Extended duration of the detectable stage by adding HPV test in cervical cancer screening

Abstract

The human papillomavirus test (HPV) test could improve the (cost−) effectiveness of cervical screening by selecting women with a very low risk for cervical cancer during a long period. An analysis of a longitudinal study suggests that women with a negative Pap smear and a negative HPV test have a strongly reduced risk of developing cervical abnormalities in the years following the test, and that HPV testing lengthens the detectable stage by 2–5 years, compared to Pap smear detection alone.

Main

One of the possible uses of the human papillomavirus test (HPV) is in primary cervical cancer screening in addition to or instead of the current Pap smear (Cuzick et al, 1999a; Cuzick, 2000; Meijer and Walboomers, 2000).

Introduction of HPV screening should be based on established (cost-) effectiveness. The (cost-) effectiveness of HPV testing is primarily determined by the duration of the detectable preclinical stage (the period from the HPV infection to clinical disease), and the sensitivity and costs of HPV testing. To estimate preclinical duration and sensitivity, longitudinal studies on the association between HPV infection and the development of neoplasias are necessary. Several large longitudinal screening studies have started, but no longterm results have been reported yet, although smaller longitudinal studies have been published (Rozendaal et al, 1996, 2000; Ho et al, 1998; Liaw et al, 1999). These studies differ with respect to HPV test used, age range of women, study design, and cytological or histological endpoint, which complicates the comparison and interpretation of these results. It is therefore too early for definite answers on the value of HPV testing in primary screening (Myers et al, 2000; Kim et al, 2002; Mandelblatt et al, 2002). However, available data can be explored to derive preliminary estimates for parameters that determine the cost-effectiveness of HPV testing. This study investigates the duration of the detectable preclinical stage using the results of Rozendaal et al (2000). For these estimates, the 5-year cumulative incidence of cervical intraepithelial neoplasia (CIN) III after a negative Pap smear, the current screening interval in the Netherlands, is compared with the cumulative incidence within a doubled screening interval of 10 years (Meijer and Walboomers, 2000) in case of a negative Pap smear and a negative HPV test.

Material and methods

The study population, and screening and follow-up results are described in Rozendaal et al (1996), (2000). Briefly, the smears obtained during routine screening from 1988 to 1991, from a cohort of 2250 women aged 34–54 years, who were either normal or who showed borderline nuclear changes were tested for a high-risk HPV. The women were followed during a mean period of 6.4 years, using screen-detected (histologically confirmed) CIN III as end point. Among the 2129 (95%) women with a negative HPV test at baseline, one case of CIN III was diagnosed at a following screening round. Of the 121 women with a positive HPV test result at baseline (5%), 12 women with CIN III were detected later. This resulted in a relative risk of 210, with a 95% confidence interval from 27 to 1600.

The disease model used in this study is schematically presented in Figure 1. Women without cervical disease or HPV may become infected with HPV. This infection may clear or it may progress to low-grade CIN. From ‘low-grade CIN’, the disease may regress spontaneously or progress to high-grade CIN (corresponding with CIN III), the end point of the model. We assume that CIN III cannot develop without HPV infection.

Figure 1
figure1

Schematic representation of the natural history model for HPV and CIN. IHPV=incidence rate of HPV infections in women without cervical disease or HPV, PCIN=probability that HPV will progress to CIN III, DHPV=duration of HPV infection preceding CIN and DCIN=duration of CIN I/II preceding CIN III.

In the model, we assumed a constant duration of the HPV infection, and an exponentially distributed duration of low-grade CIN with a mean of 6 years. The incidence rate of HPV infections in the age group considered (34–54 years) was set at five per 1000 woman years (Meijer, 1997). Using the results of the Rozendaal study, the duration of the HPV infection and the probability that the HPV infection will progress to CIN III were estimated. On the basis of these estimates, it was possible to calculate the cumulative incidence of CIN III within 5 years after the smear was taken; the current screening interval in the Netherlands, per 1000 cytologically negative women and the cumulative incidence of CIN III within 10 years after the smear was taken per 1000 cytologically negative/HPV negative women.

Initially, it was assumed that there were no diagnostic errors, that is, the results of the HPV test and the Pap smear as found by Rozendaal et al (2000) reflected the true disease stage of the women. We used this as our reference model. In alternative models, we studied the consequences of assuming diagnostic errors. We also varied the assumptions on the incidence rate of HPV infections and the duration of low-grade CIN. The mathematical description of the model is given in the Appendix.

Results

In Table 1, the results of the reference and alternative models are shown. Using the reference values for the model parameters, the duration of HPV infection before progressing to CIN was estimated at 3.8 years, resulting in a lower cumulative incidence of CIN III in 10 years for women with double-negative screening results, than in 5 years after a negative Pap smear and an unknown HPV result.

Table 1 Estimated values for the duration of HPV, the probability that the HPV infection will progress to CIN III, the 5-year cumulative incidence in women with a negative smear and the 10-year cumulative incidence in women with a negative smear and a negative HPV test

Next, we dropped the perfect-test assumptions (100% sensitivity) of the HPV test and Pap smear separately, by assuming 50% sensitivity for detecting an HPV infection and 50% sensitivity for detecting CIN I+, respectively. Furthermore, we halved and doubled our assumptions on the incidence rate of HPV infections and the duration of low-grade CIN. The estimated range for the duration of HPV before progressing to CIN widened, from 2 to 5 years. Only where the HPV test was assumed to have a sensitivity of 50% for HPV infections that will progress to CIN does the cumulative incidence 10 years after a double-negative result become slightly higher than the incidence within 5 years after a negative Pap smear.

Discussion

Our analysis shows duration of the HPV infection before it will progress into CIN of 2–5 years. For Pap smear screening, the preclinical duration was the combined duration of CIN and microinvasive cervical cancer, a period estimated at 15 years on average (Gustafsson and Adami, 1989; van Oortmarssen and Habbema, 1991, 1995; van Oortmarssen et al, 1992; Zielinski et al, 2001). Consequently, adding the HPV test to primary screening leads to duration of the detectable preclinical stage of almost 20 years in women aged 34–54 years. Furthermore, the 10-year cumulative incidence in women with a negative Pap smear and HPV test was lower than the 5-year cumulative incidence in women with a negative Pap smear and an unknown HPV result. This high negative predictive value of CIN III in double-negative women is the result of a longer preclinical duration and a better selection of women, as women with double-negative test results are at a lower risk of cervical cancer than women with only negative Pap smear results of whom part will have HPV infections.

These results suggest that an HPV test, in combination with the Pap smear, can considerably lengthen the screening interval in double-negative women (Meijer and Walboomers, 2000).

The confidence interval around the relative risk found by Rozendaal et al (1996) was large (from 27 to 1600). The upperbound of the confidence interval results in an HPV infection duration of 5.5 years. Assuming a relative risk corresponding to the lowerbound, results in a negligible duration. Even then, the 10-year cumulative incidence in double-negative women is lower than the 5-year cumulative incidence in women with a negative cytological result, assuming no diagnostic errors. This results from the very low risk of double-negative women of becoming infected with HPV infected, and subsequently developing cervical abnormalities in the years following the test. More firm estimates will be obtained on the basis of the results of the ongoing longitudinal studies.

The relative risks found in other studies (Ho et al, 1998; Liaw et al, 1999), 10.0 and 12.7, respectively, are lower than the range studied here. One of the reasons may be that the other studies concern women aged around 20 years. In young women, the occurrence of HPV infections is high (Melkert et al, 1993) and a much higher proportion of these infections are transient (Evander et al, 1995) compared to older women. Therefore, adding the HPV test in primary screening is not useful for young women (Cuzick et al, 1999b).

With the current model, it is technically not possible to lower the sensitivity of the HPV test and Pap smear simultaneously. Doing this will probably result in an estimate for the duration of the HPV infection of around 2 years. Also, the assumption of a constant duration of the HPV infection can be dropped using a more sophisticated model. However, these refinements pay off only when adequate longitudinal data on HPV detection are available for quantification of the additional parameters.

Women may develop CIN III without first passing the stages CIN I (and even CIN II). This situation has been represented by assuming a relatively short average duration of low-grade CIN of 2 years. Together with the assumption that this stage is exponentially distributed among women, this leads to a situation in which part of the women will develop CIN III shortly after having no neoplasia. Under these assumptions, the duration of the HPV infection before progressing to CIN is estimated to be relatively long, and the selection of low-risk women by adding HPV to cytology will be even better (incidence of CIN III 2.2 vs 4.5, Table 1).

The end point of the model was CIN III as imposed by the data. Invasive cancer is the end point to be preferred as prevention of invasive cancer and therefore, death is aimed at by cervical cancer screening. This end point, however, does not yield sufficient power due to the low risk for invasive cancer in Pap smear screened women, unless extremely large and long-term trials are performed. The current estimate on the duration of HPV before developing CIN is a combined estimate for the duration of HPV for women who will have a regressive CIN III lesion and those who will progress to cervical cancer. To solve the uncertainty on the confounding of regressive CIN III lesions, this prospective analysis with CIN III as end point should be accompanied by archival studies, in which retrospectively the HPV status of smears preceding a diagnosis of cervical cancer, is assessed. Zielinski et al (2001) concluded in a retrospective study of 57 women with invasive cervical cancer that the detectable preclinical stage could be prolonged by at least 2 years by adding HPV testing, which corroborates our results. This type of study, however, is susceptible to confounding biases such as selection and length time bias, which may result in an underestimation of the extension of the detectable preclinical stage as cervical cancers found after participation in a screening programme may be selective towards fast-growing cancers.

Doubling the screening interval for double-negative women will result in cost savings, as half of the screening rounds can be omitted. If, for example, the effectiveness of 10-yearly combined screening is the same as 5-yearly screening using the Pap smear, the costs of adding the HPV test to the Pap smear must be lower than these savings to be at least a cost-equal alternative. For a full cost analysis, other costs and savings should also be taken into account, such as the costs of follow-up in HPV positive/cytologically negative women, and possible savings due to a decrease in the detection of regressive cervical lesions because of a longer screening interval.

In conclusion, adding the HPV test to cytology in primary screening for cervical cancer results in an additional duration of the detectable preclinical stage of 2–5 years. Consequently, the screening interval for women with cytological and HPV-negative test results may be considerably lengthened. These results remain to be confirmed by the large longitudinal studies that are currently underway.

Change history

  • 16 November 2011

    This paper was modified 12 months after initial publication to switch to Creative Commons licence terms, as noted at publication

References

  1. Cuzick J (2000) Human papillomavirus testing for primary cervical cancer screening. JAMA 283: 108–109

  2. Cuzick J, Beverley E, Ho L, Terry G, Sapper H, Mielzynska I, Lorincz A, Chan WK, Krausz T, Soutter P (1999a) HPV testing in primary screening of older women. Br J Cancer 81: 554–558

  3. Cuzick J, Sasieni P, Davies P, Adams J, Normand C, Frater A, van Ballegooijen M, van den Akker E (1999b) A systematic review of the role of human papillomavirus testing within a cervical screening programme. Health Techno/Assess 3(i–iv): 1–196

  4. Evander M, Edlund K, Gustafsson A, Jonsson M, Karlsson R, Rylander E, Wadell G (1995) Human papillomavirus infection is transient in young women: a population-based cohort study. J Infect Dis 171: 1026–1030

  5. Gustafsson L, Adami HO (1989) Natural history of cervical neoplasia: consistent results obtained by an identification technique [see comments]. Br J Cancer 60: 132–141

  6. Ho GY, Palan PR, Basu J, Romney SL, Kadish AS, Mikhail M, Wassertheil-Smoller S, Runowicz C, Burk RD (1998) Viral characteristics of human papillomavirus infection and antioxidant levels as risk factors for cervical dysplasia. Int J Cancer 78: 594–599

  7. Kim JJ, Wright TC, Goldie SJ (2002) Cost-effectiveness of alternative triage strategies for atypical squamous cells of undetermined significance. JAMA 287: 2382–2390

  8. Liaw KL, Glass AG, Manos MM, Greer CE, Scott DR, Sherman M, Burk RD, Kurman RJ, Wacholder S, Rush BB, Cadell DM, Lawler P, Tabor D, Schiffman M (1999) Detection of human papillomavirus DNA in cytologically normal women and subsequent cervical squamous intraepithelial lesions. J Natl Cancer Inst 91: 954–960

  9. Mandelblatt JS, Lawrence WF, Womack SM, Jacobson D, Yi B, Hwang YT, Gold K, Barter J, Shah K (2002) Benefits and costs of using HPV testing to screen for cervical cancer. JAMA 287: 2372–2381

  10. Meijer CJLM (1997) Human Papillomavirus and Cervical Cancer Screening: Epidemiological Aspects of HPV Infection of Normal Cervical Epithelium in Relation to the Development of Abnormal Cytology [in Dutch]. Amsterdam: Department of Pathology, University Hospital Vrije Universiteit

  11. Meijer CJLM, Walboomers JMM (2000) Cervical cytology after 2000: where to go? J Clin Pathol 53: 41–43

  12. Melkert PW, Hopman E, van den Brule AJ, Risse EK, van Diest PJ, Bleker OP, Helmerhorst T, Schipper ME, Meijer CJ, Walboomers JM (1993) Prevalence of HPV in cytomorphologically normal cervical smears, as determined by the polymerase chain reaction, is age-dependent. Int J Cancer 53: 919–923

  13. Myers ER, McCrory DC, Nanda K, Bastian L, Matchar DB (2000) Mathematical model for the natural history of human papillomavirus infection and cervical carcinogenesis. Am J Epidemiol 151: 1158–1171

  14. Rozendaal L, Walboomers JM, van der Linden JC, Voorhorst FJ, Kenemans P, Helmerhorst TJ, van Ballegooijen M, Meijer CJ (1996) PCR-based high-risk HPV test in cervical cancer screening gives objective risk assessment of women with cytomorphologically normal cervical smears. Int J Cancer 68: 766–769

  15. Rozendaal L, Westerga J, van der Linden JC, Walboomers JM, Voorhorst FJ, Risse EK, Boon ME, Meijer CJ (2000) PCR based high risk HPV testing is superior to neural network based screening for predicting incident CIN III in women with normal cytology and borderline changes. J Clin Pathol 53: 606–611

  16. van Oortmarssen GJ, Habbema JD (1991) Epidemiological evidence for age-dependent regression of pre-invasive cervical cancer. Br J Cancer 64: 559–565

  17. van Oortmarssen GJ, Habbema JD (1995) Duration of preclinical cervical cancer and reduction in incidence of invasive cancer following negative pap smears. Int J Epidemiol 24: 300–307

  18. van Oortmarssen GJ, Habbema JD, van Ballegooijen M (1992) Predicting mortality from cervical cancer after negative smear test results. BMJ 305: 449–451

  19. Zielinski GD, Snijders PJ, Rozendaal L, Voorhorst FJ, van der Linden HC, Runsink AP, de Schipper FA, Meijer CJ (2001) HPV presence precedes abnormal cytology in women developing cervical cancer and signals false negative smears. Br J Cancer 85: 398–404

Download references

Author information

Correspondence to M E van den Akker-van Marie.

Appendix

Appendix

Mathematical formulae of the reference model

If we assume that:

  • incidence rate of HPV is IHPV,

  • probability that HPV will progress to CIN III is PCIN,

  • duration of HPV preceding CIN is DHPV and

  • duration CIN I/II preceding CIN III is exponentially distributed with mean DCIN.

Then, cumulative incidence of CIN III after x years for women without cervical lesions and without HPV infection at baseline

Cumulative incidence of CIN III after x years for women without cervical lesions but with HPV infection at baseline

Relative risk on CIN III in women not having a cervical lesion but being HPV infected compared to women without cervical lesion and HPV infection after x years

Cumulative incidence of CIN III x years after not having a cervical lesion

with pHPV− and pHPV+ the percentage of women without and with HPV infection at baseline, respectively.

Rights and permissions

From twelve months after its original publication, this work is licensed under the Creative Commons Attribution-NonCommercial-Share Alike 3.0 Unported License. To view a copy of this license, visit http://creativecommons.org/licenses/by-nc-sa/3.0/

Reprints and Permissions

About this article

Cite this article

van den Akker-van Marie, M., van Ballegooijen, M., Rozendaal, L. et al. Extended duration of the detectable stage by adding HPV test in cervical cancer screening. Br J Cancer 89, 1830–1833 (2003). https://doi.org/10.1038/sj.bjc.6601355

Download citation

Keywords

  • human papillomavirus
  • cervical cancer screening model

Further reading