HIV-infected women are at increased risk of infection with human papillomavirus (HPV) and its sequelae, cervical intraepithelial neoplasia (CIN), and invasive cervical cancer (ICC).1 Primary cervical screening using high-risk (hr) HPV testing in HIV-infected women has been shown to have satisfactory sensitivity and positive predictive value (PPV) but lower specificity2–42–42–4 than in HIV-uninfected women5 because of much higher prevalence of hrHPV infections.6,76,7 Effective triage methods are therefore especially necessary in HIV-infected women.
At present, cytology is considered the gold standard for the triage of hrHPV-positive women in high-income countries,8,98,9 whereas visual inspection with acetic acid (VIA)10 is recommended in low-income countries. However, both methods suffer from problems of accuracy and reproducibility.11 Very few studies are available on the combination of HPV testing with either cytology2 or VIA2,42,4 in HIV-infected women.
DNA hypermethylation in the promoter region of certain tumor-relevant genes can lead to gene silencing and has been identified in many cancers, including cervical cancer,12,1312,13 and in precancerous cervical lesions.14 Steenbergen et al14 make a distinction between 2 types of HPV-induced lesions in the cervical epithelium based on the nature of the HPV infection: productive CIN (comprising CIN1 and a subset of CIN2) and transforming CIN (comprising CIN3 and the remainder of CIN2). Only transforming CIN has the potential for malignant transformation but may be morphologically indistinguishable from productive CIN. Because methylation changes of host cell genes accumulate over time in a transforming HPV infection, they can help identify advanced transforming CIN and be, therefore, of special value in cervical cancer screening.14 More than 80 genes have been identified as possible targets for methylation diagnostic biomarkers of cervical cancer, although most have been analyzed in only a single study.13 Among the more promising genes13 are CADM1 (gene encoding cell adhesion molecule 1),15MAL (encoding T-lymphocyte maturation-associated protein),16 and MIR124-2 (encoding the microRNA hsa-miR-124-2).17 Methylation levels of these genes have been shown to be able to discriminate ICC and its most severe precursors from normal cervix among hrHPV-positive women and have been clinically validated18–2118–2118–2118–21 using samples from the general female population.
Methylation markers of cervical cancer have not been well studied in HIV-infected women. The association between cervical cancer and CIN and methylation levels of PEG3, PEG1/MEST, and IGF2 have been analyzed in a case–control study of cervical cancer and precancerous lesions in Tanzania, where the HIV seroprevalence in controls was 20%.22,2322,23 However, no previous study has analyzed the clinical performance of methylation markers in screening of HIV-positive women for cervical cancer. The aim of the present study was to assess for the first time the clinical performance of methylation analysis of CADM1, MAL, and MIR124-2 (using markers CADM1-m18, MAL-m1, and MIR124-2), individually or combined in a tri-marker panel, in the triage of HIV-infected hrHPV-positive women.
Participants and Study Procedures
Two hundred fifty-one HIV-infected women who were hrHPV positive as previously determined2,62,6 and who had a valid histology result were eligible for inclusion in the present report, ie, 50% of 498 HIV-infected women who were originally included at the Coptic Hope Center for Infectious Diseases, Nairobi, Kenya, in a study comparing various cervical cancer screening strategies (see study flowchart in Supplemental Digital Content 1, http://links.lww.com/QAI/A709).2,6,242,6,242,6,24 The HIV-infected hrHPV-positive women were aged between 18 and 55 years, had not undergone cervical screening in the last year, and had not been treated for cervical cancer or precancerous lesions.2,6,242,6,242,6,24
Briefly, after obtaining written informed consent, a questionnaire on women's characteristics and use of combination antiretroviral therapy (cART) was administered, VIA was performed, and cervical cells for conventional cytology were collected using a Cervex-Brush (Rovers Medical Devices, Oss, the Netherlands). After preparation of a conventional smear, the brush was placed in a vial with PreservCyt medium (Hologic, Marlborough, MA) for DNA isolation. Isolated DNA was subsequently used for HPV testing, and the hrHPV-positive ones for methylation marker analysis. A physician performed a colposcopic examination and took a biopsy from all women, either from the most abnormal area on the cervix or at 12 o'clock if no lesion was visualized. Conventional cytology slides and biopsies were read by the study pathologist at the Aga Khan University, Nairobi. Cytology was reported according to the Bethesda 1991 revised classification.25 Thirteen hrHPV-positive women with inadequate biopsies were excluded (see flowchart in Supplemental Digital Content 1, http://links.lww.com/QAI/A709). Women with CIN2 lesions or worse (CIN2+) were offered cryotherapy or more extended treatment if not eligible for cryotherapy.26
A follow-up visit was planned for all women 6 months after the baseline visit. Cervical cells were collected again to perform conventional cytology, HPV testing, and, among women who were still hrHPV positive at follow-up, a second methylation marker analysis. Women with high-grade intraepithelial lesions (HSIL) had another colposcopic examination and, if lesions were visible, another biopsy. Women who were CIN2+ at the 6-month follow-up visit were treated with cryotherapy if appropriate.26 If the lesion was too large or the woman had already been treated at the screening round, a loop electrosurgical excision procedure was performed.
The study protocol was approved by the Ethical Review Committees of the Kenyatta National Hospital, Kenya, the University of Washington, Seattle, WA, and the International Agency for Research on Cancer, France.
HPV DNA Testing
HPV DNA testing had already been done on cells at baseline and at 6-month visit as described previously.2,62,6 Briefly, the presence of HPV DNA was first determined using general primer GP5+/6+-mediated polymerase chain reaction (PCR).27 PCR products were hybridized using an enzyme immunoassay that included an oligoprobe for hrHPV types 16, 18, 31, 33, 35, 39, 45, 51, 52, 56, 58, 59, 66, and 68. HPV genotyping was performed by reverse line blot hybridization of PCR products.28 Women positive for the hr oligoprobes and for hrHPV genotyping were classified as hrHPV positive.
Methylation Marker Analysis
Methylation marker analysis of the promoter regions of the CADM1, MAL, and MIR124-2 genes was carried out in hrHPV-positive women. Quantitative methylation-specific PCR (qMSP) comprising the regions CADM1-m18, MAL-m1, and MIR124-2 was run in single separate reactions as previously described.20 As a reference, a qMSP for methylation-independent target β-actin was performed. First, extracted DNA was subjected to sodium bisulfite treatment using the EZ DNA Methylation Kit (Zymo Research, Irvine, CA). Bisulfite converts unmethylated cytosines into uracils, whereas methylated cytosines remain unchanged. This difference in sequence can be detected by qMSP. All qMSP assays were run on a 7500 Fast ABI real-time PCR system (Applied Biosystems, Carlsbad, CA). Cycle threshold (Ct) values were measured at a fixed fluorescence threshold of 0.01. Ct ratios between the Ct values of the housekeeping gene β-actin and the target gene were used to quantify methylation level using the formula: 100 × 2(Ct(β-actin)−Ct(target)). Baseline samples of 3 women were considered invalid for methylation analysis given negative results for β-actin and all gene targets. These women were excluded from further analyses. A further 8 women had invalid methylation results at 6 months and were excluded from the follow-up analysis.
The distribution of Ct ratios was highly skewed with some zero values representing no detectable methylation for the target gene. To account for this nonstandard distribution, nonparametric methods were used. Comparisons of methylation by histological diagnosis were made using the median Ct ratio, with 95% confidence intervals (CIs) and P values calculated by bootstrapping.29 Changes in median Ct ratio between baseline and follow-up were tested with the nonparametric sign test.30 Correlations between Ct ratios of the 3 methylation markers were calculated using Spearman rank correlations.
Sensitivity and specificity of the methylation markers among hrHPV-positive women were calculated for the detection of CIN2+. The 3 methylation markers were combined into a tri-marker panel by considering a woman as positive if the Ct ratio of at least one of the markers was above its respective threshold for positivity, using a different threshold for each methylation marker. Thresholds were calculated in an optimization step by first fixing the minimum value for the specificity of the combined test and then maximizing its sensitivity over all possible combinations of thresholds. This optimization step was repeated for specificity values from 10% to 90% in steps of 10%. The optimization step used simulated annealing31 and was repeated 10 times from random starting points to ensure that a global maximum was obtained for the sensitivity of the tri-marker panel. PPVs and the complement of the negative predictive value (cNPV = 100−NPV), a measure of disease risk after a negative result, were also computed for each combination of sensitivity and specificity. Receiver operating characteristic (ROC) curves32 were calculated for the individual methylation markers, and an approximate ROC curve was plotted for the tri-marker panel using 10% steps in specificity. ROC curves were also plotted for CIN3 after excluding women with CIN2 and without changing the thresholds for the combined test.
A reference test for the tri-marker panel was calculated for the target specificity ≥50%. The thresholds for positivity in this reference test were Ct ratios of 0.043 for CADM1-m18, 0.104 for MAL-m1, and 0.741 for MIR124-2.
The median age of 248 HIV-infected hrHPV-positive women was 37 years (interquartile range, IQR: 33–42), and median CD4 count was 331 cells per microliter (IQR: 217–485). Seventy-two percent of women had been using cART for a median of 669 days (IQR: 230–1113). Sixty-eight percent of women had multiple hrHPV-type infections. The most frequently detected types in either single or multiple infections were HPV35, HPV16, and HPV52. Table 1 shows Ct ratios of CADM1-m18, MAL-m1, and MIR124-2 and corresponding 95% CIs by biopsy finding at baseline. Zero values for methylation were less often found for MIR124-2 (1.2%) than for CADM1-m18 (23.8%) or MAL-m1 (24.2%). Nevertheless, a tendency of Ct ratios to increase with the severity of biopsy findings was found for all methylation markers. Median Ct ratio was elevated in CIN2 and CIN3 lesions compared with normal biopsies for CADM1-m18 (2.5 and 14 times, respectively) and MAL-m1 (6.5 and 20 times, respectively) (P ≤ 0.006). For MIR124-2, a 7.7-fold significant difference in Ct ratio compared with normal biopsies was found for CIN3 lesions (P < 0.001), but not for CIN2 (P = 0.16). Ct ratios in CIN1 were intermediate between those of normal biopsies and CIN2 but always many times lower than in CIN3.
Of 176 women with valid diagnosis at follow-up, 110 had ≤CIN1 at baseline. Table 1 also shows median baseline Ct ratios by diagnosis at 6-month follow-up among these 110 women. Women who had CIN2+ diagnosed at follow-up tended to have higher baseline Ct ratios for all methylation markers than women who remained ≤CIN1, and the difference was statistically significant for MAL-m1 (P = 0.05) and MIR124-2 (P = 0.05).
Positive correlation coefficients of ≥0.30 were found for all 2-way combinations for each methylation marker at baseline (see Table, Supplemental Digital Content 2, http://links.lww.com/QAI/A709). The strongest correlation (0.73) was found for MIR124-2 vs MAL-m1 in CIN2+ lesions.
ROC curves for the detection of CIN2+ (Fig. 1A) or CIN3 (Fig. 1B) provide a visual comparison of the sensitivity and specificity of CADM1-m18, MAL-m1, and MIR124-2 and the tri-marker panel. Point estimates of the sensitivity and specificity of cytology [atypical squamous cells of undetermined significance or worse (ASCUS+)], HPV16/18 genotyping, and VIA are also shown. The curves for CADM1-m18, MAL-m1, and MIR124-2 overlap substantially for both CIN2+ and CIN3 endpoints. The tri-marker panel and cytology (ASCUS+) show comparable sensitivity at the same level of specificity. The sensitivity of the tri-marker panel is substantially better than the sensitivity of VIA at the same specificity and is nearly 2 times better than the sensitivity of HPV16/18 (Figs. 1A, B).
Tables 2 and 3Tables 2 and 3 show the sensitivity and some additional indicators of the clinical performance of the tri-marker panel by specificity threshold for CIN2+ (Table 2) and CIN3 (Table 3). Tables 2 and 3Tables 2 and 3 also show the corresponding performance of cytology (with ASCUS, low-grade squamous intraepithelial lesion, or HSIL as threshold), HPV16/18 genotyping, and VIA. At a specificity threshold of ≥50%, for instance, the sensitivity of the tri-marker panel (positive in 65% of women) was 89% (95% CI: 81 to 95) for CIN2+ and 95% (84% to 99%) for CIN3. The sensitivity of ASCUS+ cytology (positive in 69% of women) was 95% for both CIN2+ and CIN3. The sensitivity of the tri-marker panel was significantly better than the sensitivity of HPV16/18 genotyping (40% for both CIN2+ and CIN3) or VIA (70% and 74%, respectively). PPVs of the tri-marker panel, ASCUS+ cytology, HPV16/18 genotyping, and VIA were similar, in a 50%–54% range for CIN2+ (Table 2) and 32%–36% range for CIN3 (Table 3). Conversely, cNPVs differed substantially. For the detection of CIN3, for instance, cNPV was 3% for both tri-marker panel and ASCUS+, 18% for HPV16/18 and 10% for VIA (Table 3).
A sensitivity analysis was conducted on the reference tri-marker panel (specificity ≥ 50%) to determine whether data on cART or CD4 count could affect the predictive value for CIN2+ (data not shown). When women were stratified by cART, users and nonusers had similar PPV (56% and 51%, respectively) and cNPV (12% and 11%, respectively). Likewise, women with low CD4 count (<350 cells per microliter) showed similar PPV to women with a higher CD4 count (53% and 51%, respectively), and cNPV was also similar between these 2 groups (11% and 12%, respectively).
Table 4 shows median Ct ratios for the 3 methylation markers at baseline and at 6-month follow-up among 128 women who remained hrHPV positive. Among 69 women with ≤CIN1 at baseline, median Ct ratio increased at follow-up for MAL-m1 (P = 0.006) and MIR124-2 (P = 0.007) but not for CADM1-m18 (P = 1). Among 59 women who had CIN2+ at baseline and had undergone treatment, there was no statistically significant change.
Our present report on HIV-infected women who tested positive for hrHPV in cervical cancer screening shows that markers of methylation of host cell genes involved in cervical carcinogenesis are a promising triage method in this group of women. Methylation levels of studied CADM1, MAL, and MIR124-2 loci, as measured by Ct ratios in exfoliated cervical cells, were many times higher in women with CIN2+ lesions than in those with normal biopsies. Triage of HIV-infected hrHPV-positive women using a combination of the 3 markers was comparable with cytology (ASCUS+) and clearly superior to HPV16/18 genotyping or VIA. Consistent with previous work,14 methylation markers were especially able to distinguish CIN3 compared with CIN2, 2 lesions that are very difficult to distinguish morphologically. In a subset of study women who had a 6-month follow-up visit, newly diagnosed CIN2+ at follow-up was also associated with higher methylation levels at baseline.
Support for the ability of qMSP markers CADM1-m18, MAL-m1, and MIR124-2 to discriminate the presence of CIN2+ and, especially, cervical cancer from the mere presence of hrHPV DNA has mainly accumulated from screening studies in the Netherlands in which the same methodology as in our present report was used.18–2118–2118–2118–21
Bierkens et al19 showed that methylation levels of CADM1-m18 and MAL-m1 in cervical cell scrapes from 267 hrHPV-positive women increased proportionally to both the degree and duration of the underlying cervical lesion. This indicates that more advanced lesions are characterized by the presence of increased methylation levels, which finds support from our observation of higher methylation levels in samples of women with CIN3 compared with CIN2. Hesselink et al18 used a training set of 275 women to develop a combined CADM1-m18 and MAL-m1 bi-marker panel and then assessed its clinical performance in an independent validation set of 236 women. In that study, when applying assay thresholds corresponding to 75% specificity (resulting in 68% sensitivity) or 53% specificity (resulting in 84% sensitivity), the bi-marker panel was equally discriminatory for the detection of CIN3+ in hrHPV-positive women as cytology, or cytology and HPV16/18 genotyping, respectively. We note that the 2-phase design with separate training and validation data gives a better unbiased estimate of the sensitivity and specificity, but we were unable to follow this design because of the relatively small number of women in the present study.
Subsequently, Hesselink et al20 evaluated MIR124-2,17 in addition to CADM1-m18 and MAL-m1, in self-collected cervicovaginal lavage specimens from 355 hrHPV-positive women. The area under the curve for endpoint CIN3+ was larger for MAL-m1 (0.77) and MIR124-2 (0.76) than for CADM1-m18 (0.64). The relatively poor performance of CADM1-m18 in these self-collected samples was probably related to a lower yield of cervical cells compared with cervical scrapes.20 The authors eventually focused on MAL-m1 and MIR124-2 and reported that the bi-marker panel was robust at different assay thresholds and had at least a clinical performance in self-collected specimens comparable with that of HPV16/18 genotyping. Next, this bi-marker panel was prospectively evaluated on self-collected specimens in a randomized trial comparing direct methylation-based triage of hrHPV-positive nonattendees with cytology-based triage cervical cancer screening setting in the Netherlands.21 Here, methylation analysis of MAL-m1 and MIR124-2 on hrHPV-positive self-collected specimens was found not inferior to cytology triage for the detection of CIN2+. Molecular triage was logistically convenient as it did not require a referral for cervical scrape collection and reduced time to CIN2+ diagnosis, although this was at the cost of more colposcopy referrals compared with the cytology triage.
In the absence of previous studies of CADM1, MAL, and MIR124-2 in HIV-infected women, our present findings can only be compared with previous similar assessments of these methylation markers in the general female population in the Netherlands. As expected, hrHPV infections6 and cytological and histological abnormalities2 among the HIV-infected women originally recruited in our study in Kenya were much more frequent than in HIV-uninfected women.19,2019,20 However, among hrHPV-positive women, Ct ratios of CADM1-m18, MAL-m1, and MIR124-2 at baseline by the presence of CIN lesions in HIV-infected women were similar to those found in 30- to 60-year-old women participating in population-based cervical screening in the Netherlands and so was the strong trend of increasing methylation levels with the severity of cervical lesions.
ROC curves for CADM1-m18, MAL-m1, and MIR124-2 in HIV-infected hrHPV-positive women also show sensitivity and specificity for the detection of CIN2+ or CIN3 similar to previous reports in the general population.18,2018,20 The other indicators of the clinical performance of the tri-marker panel for detection of CIN3, including area under the curve (0.85), were also at least as good as those of bi-marker combinations of CADM1-m18 and MAL-m118 or MAL-m1 and MIR124-220,2120,21 in the Netherlands. Most importantly, the tri-marker panel we used was comparable with cytology and substantially better than VIA in the triage of HIV-infected hrHPV-positive women. We note that HPV16/18 genotyping showed especially low sensitivity (40%) for both CIN2+ and CIN3 because of the relatively higher proportion of CIN2+ lesions related to hrHPV types other than HPV16/18 among HIV-infected women.33 In addition, data from a comprehensive meta-analysis among general populations suggest that testing for HPV16/18 may be a less suitable method to detect CIN2/3 in Africa, as the proportion of HPV-positive CIN2/3 because of types 16 and/or 18 was the lowest in Africa (40%), with other regions in the world ranging from 45% to 72%.34 However, HPV16/18 accounts for the majority (67%) of ICC also in HIV-infected women in sub-Saharan Africa.35
Some support for the ability of methylation analysis of at least MAL-m1 and MIR124-2 to predict the future onset of high-grade lesions in hrHPV-positive women came from a small group of women who had CIN2+ diagnosed at a 6-month follow-up visit. Repeated testing for CADM1-m18, MAL-m1, and MIR124-2 in women who had ≤CIN1 at baseline and remained hrHPV positive at 6-month follow-up revealed a tendency for methylation levels of MAL-m1 and MIR124-2 to increase. This finding is consistent with the notion that methylation level increases with the duration of hrHPV infection or CIN.14 No clear changes in methylation levels were found in women with CIN2+ at baseline, although they had received destructive cervical treatment for their lesions. We note that persistence of hrHPV was common in this population even after the treatment of CIN2+, and we did not perform any random biopsy to exclude residual disease in women whose cytology was not HSIL+.26
An important strength of the present study is the systematic collection of baseline biopsies from all women, which increased the accuracy of cervical lesion ascertainment. Cytology and histological diagnoses were centrally reviewed, and hrHPV infection was assessed using a clinically validated assay.36 Finally, Ct ratios of CADM1-m18, MAL-m1, and MIR124-2 on DNA originating from cervical exfoliated cells have previously shown high reproducibility.20
The lack of a comparison group of HIV-uninfected women from the same source population in Kenya may be considered a weakness of our present study, but it should not eclipse the generalizability of our methylation findings to HIV-infected women who have access to cART in regions other than sub-Saharan Africa. We note that methylation levels are much higher in ICC than in CIN2 and CIN3.19 The lack of ICCs in our study implies, therefore, that the performance of CADM1-m18, MAL-m1, and MIR124-2 for CIN3 in our study is, if anything, an underestimate of the performance for CIN3+ in a hypothetical population of unscreened women who may also harbor ICC. In addition, the follow-up visit was made after 6 months because this coincided with the check-up visit of women who were treated for CIN2/3. A longer follow-up would have been more informative. However, the evidence on the good performance of methylation in the present article is mainly based on strong cross-sectional data. Obviously, the present report is not informative about the value of methylation as a stand-alone screening test.
In conclusion, a major asset of methylation markers is that they can be objectively assessed in the same cervical cell sample used for HPV primary screening. If well-designed, multiplexed high-throughput tests became available,37 methylation analyses may overcome the drawbacks of cytology, HPV16/18 genotyping, and VIA in the triage of many HIV-infected women who also test positive for hrHPV infection. A full molecular screening applicable to self-collected cell samples is an especially promising strategy in low-income countries where cervical cancer screening is hampered by lack of health professionals and, in some populations, women's reluctance to undergo a gynecological examination.38 However, to reach low-income populations, it would be important to have access to a lower cost and more user-friendly version of the test.
The authors acknowledge the work of Dr. Farzana S. Rana (deceased in 2012) who performed cytological and histological evaluations for the study. She was a great scientist, collaborator, and mentor on this project and they wish to dedicate this article to her memory.
Dr. G. Clifford provided useful comment. The authors also thank the research personnel, clinic and laboratory staff, and data management teams in Nairobi, Kenya; Seattle, WA; Amsterdam, the Netherlands; and Lyon, France, for their work. The authors recognize also the Coptic Hope Center for Infectious Diseases, Nairobi, Kenya, for their cooperation and our patients for their participation and support.
1. IARC. Monographs on the Evaluation of Carcinogenic Risks to Humans Volume 100B: A Review of Human Carcinogens: Biological Agents. Lyon, France: International Agency for Research on Cancer; 2012.
2. Chung MH, McKenzie KP, De Vuyst H, et al.. Comparing papanicolau smear, visual inspection with acetic acid and human papillomavirus
cervical cancer screening
methods among HIV-positive women by immune status and antiretroviral therapy. AIDS. 2013;27:2909–2919.
3. Kuhn L, Wang C, Tsai WY, et al.. Efficacy of human papillomavirus
-based screen-and-treat for cervical cancer prevention among HIV-infected women. AIDS. 2010;24:2553–2561.
4. Joshi S, Sankaranarayanan R, Muwonge R, et al.. Screening
of cervical neoplasia in HIV-infected women in India. AIDS. 2013;27:607–615.
5. Cuzick J, Arbyn M, Sankaranarayanan R, et al.. Overview of human papillomavirus
-based and other novel options for cervical cancer screening
in developed and developing countries. Vaccine. 2008;26(suppl 10):K29–K41.
6. De Vuyst H, Mugo NR, Chung MH, et al.. Prevalence and determinants of human papillomavirus
infection and cervical lesions in HIV-positive women in Kenya. Br J Cancer. 2012;107:1624–1630.
7. Giorgi-Rossi P, Franceschi S, Ronco G. HPV prevalence and accuracy of HPV testing to detect high-grade cervical intraepithelial neoplasia
. Int J Cancer. 2012;130:1387–1394.
8. Dillner J. Primary human papillomavirus
testing in organized cervical screening
. Curr Opin Obstet Gynecol. 2013;25:11–16.
9. Dijkstra MG, van ND, Rijkaart DC, et al.. Primary hrHPV DNA testing in cervical cancer screening
: how to manage screen-positive women? A POBASCAM trial substudy. Cancer Epidemiol Biomarkers Prev. 2014;23:55–63.
10. WHO. WHO Guidelines for Screening
and Treatment of Precancerous Lesions for Cervical Cancer Prevention. Geneva, Switzerland: WHO; 2013.
11. Stoler MH, Schiffman M. Interobserver reproducibility of cervical cytologic and histologic interpretations: realistic estimates from the ASCUS-LSIL Triage
Study. JAMA. 2001;285:1500–1505.
12. Wentzensen N, Sherman ME, Schiffman M, et al.. Utility of methylation
markers in cervical cancer early detection: appraisal of the state-of-the-science. Gynecol Oncol. 2009;112:293–299.
13. Lorincz AT. Cancer diagnostic classifiers based on quantitative DNA methylation
. Expert Rev Mol Diagn. 2014;14:293–305.
14. Steenbergen RD, Snijders PJ, Heideman DA, et al.. Clinical implications of (epi)genetic changes in HPV-induced cervical precancerous lesions. Nat Rev Cancer. 2014;14:395–405.
15. Overmeer RM, Henken FE, Snijders PJ, et al.. Association between dense CADM1 promoter methylation
and reduced protein expression in high-grade CIN and cervical SCC. J Pathol. 2008;215:388–397.
16. Overmeer RM, Henken FE, Bierkens M, et al.. Repression of MAL tumour suppressor activity by promoter methylation
during cervical carcinogenesis. J Pathol. 2009;219:327–336.
17. Wilting SM, van Boerdonk RA, Henken FE, et al.. Methylation
-mediated silencing and tumour suppressive function of hsa-miR-124 in cervical cancer. Mol Cancer. 2010;9:167.
18. Hesselink AT, Heideman DA, Steenbergen RD, et al.. Combined promoter methylation
analysis of CADM1 and MAL: an objective triage
tool for high-risk human papillomavirus
DNA-positive women. Clin Cancer Res. 2011;17:2459–2465.
19. Bierkens M, Hesselink AT, Meijer CJ, et al.. CADM1 and MAL promoter methylation
levels in hrHPV-positive cervical scrapes increase proportional to degree and duration of underlying cervical disease. Int J Cancer. 2013;133:1293–1299.
20. Hesselink AT, Heideman DA, Steenbergen RD, et al.. Methylation
marker analysis of self-sampled cervico-vaginal lavage specimens to triage
high-risk HPV-positive women for colposcopy. Int J Cancer. 2014;135:880–886.
21. Verhoef VM, Bosgraaf RP, van Kemenade FJ, et al.. Triage
-marker testing versus cytology in women who test HPV-positive on self-collected cervicovaginal specimens (PROHTECT-3): a randomised controlled non-inferiority trial. Lancet Oncol. 2014;15:315–322.
22. Nye MD, Hoyo C, Huang Z, et al.. Associations between methylation
of paternally expressed gene 3 (PEG3), cervical intraepithelial neoplasia
and invasive cervical cancer. PLoS One. 2013;8:e56325.
23. Vidal AC, Henry NM, Murphy SK, et al.. PEG1/MEST and IGF2 DNA methylation
in CIN and in cervical cancer. Clin Transl Oncol. 2014;16:266–272.
24. Chung MH, McKenzie KP, Richardson BA, et al.. Cervical HIV-1 RNA shedding after cryotherapy among HIV-positive women with cervical intraepithelial neoplasia
stage 2 or 3. AIDS. 2011;25:1915–1919.
25. Luff RD. The Bethesda System for reporting cervical/vaginal cytologic diagnoses. Report of the 1991 Bethesda workshop. Am J Clin Pathol. 1992;98:152–154.
26. De Vuyst H, Mugo NR, Franceschi S, et al.. Residual disease and HPV persistence after cryotherapy for cervical intraepithelial neoplasia
Grade 2/3 in HIV-positive women in Kenya. PLoS One. 2014;9:e111037.
27. Jacobs MV, Walboomers JM, Snijders PJ, et al.. Distribution of 37 mucosotropic HPV types in women with cytologically normal cervical smears: the age-related patterns for high-risk and low-risk types. Int J Cancer. 2000;87:221–227.
28. van den Brule AJ, Pol R, Fransen-Daalmeijer N, et al.. GP5+/6+ PCR followed by reverse line blot analysis enables rapid and high-throughput identification of human papillomavirus
genotypes. J Clin Microbiol. 2002;40:779–787.
30. Mendenhall W, Wackerly DD, Scheaffer RL. Nonparametric statistics. In: Mathematical Statistics With Applications. 4th ed. Boston, MA: PWS-Kent; 1989:674–679.
31. Belisle CJP. Convergence theorems for a class of simulated annealing algorithms on Rd. J Appl Prob. 1992;29:885–895.
32. Robin X, Turck N, Hainard A, et al.. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics. 2011;12:77.
33. Clifford GM, Goncalves MA, Franceschi S, et al.. Human papillomavirus
types among women infected with HIV: a meta-analysis. AIDS. 2006;20:2337–2344.
34. Guan P, Howell-Jones R, Li N, et al.. Human papillomavirus
types in 115,789 HPV-positive women: a meta-analysis from cervical infection to cancer. Int J Cancer. 2012;131:2349–2359.
35. De Vuyst H, Ndirangu G, Moodley M, et al.. Prevalence of human papillomavirus
in women with invasive cervical carcinoma by HIV status in Kenya and South Africa. Int J Cancer. 2012;131:949–955.
36. Bulkmans N, 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;370:1764–1772.
37. Snellenberg S, De Strooper LM, Hesselink AT, et al.. Development of a multiplex methylation
-specific PCR as candidate triage
test for women with an HPV-positive cervical scrape. BMC Cancer. 2012;12:551.
38. Gravitt PE, Rositch AF. HPV self-testing and cervical cancer screening
coverage. Lancet Oncol. 2014;15:128–129.
methylation; cervical intraepithelial neoplasia; human papillomavirus; triage; screening
Supplemental Digital Content
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.