Skip Navigation


American Journal of Epidemiology Advance Access originally published online on May 15, 2008
American Journal of Epidemiology 2008 168(2):145-148; doi:10.1093/aje/kwn129
This Article
Right arrow Extract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
168/2/145    most recent
kwn129v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Koshiol, J.
Right arrow Articles by Smith, J. S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Koshiol, J.
Right arrow Articles by Smith, J. S.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology Published by the Johns Hopkins Bloomberg School of Public Health 2008.

Response to Invited Commentary

The Authors Respond to "HPV Persistence and Cervical Cancer Screening"

Jill Koshiol1, Charles Poole2, Haitao Chu3,4, Jeanne M. Pimenta5, Lisa Lindsay6, David Jenkins6 and Jennifer S. Smith2

1 Cancer Prevention Fellowship Program, Division of Cancer Prevention, National Cancer Institute, Bethesda, MD
2 Department of Epidemiology, School of Public Health, University of North Carolina, Chapel Hill, NC
3 Department of Biostatistics, School of Public Health, University of North Carolina, Chapel Hill, NC
4 Lineberger Comprehensive Cancer Center, School of Medicine, University of North Carolina, Chapel Hill, NC
5 Worldwide Epidemiology, GlaxoSmithKline, Greenford, United Kingdom
6 GlaxoSmithKline Biologicals, Rixensart, Belgium

Correspondence to Dr. Jill Koshiol, National Cancer Institute, 6120 Executive Blvd., MSC 7236, Bethesda, MD 20892-7236 (e-mail: koshiolj{at}mail.nih.gov) or Dr. Jennifer S. Smith, CB 7435, McGavran-Greenberg Building, School of Public Health, University of North Carolina–Chapel Hill, Chapel Hill, NC 27599-7435 (e-mail: jennifers{at}unc.edu).

Received for publication March 27, 2008. Accepted for publication April 14, 2008.

Abbreviations: AUC, area under the ROC curve; CI, confidence interval; CIN, cervical intraepithelial neoplasia; HPV, human papillomavirus; HSIL, high-grade squamous intraepithelial lesions; ROC, receiver operating characteristic

We thank Dr. Castle for his constructive and contributory commentary (1) on our meta-analysis of human papillomavirus (HPV) persistence and cervical neoplasia (2). We agree that ascertaining HPV persistence may have clinical utility and that several issues must be addressed before persistence can be effectively implemented in cervical screening programs. As Dr. Castle mentioned, a single HPV test is more sensitive but less specific than a single Papanicolaou test for the detection of cervical precancer and cancer. In fact, HPV testing at a single time point has approximately 20–40 percent higher sensitivity than Papanicolaou testing and 5–10 percent lower specificity (3). Incorporating HPV persistence into routine cervical cancer screening may help to improve specificity without unduly reducing sensitivity by distinguishing women with short-term infections, and therefore lower risk of cervical precancer and cancer, from women with long-term infections, and therefore higher risk. To address this issue, we evaluated the diagnostic performance of HPV persistence as a predictor or marker for cervical intraepithelial neoplasia (CIN) 2–3 and high-grade squamous intraepithelial lesions (HSIL) or higher-level disease, including invasive cancer (termed CIN2–3/HSIL+). Although we combined the histologic (CIN2–3) and cytologic (HSIL) outcomes to maximize the number of studies available for meta-analysis (2), it is important to recognize that many of these outcomes may regress and not be linked to persistent carcinogenic HPV infection. Thus, the diagnostic performance of HPV persistence may in fact be better than presented here.

Using the studies that contributed to our meta-analysis of CIN2–3/HSIL+ (2), we estimated the sensitivity and specificity of each study and summary receiver operating characteristic (ROC) curves (4) for HPV persistence as a diagnostic marker for CIN2–3/HSIL+. We used the following definitions: 1) true positives: the number of women with HPV persistence and a diagnosis of CIN2–3/HSIL+; 2) false negatives: the number of women without HPV persistence but with a diagnosis of CIN2–3/HSIL+; 3) true negatives: the number of women without HPV persistence who were not diagnosed with CIN2–3/HSIL+; 4) false positives: the number of women with HPV persistence who were not diagnosed with CIN2–3/HSIL+. Sensitivity was calculated as true positives/(true positives + false negatives) and specificity as true negatives/(true negatives + false positives). The ROC curves were plotted in the usual way as sensitivity on the ordinate and 1 minus specificity (false-positivity rate) on the abscissa. Three ROC curves were estimated, depending on whether women with persistent infection were compared with 1) women who were consistently negative for the HPV types used to define persistence ("HPV-negative referent group"), 2) women with transient HPV infection ("transient HPV referent group"), or 3) a mix of women who were HPV-negative or had transient and/or incident HPV infection ("mixed HPV referent group"). The ROC curves were created using bivariate generalized linear mixed-effects models (5–7). The size of the symbol for each sensitivity-specificity pair (see figure 1) is proportional to the log of the sample size.


Figure 1
View larger version (13K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 1. Receiver operating characteristic curve for human papillomavirus (HPV) persistence as a predictor of high-grade cervical intraepithelial neoplasia, high-grade squamous intraepithelial lesions, and invasive cervical cancer, by HPV referent group, in a meta-analysis of the published literature. Data were obtained from the article by Koshiol et al. (2).

 
Sixteen studies had sufficient data available for calculation of both the sensitivity and specificity of HPV persistence in detecting CIN2–3/HSIL+ (8–23), including six studies with data for all three HPV referent groups (8, 12, 13, 15, 16, 22), one with data for the HPV-negative referent group only (17), three with data for the mixed HPV referent group only (18, 20, 23), and six with data for the transient HPV referent group only (9–11, 14, 19, 21) (figure 1). The mean or median age of study participants ranged from 25 years to 44 years (median, 37) for studies contributing to the HPV-negative referent group, from 25 years to 44 years (median, 36) for the mixed referent group, and from 24 years to 44 years (median, 38) for the transient HPV referent group.

Sensitivity for HPV persistence ranged from 42.8 percent to 100 percent (median, 100) for the HPV-negative referent group, from 26.0 percent to 100 percent (median, 77.3) for the mixed HPV referent group, and from 69.3 percent to 100 percent (median, 97.2) for the transient HPV referent group. Specificity ranged from 44.6 percent to 99.0 percent (median, 91.8) for the HPV-negative referent group, from 63.3 percent to 99.4 percent (median, 93.2) for the mixed HPV referent group, and from 45.8 percent to 97.2 percent (median, 54.0) for the transient HPV referent group. Variability in sensitivity and specificity estimates may reflect underlying differences in HPV persistence and cervical neoplasia definitions, study designs, and diagnostic techniques. The highest sensitivity summary estimate from the generalized linear mixed-effects model was 99.7 percent (95 percent confidence interval (CI): 68.7, 100.0) for the HPV-negative referent group, and the highest specificity summary estimate was 94.8 percent (95 percent CI: 87.2, 98.0) for the mixed HPV referent group (table 1).


View this table:
[in this window]
[in a new window]

  		TABLE 1. Summary estimates of the sensitivity and specificity (generalized linear mixed-effects model) of human papillomavirus (HPV) persistence as a predictor of high-grade cervical intraepithelial neoplasia, high-grade squamous intraepithelial lesions, and invasive cervical cancer, by HPV referent group, in a meta-analysis of the published literature*

 
As figure 1 shows, both the sensitivity and the specificity of HPV persistence were higher with the HPV-negative referent group than with either of the other two referent groups, as was the estimated area under the ROC curve (AUC) (HPV-negative referent group: 97.8 percent, 95 percent CI: 95.3, 99.1; mixed HPV referent group: 96.0 percent, 95 percent CI: 89.5, 98.1; transient HPV referent group: 87.0 percent, 95 percent CI: 72.6, 97.5). Using a novel alternative method that utilizes the estimated correlation between logit(sensitivity) and logit(specificity) for the calculation of estimated AUC (H. Chu and H. Guo, University of North Carolina, unpublished manuscript), the estimated AUC is 0.93 (95 percent CI: 0.75, 0.98) for the transient HPV referent group and the other two AUC estimates are essentially unchanged. With either the HPV-negative referent group or the mixed HPV referent group, ascertaining HPV persistence could potentially result in a specificity of 90 percent or higher while simultaneously maintaining a sensitivity of 90 percent or higher.

Women with transient or mixed HPV status may be the most clinically relevant comparison group, since future guidelines for diagnostic follow-up could potentially be based on persistent versus nonpersistent HPV infection. The comparatively poorer performance with the transient referent group is not surprising, since persistent and transient HPV infections are less likely to differ in predicting cervical CIN2–3/HSIL+ than persistent infection would be expected to differ from either mixed infection or no infection over the course of follow-up.

Incorporating HPV persistence into cervical screening may improve the balance of sensitivity and specificity. In a recent summary of single-time-point HPV testing, Cuzick et al. (24) reported an overall sensitivity for CIN2+ of 96.1 percent (95 percent CI: 94.2, 97.4) as compared with an overall sensitivity of 53.0 percent (95 percent CI: 48.6, 57.4) for cytologic screening. However, the overall specificity for <CIN2 was 90.7 percent (95 percent CI: 90.4, 91.1) as compared with 96.3 percent (95 percent CI: 96.1, 96.5) for cytologic screening (24). As presented here, the summary sensitivity and specificity for HPV persistence were 99.7 percent and 90.8 percent, respectively, for the HPV-negative referent group and 81.0 percent and 94.8 percent, respectively, for the mixed HPV referent group. Thus, the sensitivity of HPV persistence is higher than that of cytologic screening, while the specificity is higher than that of single-time-point HPV testing and may be similar to that of cytologic screening. The specificity of HPV testing may also be improved by screening older women (e.g., women over 30–35 years of age) (24).

These results must be interpreted with caution given the fact that the contributing studies had a variety of settings, study purposes, and follow-up times, and the analyses do not account for verification bias. In addition, up to 50 percent of HSIL regress, as does a high proportion of CIN2 (25). Thus, our estimates of sensitivity and specificity may be low because of the composite nature of the endpoint, which may include large numbers of transient lesions. Sensitivity and specificity are likely to be higher if the outcome is limited to CIN3 and higher-level disease. However, these results do suggest that HPV persistence may increase the specificity of HPV testing to near that of cytologic screening while maintaining an acceptably high sensitivity for the detection of cervical precancer and cancer.


   ACKNOWLEDGMENTS
 
Dr. Jill Koshiol's work on the original meta-analysis (2) was supported by the GlaxoSmithKline (GSK)/University of North Carolina (UNC) Center of Excellence from 2004 to 2005 and was directly supported by GSK for a brief period in 2005. Dr. Koshiol completed this work as a Cancer Prevention Fellow at the National Cancer Institute.

The authors thank Danielle Backes, Elizabeth Hodgson, and Dominique Luyts for their considerable aid in collecting and cleaning the data for the original meta-analysis (2).

All authors participated in designing this study and writing the manuscript. All analyses were conducted by Drs. Jill Koshiol and Haitao Chu.

Dr. Jill Koshiol was employed part-time by GSK from May 2001 to June 2005. Dr. Jennifer Smith has received grants or consulting fees from GSK. Dr. Charles Poole received partial salary support from the GSK/UNC Center for Excellence and an honorarium for one talk given at GSK. Drs. Jeanne M. Pimenta, Lisa Lindsay, and David Jenkins were full-time employees of GSK while working on the original meta-analysis (2). Dr. Jeanne M. Pimenta was a full-time employee of GSK and Dr. David Jenkins was a part-time employee of GSK while working on this article.


   References
TOP
 References
 

  1. Castle PE. Invited commentary: is monitoring of human papillomavirus infection for viral persistence ready for use in cervical cancer screening? Am J Epidemiol (2008) 168:138–44.[Abstract/Free Full Text]
  2. Koshiol J, Lindsay L, Pimenta JM, et al. Persistent human papillomavirus infection and cervical neoplasia: a systematic review and meta-analysis. Am J Epidemiol (2008) 168:123–37.[Abstract/Free Full Text]
  3. Franco EL, Cuzick J, Hildesheim A, et al. Chapter 20: issues in planning cervical cancer screening in the era of HPV vaccination. Vaccine (2006) 24((suppl 3)):S171–7.[CrossRef][Medline]
  4. Lasko TA, Bhagwat JG, Zou KH, et al. The use of receiver operating characteristic curves in biomedical informatics. J Biomed Inform (2005) 38:404–15.[CrossRef][Web of Science][Medline]
  5. Chu H, Cole SR. Bivariate meta-analysis of sensitivity and specificity with sparse data: a generalized linear mixed model approach. J Clin Epidemiol (2006) 59:1331–2.[CrossRef][Web of Science][Medline]
  6. Harbord RM, Deeks JJ, Egger M, et al. A unification of models for meta-analysis of diagnostic accuracy studies. Biostatistics (2007) 8:239–51.[Abstract/Free Full Text]
  7. Rutter CM, Gatsonis CA. A hierarchical regression approach to meta-analysis of diagnostic test accuracy evaluations. Stat Med (2001) 20:2865–84.[CrossRef][Web of Science][Medline]
  8. Bais AG, Rebolj M, Snijders PJ, et al. Triage using HPV-testing in persistent borderline and mildly dyskaryotic smears: proposal for new guidelines. Int J Cancer (2005) 116:122–9.[CrossRef][Web of Science][Medline]
  9. Beskow AH, Josefsson AM, Gyllensten UB. HLA class II alleles associated with infection by HPV16 in cervical cancer in situ. Int J Cancer (2001) 93:817–22.[CrossRef][Web of Science][Medline]
  10. Bory JP, Cucherousset J, Lorenzato M, et al. Recurrent human papillomavirus infection detected with the hybrid capture II assay selects women with normal cervical smears at risk for developing high grade cervical lesions: a longitudinal study of 3,091 women. Int J Cancer (2002) 102:519–25.[CrossRef][Web of Science][Medline]
  11. Cuschieri KS, Cubie HA, Whitley MW, et al. Persistent high risk HPV infection associated with development of cervical neoplasia in a prospective population study. J Clin Pathol (2005) 58:946–50.[Abstract/Free Full Text]
  12. Cuzick J, Szarewski A, Cubie H, et al. Management of women who test positive for high-risk types of human papillomavirus: The HART Study. Lancet (2003) 362:1871–6.[CrossRef][Web of Science][Medline]
  13. Dalstein W, Riethmuller D, Pretet JL, et al. Persistence and load of high-risk HPV are predictors for development of high-grade cervical lesions: a longitudinal French cohort study. Int J Cancer (2003) 106:396–403.[CrossRef][Web of Science][Medline]
  14. Harris TG, Kulasingam SL, Kiviat NB, et al. Cigarette smoking, oncogenic human papillomavirus, Ki-67 antigen, and cervical intraepithelial neoplasia. Am J Epidemiol (2004) 159:834–42.[Abstract/Free Full Text]
  15. Kjaer SK, van den Brule AJ, Paull G, et al. Type specific persistence of high risk human papillomavirus (HPV) as indicator of high grade cervical squamous intraepithelial lesions in young women: population based prospective follow up study. BMJ (2002) 325:572.[Abstract/Free Full Text]
  16. Liaw KL, Glass AG, Manos MM, et al. Detection of human papillomavirus DNA in cytologically normal women and subsequent cervical squamous intraepithelial lesions. J Natl Cancer Inst (1999) 91:954–60.[Abstract/Free Full Text]
  17. Moberg M, Gustavsson I, Gyllensten U. Type-specific associations of human papillomavirus load with risk of developing cervical carcinoma in situ. Int J Cancer (2004) 112:854–9.[CrossRef][Web of Science][Medline]
  18. Paraskevaidis E, Kaponis A, Malamou-Mitsi V, et al. The natural history of HPV infection of the uterine cervix. Long-term observational and histological data. Anticancer Res (2002) 22:1177–81.[Web of Science][Medline]
  19. Peto J, Gilham C, Deacon J, et al. Cervical HPV infection and neoplasia in a large population-based prospective study: the Manchester cohort. Br J Cancer (2004) 91:942–53.[Web of Science][Medline]
  20. Schiffman M, Herrero R, DeSalle R, et al. The carcinogenicity of human papillomavirus types reflects viral evolution. Virology (2005) 337:76–84.[CrossRef][Web of Science][Medline]
  21. ter Harmsel B, Smedts F, Kuijpers J, et al. Relationship between human papillomavirus type 16 in the cervix and intraepithelial neoplasia. Obstet Gynecol (1999) 93:46–50.[CrossRef][Web of Science][Medline]
  22. Wallin KL, Wiklund F, Angstrom T, et al. Type-specific persistence of human papillomavirus DNA before the development of invasive cervical cancer. N Engl J Med (1999) 341:1633–8.[Abstract/Free Full Text]
  23. Ylitalo N, Josefsson A, Melbye M, et al. A prospective study showing long-term infection with human papillomavirus 16 before the development of cervical carcinoma in situ. Cancer Res (2000) 60:6027–32.[Abstract/Free Full Text]
  24. Cuzick J, Clavel C, Petry KU, et al. Overview of the European and North American studies on HPV testing in primary cervical cancer screening. Int J Cancer (2006) 119:1095–101.[CrossRef][Web of Science][Medline]
  25. Jenkins D. Histopathology and cytopathology of cervical cancer. Dis Markers (2007) 23:199–212.[Web of Science][Medline]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?

Related articles in Am. J. Epidemiol.:

Persistent Human Papillomavirus Infection and Cervical Neoplasia: A Systematic Review and Meta-Analysis
Jill Koshiol, Lisa Lindsay, Jeanne M. Pimenta, Charles Poole, David Jenkins, and Jennifer S. Smith
Am. J. Epidemiol. 2008 168: 123-137. [Abstract] [FREE Full Text]  




This Article
Right arrow Extract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
168/2/145    most recent
kwn129v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Related articles in Am. J. Epidemiol.
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Koshiol, J.
Right arrow Articles by Smith, J. S.
Right arrow Search for Related Content
PubMed
Right arrow Articles by Koshiol, J.
Right arrow Articles by Smith, J. S.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?