV positive women to colposcopy can cause additional problems in a CCSP by over-burdening the professional services. We observed a wide variation in the detection rate of CIN21 (4.1 vs. 1.2) among the colposcopy clinic participants. The major reason for this variation was likely the different number of biopsies taken among the colposcopy groups (41.3– 9.1%). Our data suggest that lack of standardization in biopsy collection could be an important cause of failure in the Mexican CCSP, and improvements are needed if we are to realize C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
115
Lazcano-Ponce et al.
Table 3. Colposcopy performance in the PASS study to detect CIN21 among Mexican women aged 25–75 years Evaluated women
Proportion of biopsies collected
CIN21 detected
Colposcopy clinics
HPV pos (n 5 8,267)
HPV neg (n 5 2,399)
HPV pos (n 5 1081)
HPV neg (n 5 245)
HPV pos (n 5 162) [95%CI]
HPV neg (n 5 3)
1
269
115
41.3% (111)
30.4%
4.1% (11)
0% (0)
(35)
[1.7–6.5]
2
1,091
628
18.1% (197)
9.9%
3.3% (36)
(62)
[2.2–4.4]
3
553
247
16.5% (91)
10.9%
2.7% (15)
(27)
[1.4–4.1]
4
2,839
261
11.9% (339)
8.8% (23)
1.9% (53)
5
1019
523
11.3% (115)
8.6% (45)
6
2496
625
9.1% (228)
8.5% (53)
0.2% (1)
0% (0)
0% (0)
[1.4–2.4] 1.7% (17)
0.2% (1)
[0.9–2.5] 1.2% (30)
0.2% (1)
the full benefits of hrHPV DNA testing. Colposcopy can be improved if technical accreditation of practitioners is required and by introducing the practice of multiple biopsies. There has been debate over how many biopsies are needed, with up to four in total being recommended for screenpositive women.33 It is reasonable to suggest that at least one biopsy should be systematically collected from the most suspicious area of squamocolumnar junction in all hrHPV positive women triaged to colposcopy; this may substantially increase the cervical disease detection rate and would also increase the apparent PPV of an hrHPV positive test.33,34 The costs arising from the collection of multiple biopsies can be controlled if these biopsies are included and processed in the same paraffin block. In contrast with our previous studies, the estimated adjusted PPV for detection of CIN21 was very low. We observed only a 2.4% probability of detecting CIN21 lesions among women positive for hrHPV, meaning that hrHPV testing identified a large number of women who, by standard interpretation, did not require further evaluation to exclude high-grade CIN abnormalities; however, this low PPV needs to be considered in the context of the accuracy of colposcopy. It is likely that less evident lesions were often missed and biopsies were mainly collected in the more visible advanced cervical disease cases; this is also reflected in the relatively low number of CIN2 cases in our study when compared with CIN3 or cancer cases. As previously reported, overloading colposcopy services leads to lower biopsy collection rates.35 Our results suggest that colposcopy should be deployed only after other conventional triage alternatives like cervical cytology, genotyping for HPV16 (possibly also HPV18) or the use of biomarkers. Another alternative is to use a higher threshold of the HC2 test. For example, in our earlier MARCH study a HC2 RLU/CO cutoff of two gave a PPV of 20% and a cutoff of five gave a PPV of 26%, both these cut-
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
offs were associated with acceptable detection rates of CIN21 when compared with routine cytology.13 It is possible that a higher threshold of five for HC2 is a reasonable alternative to genotyping for HPV16 and 18, giving similar sensitivity and PPV.36 For both genotyping or using an HC2 RLU/CO of 5, care needs to be taken to ensure good followup and possibly frequent repeat testing for the women who are hrHPV positive for types other than 16 or 18, or are HC2 positive in the RLU/CO range of less than 5. The main dilemma in middle-income regions is how to improve both the quality and accessibility of triage after a positive HPV DNA screening test. In the current Mexican CCSP implemented within the “Seguro Popular”, guidelines for HPV testing-based screening recommend using the Pap test only in hrHPV2positive women. The Pap test must be of high quality and in particular also highly sensitive (90% or better) for triage to be effective, which is a major challenge in most developing countries. The hrHPV testing as primary screening may decrease the numbers of necessary cytology evaluations by close to 90%, thus allowing cytologists to focus more carefully on the remaining specimens enriched for abnormalities, as well as permitting the use of cytology double-reading procedures in order to improve the sensitivity and overall quality of cytology evaluation. However, this triage alternative is not well established and may require further evaluations in developing countries. Ideally, triage and follow-up would be conducted in the context of an organized registry-based program that could apply appropriate patient call and recall procedures. Testing cervical lesions for carcinogenic HPV types including 16, 18 and 45 may help identify cases that are more likely to progress to cancer.36 Currently, several triage alternatives for hrHPV positive women are being evaluated, focusing on HPV genotyping, E6 protein detection, and incorporation of cytological biomarkers such as p16 and Ki-67 or DNA
Early Detection and Diagnosis
[0.8–1.6]
116
HPV DNA screening in self-collected specimens
methylation.37 Meanwhile, we are striving to both improve our CCSP infrastructure and to modify existing practices by scaling up the new, currently validated strategies that appear appropriate to our socio-cultural context and technological capabilities.
Acknowledgements Drs. Enrique Carmona Balandrano, Patricia Alonso and the Association of Colposcopy of Morelos State, Mexico provided pathology and colposcopy expertise. The authors gratefully recognize the patients who participated in this study.
References 1.
2.
3.
4.
5.
Early Detection and Diagnosis
6.
7.
8.
9.
10.
11.
12.
13.
Szarewski A. Ushering in a new era: human papillomavirus (HPV) testing comes to the NHS cervical screening programme. J Fam Plann Reprod Health Care 2011;37:64–7. Ronco G, Giorgi-Rossi P, Carozzi F, et al. Efficacy of human papillomavirus testing for the detection of invasive cervical cancers and cervical intraepithelial neoplasia: a randomised controlled trial. Lancet Oncol 2010;11: 249–57. Bulkmans NW, Berkhof J, Rozendaal L, et al. Human papillomavirus DNA testing for the detection of cervical intraepithelial neoplasia grade 3 and cancer: 5-year follow-up of a randomised controlled implementation trial. Lancet 2007;370:1764–72. Mayrand MH, Duarte-Franco E, Rodrigues I, et al; Canadian Cervical Cancer Screening Trial Study Group. Human papillomavirus DNA versus Papanicolaou screening tests for cervical cancer. N Engl J Med 2007;357:1579–88. Naucler P, Ryd W, T€ ornberg S, et al. Human papillomavirus and Papanicolaou tests to screen for cervical cancer. N Engl J Med. 2007;357: 1589–97. Sankaranarayanan R, Nene BM, Shastri SS, et al. HPV screening for cervical cancer in rural India. N Engl J Med 2009;360:1385–94. Kitchener HC, Almonte M, Thomson C, et al. HPV testing in combination with liquid-based cytology in primary cervical screening (ARTISTIC): a randomised controlled trial. Lancet Oncol 2009;10:672–82. Anttila A, Kotaniemi-Talonen L, Leinonen M, et al. Rate of cervical cancer, severe intraepithelial neoplasia, and adenocarcinoma in situ in primary HPV DNA screening with cytology triage: randomised study within organised screening programme. BMJ 2010;340:c1804. Hernandez-Avila M, Lazcano-Ponce E, Martınez O. Public policies for HPV screening and vaccination in Mexico. In the 27th International Papillomavirus Conference and Clinical Workshop. Berlin, Germany, September 17–23, 2011. Secretarıa de Salud. Programa de acci on especıfico 2007–2012 Cancer Cervicouterino, 1st edn. Mexico, D.F.: Secretarıa de Salud, 2008. Lazcano-Ponce E, L€ orincz AT, Salmer on J, et al. A pilot study of HPV DNA and cytology testing in 50,159 women in the routine Mexican social security program. Cancer Causes Control 2010;21: 1693–700. Salmeron J, Lazcano-Ponce E, Lorincz A, et al. Comparison of HPV2based assays with Papanicolaou smears for cervical cancer screening in Morelos State, Mexico. Cancer Causes Control 2003;14:505–12. Lazcano-Ponce E, Lorincz AT, Cruz-Valdez A, et al. Self-collection of vaginal specimens for human papillomavirus testing in cervical cancer prevention (MARCH): a community-based
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
randomised controlled trial. Lancet 2011;378: 1868–73. Sowjanya AP, Paul P, Vedantham H, et al. Community Access to Cervical Health Study Group. Suitability of self-collected vaginal samples for cervical cancer screening in periurban villages in Andhra Pradesh, India. Cancer Epidemiol Biomarkers Prev 2009;18:1373–8. Gravitt PE, Belinson JL, Salmeron J, et al. Looking ahead: a case for human papillomavirus testing of self-sampled vaginal specimens as a cervical cancer screening strategy. Int J Cancer 2011;129:517–27. Petignat P, Vassilakos P. Is It Time to Introduce HPV Self-Sampling for primary cervical cancer screening? JNCI 2012;194:166–7. Waller J, McCaffery K, Forrest S, et al. Acceptability of unsupervised HPV self-sampling using written instructions. J Med Screen 2006;13:208–13. Mitchell S, Ogilvie G, Steinberg M, et al. Assessing women’s willingness to collect their own cervical samples for HPV testing as part of the ASPIRE cervical cancer screening project in Uganda. Int J Gynaecol Obstet 2011;114: 111–5. Sellors JW, Lorincz AT, Mahony JB, et al. Comparison of self-collected vaginal, vulvar and urine samples with physician-collected cervical samples for human papillomavirus testing to detect highgrade squamous intraepithelial lesions. CMAJ 2000;163:513–8. Stewart DE, Gagliardi A, Johnston M, et al. Selfcollected samples for testing of oncogenic human papillomavirus: a systematic review. J Obstet Gynaecol Can. 2007;29:817–28. Dzuba IG, Dıaz EY, Allen B, et al. The acceptability of self-collected samples for HPV testing vs. the pap test as alternatives in cervical cancer screening. Womens Health Gend Based Med 2002;11:265–75. National Institute of Statistics and Geography. Data bank. Interactive data consulting. Mexico Population and Housing Census 2010. Female population aged over 12 by age and Seguro Popular affiliation. National Institute of Statistics and Geography, 2010. Available at: http:// www.inegi.org.mx/lib/olap/consulta/general_ver4/ MDXQueryDatos.asp?#Regreso&c527874; in Spanish. Last accessed on May 31 2012, Mexico City. G omez Dantes O, Sesma S, Becerril VM, et al. The health system of Mexico. Salud Publica Mex. 2011;53:s220–s32. Knaul FM, Gonzalez-Pier E, G omez-Dantes O, et al. The quest for universal health coverage: achieving social protection for all in Mexico. Lancet 2012;380:1259–79. Kupets R, Paszat L. Physician and patient factors associated with follow up of high grade dysplasias of the cervix: a population-based study. Gynecol Oncol 2011;120:63–7.
26. Lorincz, AT. Hybrid CaptureTM method for detection of human papillomavirus DNA in clinical specimens: a tool for clinical management of equivocal Pap smears and for population screening. J Obstet Gynaecol Res 1996;22:629–636. 27. Reid R. Colposcopy of cervical preinvasive neoplasia. In: Singer A, ed. Premalignant lesions of the lower genital tract, vol. 2. New York: Elsevier, 1990. 87–116. 28. United Mexican States-Ministry of Health. Amendment to Mexican Official Standard NOM-014-SSA2–1994, for the prevention, treatment and management of cervical cancer and breast cancer in primary care, to read as NOM014-SSA2-1994, for the prevention, detection diagnosis, treatment, control and surveillance of cervical cancer. United Mexican States-Ministry of Health, 1995. Available at: http://www.salud. gob.mx/unidades/cdi/nom/m014ssa24.html; in Spanish. Last accessed on Oct 7, 2011, Mexico City. 29. Cronin AM, Vickers AJ. Statistical methods to correct for verification bias in diagnostic studies are inadequate when there are few false negatives: a simulation study. BMC Medical Research Methodol 2008;8:75. 30. Begg CB, Greenes RA. Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics 1983;39:207–215. 31. Belinson JL, Du H, Yang B, et al. Improved sensitivity of vaginal self-collection and high-risk human papillomavirus testing. Int J Cancer 2012; 130:1855–60. 32. Flores Y, Bishai D, Lazcano E, et al. Improving cervical cancer screening in Mexico: results from the Morelos HPV Study. Salud Publica Mex 2003; 45:S388–S98. 33. Stoler MH, Vichnin MD, Ferenczy A, et al. The accuracy of colposcopic biopsy: Analyses from the placebo arm of the Gardasil clinical trials. Int J Cancer 2011;128:1354–62. 34. Pretorius RG, Peterson P, Azizi F, et al. Subsequent risk and presentation of cervical intraepithelial neoplasia (CIN) 3 or cancer after a colposcopic diagnosis of CIN 1 or less. Am J Obstet Gynecol 2006;195:1260–5. 35. Arbyn M, Simoens C, Van Oyen H, et al. Analysis of 13 million individual patient records pertaining to Pap smears, colposcopies, biopsies and surgery on the uterine cervix (Belgium, 1996– 2000). Prev Med 2009;48:438–43. 36. Khan MJ, Castle PE, Lorincz AT, et al. The elevated 10-year risk of cervical precancer and cancer in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of typespecific HPV testing in clinical practice. J Natl Cancer Inst 2005;97:1072–9. 37. Galgano MT, Castle PE, Atkins KA, et al. Using biomarkers as objective standards in the diagnosis of cervical biopsies. Am J SurgPathol 2010;34: 1077–87.
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
Specimen self-collection and HPV DNA screening in a pilot study of 100,242 women }rincz2, Leticia Torres1, Jorge Salmero n1,3, Aurelio Cruz1, Rosalba Rojas1, Eduardo Lazcano-Ponce1, Attila T. Lo 1 1 ndez and Mauricio Herna ndez Pilar Herna 1
blica de Mexico. Cuernavaca Morelos, Mexico Centro de Investigacion en Salud Poblacional, Instituto Nacional de Salud Pu Centre for Cancer Prevention, Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine, Queen Mary University of London, London, United Kingdom 3 Unidad de Investigacion Epidemiologica y en Servicios de Salud, Instituto Mexicano del Seguro Social, Cuernavaca Morelos, Mexico 2
Key words: high-risk HPV testing, vaginal self-collected samples, cervical cancer, Mexico Abbreviations: ASe: adjusted sensitivity; ASp: adjusted specificity; CCSP: Mexican cervical cancer screening program; CI: confidence interval; CIN21: cervical intraepithelial neoplasia grade 2 or higher; CIN31: cervical intraepithelial neoplasia grade 3 or higher; HC2: Hybrid Capture 2; HPV: Human Papillomavirus; hrHPV: high-risk HPV; LR1: positive likelihood ratio; LR2: negative likelihood ratio; NPV: negative predictive value; PASS: Papillomavirus assessment in self-collected specimens; PPV: positive predictive value This article was published online on 7 December 2013. An error was subsequently identified. This notice is included in the online and print versions to indicate that both have been corrected 24 February 2014. Grant sponsor: Cancer Research UK; Grant number: C569/A10404; Grant sponsors: National Institute of Public Health of Mexico, the Health Ministry of Mexico, and Digene Corp. (now QIAGEN Corp.) DOI: 10.1002/ijc.28639 History: Received 13 May 2013; Accepted 11 Nov 2013; Online 28 Nov 2013 Correspondence to: Attila T. L}orincz, Wolfson Institute of Preventive Medicine, Barts and The London School of Medicine, Queen Mary University of London, London, United Kingdom, Tel.: 1[44 (0)20 7882 3540], E-mail: [email protected]
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
High-income nations are increasingly adopting High-Risk HPV DNA (hrHPV) testing as an adjunct to primary cytology screening, since epidemiologic studies have shown that this strategy may enhance screening effectiveness.1–8 Cervical cancer remains an important problem in Mexico despite a decades-old cytology program that in 2010 screened approximately eight million women.9 Recognizing the poor effectiveness of the Mexican cytology program, the government health insurance system for low-income citizens “Seguro Popular” recently introduced an alternative cervical cancer screening program (CCSP) based on hrHPV testing for affiliated women aged 35–64 years. The goal of the new CCSP is to achieve 85% coverage of six million women over a period of 5 years.10 To date, according to the Mexican Women’s Cancer Information System (SICAM) almost three million eligible women have been tested for hrHPV nationwide. Cytological evaluation is used subsequently as a triage test in hrHPV positive women to identify those at higher risk of cervical intraepithelial neoplasia who must be referred for colposcopy.10 Prior research suggests that hrHPV screening will be effective in Mexico, including findings from large randomized controlled trials (RCT) and multiple local demonstration studies,11,12 including the MARCH RCT that validated vaginal self-sampling at home as clinically useful among participants in a state-sponsored poverty reduction program living in
Early Detection and Diagnosis
Since cervical cancer remains common in Mexico despite an established cytology screening program, the Ministry of Health recently introduced pilot front-line HPV testing into the Mexican cervical cancer screening program (CCSP). Here, we present the key field performance metrics of this population-based study. High-risk HPV DNA (hrHPV) testing was conducted on selfcollected vaginal specimens from 100,242 women aged 25–75 years residing in Morelos State. All hrHPV positive women and a random sample of 3.2% (n 5 2,864) of hrHPV negative participants were referred for colposcopic examination. The main disease endpoint of interest was cervical intraepithelial neoplasia grade 2 or higher (CIN21). We calculated relative risk, positive predictive value and negative predictive value adjusted for screening test verification bias. The overall prevalence of hrHPV was 10.8% (95%CI 10.6–11.0). Women positive for hrHPV had a relative risk of 15.7 for histologically detectable CIN21. The adjusted positive predictive value of the hrHPV test was 2.4% (95%CI 2.1–2.7); whereas the adjusted negative predictive value was 99.8% (95%CI 99.8–99.9). These findings suggest that large-scale vaginal hrHPV testing in a middle-income country can identify women at greater risk of advanced cervical abnormalities in a programmatically meaningful way but care is warranted to ensure that disease not detectable at colposcopy is kept to a minimum. PASS shows areas that need improvement and sets the stage for wider use of hrHPV screening of self-collected vaginal specimens in Mexico.
110
HPV DNA screening in self-collected specimens
Early Detection and Diagnosis
What’s new? DNA testing for high-risk human papillomavirus (hrHPV) in self-collected vaginal specimens may be useful for the routine detection of cervical precancers. In this investigation of 100,242 women in Mexico, DNA testing of self-collected vaginal specimens was found to have an adjusted positive predictive value of 2.4% and an adjusted negative predictive value of 99.8%. The study sheds light on potential scale-up problems associated with HPV DNA testing and reveals that a triage test is needed to prevent excessive referrals of women to limited colposcopy services.
underdeveloped areas of the country.13 The hrHPV testing of self-collected vaginal specimens is desirable because it is an efficient screening procedure that poses fewer barriers to access for women living in areas with little access to health services. Home-based collection of self-collected vaginal specimens for hrHPV screening has been shown to significantly increase participation in screening in comparison with clinic-based selfcollection in previous studies.14 At the same time, it has the advantage of reducing the need for screening clinics and can free-up trained personnel to concentrate on more effective follow-up of women who test positive.15 Women generally had good acceptance of self-sampling, although concerns around obtaining a proper sample collection by women was a common concern and seems to be the reason why women preferred self-sampling at a clinic than at home.16 Even clinicbased vaginal self-sampling is less demanding on professional staff and eliminates the need for insertion of a speculum; many studies have shown that vaginal self-sampling is more acceptable to women than speculum-based screening.17–21 We carried out the PASS (Papillomavirus assessment in self-collected specimens) study to evaluate the performance, feasibility and logistics of hrHPV DNA testing on selfcollected vaginal samples while referring all positive women to colposcopy, at the population level under real “Seguro Popular” screening program conditions in the State of Morelos, Mexico. The PASS study employed hrHPV DNA screening of vaginal specimens among women attending clinics participating in the “Seguro Popular” CCSP program. The study’s aim was to collect evidence on the performance of hrHPV testing in a large regional pilot program to identify and refine the most effective strategy for employing hrHPV testing nationwide. Our long-term goal is to create a screening program that will minimize inequities in the quality of screening tests offered throughout Mexico.
Material and Methods Patients and eligibility
The study began in February 2007 and patient follow-up ended in July 2010. The target population for screening using self-collected samples was 150,535 women between 25 and 75 years of age who had “Seguro Popular” insurance coverage in the State of Morelos.22 “Seguro Popular” was created in 2003 to address healthcare needs of Mexicans without formal employment. The program seeks to provide health service coverage through voluntary public insurance for low-income persons who are not covered by other public insurance
schemes, regardless of employment status.23,24 A total of 121,650 women from the target population were invited to participate using the recruitment strategies described below. A total of 9,869 women were excluded because they had a prior cervical cancer screening in the previous year, an additional 893 women were excluded because they had a hysterectomy or were currently pregnant. Of the 10,646 (9.6%) women who refused to participate in the study, some declined because they simply were not interested whereas others preferred a cervical cancer screening service where a clinician obtained a cervical specimen for hrHPV testing. Thus, the final sample size for the study was 100,242 women who underwent hrHPV testing by vaginal self-sampling at local screening clinics. There were 10,863 women who tested positive for hrHPV and they were contacted by up to three visits to their homes, 8,267 (76.1%) subsequently attended colposcopy services. In addition, a randomly selected control group of 2,864 women from the hrHPV negative group (89,379) were also referred to colposcopy, 2,399 of whom were successfully evaluated, offering the necessary information for partial verification-bias adjustment.25 Settings and data collection sites
Over an eighteen-month period, the PASS study employed three simultaneous recruitment strategies. The first took place at homes and used a list of the potential population affiliated with “Seguro Popular.” Twenty community brigades (groups of at least two nurses, who also provided information on the benefits and procedure of specimen self-collection) offered vaginal self-sampling at home in 754 communities. In the second recruitment strategy, female medical service users of all 288 “Seguro Popular” primary health care centers were invited to participate in vaginal self-sampling at the health center with the same sampling method used for the first group at home. A third group of women were invited while they were attending health meetings during local government-organized health weeks. These national health weeks implement multiple disease prevention activities, including vaccination of children and adults. During these weeks, radio and television advertisements described the project, explained the usefulness of HPV testing and promoted community participation. Self collection of vaginal specimens
Collection of specimens followed the standard procedures for self-collection of vaginal samples that have been described in C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
previous studies.13 The standard Digene conical collection brush (Qiagen Inc., Gaithersburg, MD) was used to obtain the vaginal specimens. Women were instructed to grasp the brush midway along the shaft, insert the brush head as far as possible into the vagina and rotate it to the left and right. They were then instructed to place the brush immediately into the Digene cervical sampler tube that contains specimen transport medium (STM) and seal the tube with the provided cap. Specimens were collected daily and sent to the Cancer Prevention and Control Program’s coordination office in Cuernavaca, Morelos, transported weekly to the HPV laboratory at the Instituto Nacional de Salud P ublica (INSP) in Cuernavaca, Morelos, Mexico. hrHPV DNA detection by hybrid capture 2 (HC2)
The hrHPV DNA was detected with a standard commercial kit employing in vitro nucleic acid hybridization and chemiluminescent microplate-based signal amplification. This HC2 test has been described in detail by Lorincz,26 and was conducted in a validated and accredited routine HPV testing laboratory at the INSP, which is formally recognized as a regional center of expertise in CCSP HPV testing in central Mexico. HC2 RLU/CO cutoff of 1 was considered a positive test. The hrHPV positive women and a randomly selected subset of 3.2% of hrHPV negative participants serving as controls were referred to colposcopic examination for diagnosis confirmation. Colposcopy and histological confirmation
Six colposcopy clinics from the Morelos CCSP performed all colposcopy procedures. They were blinded to hrHPV status, although they knew that most of the women evaluated were hrHPV positive. Lesions were graded using the Reid Index,27 and colposcopy-directed biopsies were collected from areas of suspected cervical abnormalities. Biopsies were reviewed and diagnosed independently by two pathologists who were blinded to the colposcopy data and hrHPV testing results. If there was disagreement, a third pathologist did a reading and the final diagnosis was rendered by majority agreement. CIN21 cases were treated according to Mexico’s Cervical Cancer Screening Program guidelines.28
jects verified by histology, after which the values of the adjusted variables are calculated in the standard way. However, the method of Begg and Greenes cannot correct for an inability of the reference standard (colposcopy) to detect all disease; hence, we regard it as a partial verification bias adjustment. An approach to full verification bias correction is realistic only with long follow-up, multiple applications of the best reference methods, and the ethical clarity and permission to apply them. Crude and adjusted positive and negative predictive values (PPV and NPV) were calculated for hrHPV as a function of the probability of presence for each of CIN21 and CIN31 lesions among positive or negative hrHPV groups respectively. It is obvious that these estimates are also biased by the observed prevalence of disease and cannot be fully corrected because they are directly dependent on both the true prevalence of disease and the ability of downstream clinical process to detect and classify disease correctly. We also calculated adjusted positive and negative Likelihood Ratios (LR): 1LR 5 ASe/ (1-ASp); 2LR 5 (1-ASe/ ASp). A value of LR 5 1 indicates that the test result is equally likely in patients with and without the disease, values of 1LR > 1 and 2LR < 1 indicate the magnitude to which test results are likely to occur in patients with disease or without disease respectively, whereas LR1/LR2 gives the fold improvement for detection of the endpoint (CIN21 or CIN31) in women who are hrHPV1 versus hrHPV2. Unadjusted relative risk was calculated directly by dividing the proportion of disease in hrHPV positive women by the proportion of disease among the hrHPV negative control women. Power calculations were not conducted because the study was designed as a pilot implementation to be much larger than other adequately powered studies with similar endpoints.31 Calculations were conducted using Stata version 11.0 (Stata Corp. College Station). Ethics
The study was approved by the Research, Biosafety, and Ethics Committees at the National Institute of Public Health of Mexico, and was assigned registry number 657. All patients in the study gave informed consent.
Statistical analysis
Results
Our primary goal was to determine whether hrHPV testing, using the standard cutoff of 1 pg/ml for HC2 in routine implementation, would perform similarly to previously successful, adequately powered, smaller studies.11 We calculated the partial verification-bias of the hrHPV screening test, due to nonreferral of hrHPV negative women, using the method of Begg and Greenes; this method corrects for differential incomplete assessment of CIN21 and allowed us to derive an estimate of the adjusted sensitivity (ASe) and adjusted specificity) (ASp) for CIN21 or CIN31.29,30 The method initially adjusts the observed test counts in each contingency cell by dividing the subject count by the proportion of sub-
The study included 100,242 “Seguro Popular” affiliated women in the State of Morelos (Table 1). The overall prevalence of hrHPV DNA was 10.8% (95% CI 10.6–11.0). Among the 8,267 hrHPV positive women who attended the colposcopy service, 162 were histologically diagnosed with CIN21 (1.96%), which included 31 cases of CIN2, 76 of CIN3, and 55 invasive cancers. Among the control group of 2,399 hrHPV negative women, only three cases of CIN21 (0.125%; 2 cases of CIN2 and 1 case of CIN3) were identified (Fig. 1). Owing to the design of our study we were unable to calculate an unbiased estimate of clinical sensitivity; however, we provide estimates of ASe and ASp in Table 2. Comparing
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
Early Detection and Diagnosis
111
Lazcano-Ponce et al.
112
HPV DNA screening in self-collected specimens
Table 1. Prevalence of hrHPV and CIN21 among 100,242 Mexican women aged 25–75 years, stratified by participants’ characteristics 1
Total N 5 100,242
hrHPV prevalence2 positive (n 5 10,863)
CIN21 prevalence2 N 5 165
n
%
%
95% CI
%
95% CI
25–29
1,106
1.1
18.7
16.4–21.0
0.36
0.01–0.72
30–34
13,905
13.9
12.5
11.9–13.0
0.14
0.08–0.21
35–39
21,406
21.4
10.3
9.9–10.7
0.14
0.09–0.18
40–44
16,474
16.4
9.7
9.2–10.1
0.10
0.05–0.15
45–49
13,794
13.8
9.4
8.9–9.9
0.19
0.12–0.26
50–54
11,016
11.0
9.8
9.2–10.3
0.22
0.13–0.30
55–59
8,069
8.1
10.8
10.1–11.4
0.35
0.22–0.48
60–64
5,519
5.5
12.3
11.4–13.2
0.09
0.01–0.17
65–69
3,897
3.9
14.0
12.9–15.0
0.21
0.06–0.35
70–75
5,056
5.0
13.1
12.1–14.0
0.08
0.00–0.16
Homemaker/informal employee
65,404
65.3
9.5
9.2–9.7
0.14
0.11–0.17
Formal employee
34,240
34.2
12.4
12.0–12.7
0.19
0.14–0.24
Never
10,707
10.7
12.5
11.8–13.1
0.21
0.13–0.30
Ever
89,513
89.3
10.6
10.4–10.8
0.16
0.13–0.18
1 sexual partner
82,736
82.5
7.1
7.0–7.3
0.08
0.06–0.10
2 or more sexual partners
17,486
17.4
28.3
27.7–29.0
0.55
0.44–0.66
Urban
65,006
64.9
10.9
10.7–11.2
0.16
0.13–0.19
Suburban
19,504
19.5
9.8
9.6–10.6
0.20
0.14–0.26
Rural
15,134
15.1
9.6
9.5–10.4
0.11
0.05–0.16
Cervical clinician collected
13,714
13.7
10.1
9.6–10.6
0.13
0.07–0.19
Self sampling
85,930
85.7
10.5
10.3–10.7
0.16
0.13–0.19
Variable Age groups (years)
Occupation
Pap test history
Sexual partner history
Early Detection and Diagnosis
Place of Residence
Screening test preference for use in the future
Cells that do not add up to 100% are due to missing values. The prevalence of CIN21 shown in each row category was computed from the number of cases of CIN21 in each row category divided by the total of participants in each row category. 1 Percentages on the total are based on the column. 2 Percentages for hrHPV and CIN21 are based on the rows. Abbreviations: CIN21: cervical intraepithelial neoplasia grade 2 or worse; hrHPV: high-risk human papillomavirus.
hrHPV positive versus hrHPV negative women, the relative risk of CIN21 was 15.67 (0.01959/0.00125). Figure 2 shows the age-specific hrHPV DNA prevalence for the PASS study when compared with the age stratified distribution of hrHPV in vaginal self-sample specimens obtained from one of our previous studies in Morelos.32 In both groups we observed two peaks of hrHPV prevalence, the first among women aged 30–34 years, and a second among women over the age of 55 (Fig. 2). Participants with formal employment (12.4% vs. 9.5%), as well as women having a history of 2 or more lifetime sexual partners (28.3% vs. 7.1%) had a significantly higher rate of hrHPV detection. The hrHPV prevalence was also significantly higher in the group of urban participants (10.9%) in
comparison with rural women at 9.6% (Table 1). Verification bias corrected estimates as well as the crude estimates for sensitivity and specificity, PPV and NPV are shown in Table 2. The partial verification-bias-corrected estimate of adjusted sensitivity for CIN21 was 64.4% (95%CI 59.5–69.1). The corrected estimate for adjusted specificity in women 25 years was 89.4% (95%CI 89.2–89.6) for CIN21 (Table 2). ). The estimates were similar across different age groups [25 and 35 years (data not shown)]. High adjusted negative predictive values were observed for both CIN21 (99.8%) and CIN31 (99.9%). The positive likelihood ratios of hrHPV testing for CIN21 and CIN31 detection were 6.1 and 7.7, respectively (Table 2). Negative likelihood ratios for CIN21 and CIN31 detection were 0.40 and 0.21, respectively. C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
113
Early Detection and Diagnosis
Lazcano-Ponce et al.
Figure 1. Flowchart of the screening profiles of women in the PASS study. The target population was all women residing in the state of Morelos who were affiliated with the Seguro Popular insurance program, directed to address the healthcare needs of Mexicans without formal employment. The study protocol specified that all formally enrolled high-risk human papillomavirus (hrHPV) DNA–positive women would be referred to colposcopy. In addition, 3.2% of the hrHPV negative women were also sent to colposcopy to act as controls. In all, 1.96% (n 5 162) of the 8,267 hrHPV positive women who completed colposcopy had a histopathological diagnosis of CIN21; in comparison 0.125% (3) of women who completed colposcopy with negative hrHPV were found to have CIN21.
The proportion of biopsies collected from hrHPV positive women referred to different colposcopy clinics varied from 41.3% (111/269) to 9.1% (228/2496) (Table 3).
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
In addition, there was a clear positive correlation between the proportion of biopsies collected in the hrHPV positive women by each colposcopy clinic and the proportion of
114
HPV DNA screening in self-collected specimens
Table 2. Crude and verification bias adjusted estimates of test performance characteristics for hrHPV testing on self-collected vaginal specimens as primary screening of CIN21 and CIN31 among Mexican women aged 25–75 years Crude estimates CIN
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
LR1
95% CI
LR2
95% CI
21
98.2%
94.8–99.6
89.3%
89.1–89.5
1.5%
1.3–1.7
100%
100–100
9.2
8.9–9.4
0.02
0.006–0.06
31
99.2%
95.9–100
89.3%
89.1–89.5
1.2%
1.0–1.4
100%
100–100
9.2
9.0–9.5
0.008
0.001–0.06
Verification bias adjusted estimates CIN
Sensitivity
95% CI
Specificity
95% CI
PPV
95% CI
NPV
95% CI
LR1
95% CI
LR2
95% CI
21
64.4%
59.5–69.1
89.4%
89.2–89.6
2.4%
2.1–2.7
99.8%
99.8–99.9
6.1
5.6–6.5
0.40
0.35–0.45
31
81.5%
76.2–86.1
89.3%
89.1–89.5
1.9%
1.7–2.2
99.9
99.9–100
7.7
7.2–8.1
0.21
0.16–0.27
Early Detection and Diagnosis
The adjusted estimates are based on histological results of a large set of women randomly selected from among the hrHPV negative participants. This allowed adjustment for verification bias caused by preferential colposcopy referral of hrHPV positive cases. Corrected estimates were computed according to Begg CB, Greenes RA30
Figure 2. Comparison of hrHPV prevalence in women stratified by age in the years 2000 and 2007, in two studies from the state of Morelos. The patterns of hrHPV prevalence by age were quite similar in these two studies from the same geographic area conducted 7 years apart. The hrHPV prevalence was high, reaching almost 13% in women aged 30–34 years. The prevalence decreased significantly in women past the childbearing age until a nadir of 9% in women aged 40–49 years. The hrHPV prevalence increased again in women aged 65–69 to 15%.
CIN21 cases detected. For example, Table 3 compares 41.3% of hrHPV positive women biopsied and 4.1% of CIN21 cases detected, versus 9.1% women biopsied and 1.2% CIN21 detected.
Discussion Our study describes the first large-scale introduction of hrHPV testing into a middle-income country’s routine CCSP. The pattern of hrHPV prevalence by age was similar to that found in a previous study in the same geographic area conducted almost 7 years earlier; hrHPV prevalence was high in younger women, decreased significantly in women past the childbearing age until 49 years, and increased again in older women12 (Fig. 2). The hrHPV testing of self-collected vaginal samples demonstrates a high NPV for relevant cervical disease, highlighting this test’s benefit as a primary screening procedure in
developing countries, as women who had a negative hrHPV result have virtually no risk of CIN31 lesions. Self-collected hrHPV testing may be particularly attractive for increasing screening intervals, and it appears to be an acceptable screening option in large-scale routine settings. In addition, given the magnitude of the relative risk and positive likelihood ratio, our findings demonstrate that hrHPV testing of self-collected samples is a satisfactory screening test for identification of women with target CIN21 lesions. Increasing screening coverage with hrHPV testing of selfcollected vaginal samples is appealing because this system is easier to implement in settings lacking infrastructure and has also been shown to be more sensitive than routine cervical cytology.13 However, if routine testing of self-collected vaginal specimens is to be used for primary screening, we need to account for potential problems with that system. For example, the logistical difficulties posed by Mexico’s current system are exemplified by the fact that 23.9% of hrHPV positive participants did not receive colposcopy. The relatively low follow-up rate is due to a number of factors, primarily difficulties regarding transportation, clinic hours, costs, and child care needs that impede women’s access to health services. Other factors include lack of clinician recommendations and information, a culture of reluctance, and high levels of migration or change of residence in deprived Mexican populations. Thus, an effective screening program must include a system for comprehensive follow-up of hrHPV positive women regardless of whether the specimens are self or clinician-collected. The variation we found in follow-up colposcopy also suggests that directly referring all hrHPV positive women to colposcopy can cause additional problems in a CCSP by over-burdening the professional services. We observed a wide variation in the detection rate of CIN21 (4.1 vs. 1.2) among the colposcopy clinic participants. The major reason for this variation was likely the different number of biopsies taken among the colposcopy groups (41.3– 9.1%). Our data suggest that lack of standardization in biopsy collection could be an important cause of failure in the Mexican CCSP, and improvements are needed if we are to realize C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
115
Lazcano-Ponce et al.
Table 3. Colposcopy performance in the PASS study to detect CIN21 among Mexican women aged 25–75 years Evaluated women
Proportion of biopsies collected
CIN21 detected
Colposcopy clinics
HPV pos (n 5 8,267)
HPV neg (n 5 2,399)
HPV pos (n 5 1081)
HPV neg (n 5 245)
HPV pos (n 5 162) [95%CI]
HPV neg (n 5 3)
1
269
115
41.3% (111)
30.4%
4.1% (11)
0% (0)
(35)
[1.7–6.5]
2
1,091
628
18.1% (197)
9.9%
3.3% (36)
(62)
[2.2–4.4]
3
553
247
16.5% (91)
10.9%
2.7% (15)
(27)
[1.4–4.1]
4
2,839
261
11.9% (339)
8.8% (23)
1.9% (53)
5
1019
523
11.3% (115)
8.6% (45)
6
2496
625
9.1% (228)
8.5% (53)
0.2% (1)
0% (0)
0% (0)
[1.4–2.4] 1.7% (17)
0.2% (1)
[0.9–2.5] 1.2% (30)
0.2% (1)
the full benefits of hrHPV DNA testing. Colposcopy can be improved if technical accreditation of practitioners is required and by introducing the practice of multiple biopsies. There has been debate over how many biopsies are needed, with up to four in total being recommended for screenpositive women.33 It is reasonable to suggest that at least one biopsy should be systematically collected from the most suspicious area of squamocolumnar junction in all hrHPV positive women triaged to colposcopy; this may substantially increase the cervical disease detection rate and would also increase the apparent PPV of an hrHPV positive test.33,34 The costs arising from the collection of multiple biopsies can be controlled if these biopsies are included and processed in the same paraffin block. In contrast with our previous studies, the estimated adjusted PPV for detection of CIN21 was very low. We observed only a 2.4% probability of detecting CIN21 lesions among women positive for hrHPV, meaning that hrHPV testing identified a large number of women who, by standard interpretation, did not require further evaluation to exclude high-grade CIN abnormalities; however, this low PPV needs to be considered in the context of the accuracy of colposcopy. It is likely that less evident lesions were often missed and biopsies were mainly collected in the more visible advanced cervical disease cases; this is also reflected in the relatively low number of CIN2 cases in our study when compared with CIN3 or cancer cases. As previously reported, overloading colposcopy services leads to lower biopsy collection rates.35 Our results suggest that colposcopy should be deployed only after other conventional triage alternatives like cervical cytology, genotyping for HPV16 (possibly also HPV18) or the use of biomarkers. Another alternative is to use a higher threshold of the HC2 test. For example, in our earlier MARCH study a HC2 RLU/CO cutoff of two gave a PPV of 20% and a cutoff of five gave a PPV of 26%, both these cut-
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
offs were associated with acceptable detection rates of CIN21 when compared with routine cytology.13 It is possible that a higher threshold of five for HC2 is a reasonable alternative to genotyping for HPV16 and 18, giving similar sensitivity and PPV.36 For both genotyping or using an HC2 RLU/CO of 5, care needs to be taken to ensure good followup and possibly frequent repeat testing for the women who are hrHPV positive for types other than 16 or 18, or are HC2 positive in the RLU/CO range of less than 5. The main dilemma in middle-income regions is how to improve both the quality and accessibility of triage after a positive HPV DNA screening test. In the current Mexican CCSP implemented within the “Seguro Popular”, guidelines for HPV testing-based screening recommend using the Pap test only in hrHPV2positive women. The Pap test must be of high quality and in particular also highly sensitive (90% or better) for triage to be effective, which is a major challenge in most developing countries. The hrHPV testing as primary screening may decrease the numbers of necessary cytology evaluations by close to 90%, thus allowing cytologists to focus more carefully on the remaining specimens enriched for abnormalities, as well as permitting the use of cytology double-reading procedures in order to improve the sensitivity and overall quality of cytology evaluation. However, this triage alternative is not well established and may require further evaluations in developing countries. Ideally, triage and follow-up would be conducted in the context of an organized registry-based program that could apply appropriate patient call and recall procedures. Testing cervical lesions for carcinogenic HPV types including 16, 18 and 45 may help identify cases that are more likely to progress to cancer.36 Currently, several triage alternatives for hrHPV positive women are being evaluated, focusing on HPV genotyping, E6 protein detection, and incorporation of cytological biomarkers such as p16 and Ki-67 or DNA
Early Detection and Diagnosis
[0.8–1.6]
116
HPV DNA screening in self-collected specimens
methylation.37 Meanwhile, we are striving to both improve our CCSP infrastructure and to modify existing practices by scaling up the new, currently validated strategies that appear appropriate to our socio-cultural context and technological capabilities.
Acknowledgements Drs. Enrique Carmona Balandrano, Patricia Alonso and the Association of Colposcopy of Morelos State, Mexico provided pathology and colposcopy expertise. The authors gratefully recognize the patients who participated in this study.
References 1.
2.
3.
4.
5.
Early Detection and Diagnosis
6.
7.
8.
9.
10.
11.
12.
13.
Szarewski A. Ushering in a new era: human papillomavirus (HPV) testing comes to the NHS cervical screening programme. J Fam Plann Reprod Health Care 2011;37:64–7. Ronco G, Giorgi-Rossi P, Carozzi F, et al. Efficacy of human papillomavirus testing for the detection of invasive cervical cancers and cervical intraepithelial neoplasia: a randomised controlled trial. Lancet Oncol 2010;11: 249–57. Bulkmans NW, Berkhof J, Rozendaal L, et al. Human papillomavirus DNA testing for the detection of cervical intraepithelial neoplasia grade 3 and cancer: 5-year follow-up of a randomised controlled implementation trial. Lancet 2007;370:1764–72. Mayrand MH, Duarte-Franco E, Rodrigues I, et al; Canadian Cervical Cancer Screening Trial Study Group. Human papillomavirus DNA versus Papanicolaou screening tests for cervical cancer. N Engl J Med 2007;357:1579–88. Naucler P, Ryd W, T€ ornberg S, et al. Human papillomavirus and Papanicolaou tests to screen for cervical cancer. N Engl J Med. 2007;357: 1589–97. Sankaranarayanan R, Nene BM, Shastri SS, et al. HPV screening for cervical cancer in rural India. N Engl J Med 2009;360:1385–94. Kitchener HC, Almonte M, Thomson C, et al. HPV testing in combination with liquid-based cytology in primary cervical screening (ARTISTIC): a randomised controlled trial. Lancet Oncol 2009;10:672–82. Anttila A, Kotaniemi-Talonen L, Leinonen M, et al. Rate of cervical cancer, severe intraepithelial neoplasia, and adenocarcinoma in situ in primary HPV DNA screening with cytology triage: randomised study within organised screening programme. BMJ 2010;340:c1804. Hernandez-Avila M, Lazcano-Ponce E, Martınez O. Public policies for HPV screening and vaccination in Mexico. In the 27th International Papillomavirus Conference and Clinical Workshop. Berlin, Germany, September 17–23, 2011. Secretarıa de Salud. Programa de acci on especıfico 2007–2012 Cancer Cervicouterino, 1st edn. Mexico, D.F.: Secretarıa de Salud, 2008. Lazcano-Ponce E, L€ orincz AT, Salmer on J, et al. A pilot study of HPV DNA and cytology testing in 50,159 women in the routine Mexican social security program. Cancer Causes Control 2010;21: 1693–700. Salmeron J, Lazcano-Ponce E, Lorincz A, et al. Comparison of HPV2based assays with Papanicolaou smears for cervical cancer screening in Morelos State, Mexico. Cancer Causes Control 2003;14:505–12. Lazcano-Ponce E, Lorincz AT, Cruz-Valdez A, et al. Self-collection of vaginal specimens for human papillomavirus testing in cervical cancer prevention (MARCH): a community-based
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
randomised controlled trial. Lancet 2011;378: 1868–73. Sowjanya AP, Paul P, Vedantham H, et al. Community Access to Cervical Health Study Group. Suitability of self-collected vaginal samples for cervical cancer screening in periurban villages in Andhra Pradesh, India. Cancer Epidemiol Biomarkers Prev 2009;18:1373–8. Gravitt PE, Belinson JL, Salmeron J, et al. Looking ahead: a case for human papillomavirus testing of self-sampled vaginal specimens as a cervical cancer screening strategy. Int J Cancer 2011;129:517–27. Petignat P, Vassilakos P. Is It Time to Introduce HPV Self-Sampling for primary cervical cancer screening? JNCI 2012;194:166–7. Waller J, McCaffery K, Forrest S, et al. Acceptability of unsupervised HPV self-sampling using written instructions. J Med Screen 2006;13:208–13. Mitchell S, Ogilvie G, Steinberg M, et al. Assessing women’s willingness to collect their own cervical samples for HPV testing as part of the ASPIRE cervical cancer screening project in Uganda. Int J Gynaecol Obstet 2011;114: 111–5. Sellors JW, Lorincz AT, Mahony JB, et al. Comparison of self-collected vaginal, vulvar and urine samples with physician-collected cervical samples for human papillomavirus testing to detect highgrade squamous intraepithelial lesions. CMAJ 2000;163:513–8. Stewart DE, Gagliardi A, Johnston M, et al. Selfcollected samples for testing of oncogenic human papillomavirus: a systematic review. J Obstet Gynaecol Can. 2007;29:817–28. Dzuba IG, Dıaz EY, Allen B, et al. The acceptability of self-collected samples for HPV testing vs. the pap test as alternatives in cervical cancer screening. Womens Health Gend Based Med 2002;11:265–75. National Institute of Statistics and Geography. Data bank. Interactive data consulting. Mexico Population and Housing Census 2010. Female population aged over 12 by age and Seguro Popular affiliation. National Institute of Statistics and Geography, 2010. Available at: http:// www.inegi.org.mx/lib/olap/consulta/general_ver4/ MDXQueryDatos.asp?#Regreso&c527874; in Spanish. Last accessed on May 31 2012, Mexico City. G omez Dantes O, Sesma S, Becerril VM, et al. The health system of Mexico. Salud Publica Mex. 2011;53:s220–s32. Knaul FM, Gonzalez-Pier E, G omez-Dantes O, et al. The quest for universal health coverage: achieving social protection for all in Mexico. Lancet 2012;380:1259–79. Kupets R, Paszat L. Physician and patient factors associated with follow up of high grade dysplasias of the cervix: a population-based study. Gynecol Oncol 2011;120:63–7.
26. Lorincz, AT. Hybrid CaptureTM method for detection of human papillomavirus DNA in clinical specimens: a tool for clinical management of equivocal Pap smears and for population screening. J Obstet Gynaecol Res 1996;22:629–636. 27. Reid R. Colposcopy of cervical preinvasive neoplasia. In: Singer A, ed. Premalignant lesions of the lower genital tract, vol. 2. New York: Elsevier, 1990. 87–116. 28. United Mexican States-Ministry of Health. Amendment to Mexican Official Standard NOM-014-SSA2–1994, for the prevention, treatment and management of cervical cancer and breast cancer in primary care, to read as NOM014-SSA2-1994, for the prevention, detection diagnosis, treatment, control and surveillance of cervical cancer. United Mexican States-Ministry of Health, 1995. Available at: http://www.salud. gob.mx/unidades/cdi/nom/m014ssa24.html; in Spanish. Last accessed on Oct 7, 2011, Mexico City. 29. Cronin AM, Vickers AJ. Statistical methods to correct for verification bias in diagnostic studies are inadequate when there are few false negatives: a simulation study. BMC Medical Research Methodol 2008;8:75. 30. Begg CB, Greenes RA. Assessment of diagnostic tests when disease verification is subject to selection bias. Biometrics 1983;39:207–215. 31. Belinson JL, Du H, Yang B, et al. Improved sensitivity of vaginal self-collection and high-risk human papillomavirus testing. Int J Cancer 2012; 130:1855–60. 32. Flores Y, Bishai D, Lazcano E, et al. Improving cervical cancer screening in Mexico: results from the Morelos HPV Study. Salud Publica Mex 2003; 45:S388–S98. 33. Stoler MH, Vichnin MD, Ferenczy A, et al. The accuracy of colposcopic biopsy: Analyses from the placebo arm of the Gardasil clinical trials. Int J Cancer 2011;128:1354–62. 34. Pretorius RG, Peterson P, Azizi F, et al. Subsequent risk and presentation of cervical intraepithelial neoplasia (CIN) 3 or cancer after a colposcopic diagnosis of CIN 1 or less. Am J Obstet Gynecol 2006;195:1260–5. 35. Arbyn M, Simoens C, Van Oyen H, et al. Analysis of 13 million individual patient records pertaining to Pap smears, colposcopies, biopsies and surgery on the uterine cervix (Belgium, 1996– 2000). Prev Med 2009;48:438–43. 36. Khan MJ, Castle PE, Lorincz AT, et al. The elevated 10-year risk of cervical precancer and cancer in women with human papillomavirus (HPV) type 16 or 18 and the possible utility of typespecific HPV testing in clinical practice. J Natl Cancer Inst 2005;97:1072–9. 37. Galgano MT, Castle PE, Atkins KA, et al. Using biomarkers as objective standards in the diagnosis of cervical biopsies. Am J SurgPathol 2010;34: 1077–87.
C 2013 UICC Int. J. Cancer: 135, 109–116 (2014) V
