Human papillomavirus infection, risk for subsequent development of cervical neoplasia and associated population attributable fraction
Introduction
Human papillomavirus (HPV) types 16 and 18 are carcinogenic to humans (IARC, 1995). Presence of HPV DNA confers the potential for development of a malignant phenotype for the cervical epithelial cell (zur Hausen, 1994). However, difficulty to assess cumulative HPV exposure in disease-free women, and increased detectability of HPV in the diseased women, due to replication/amplification of persisting viral DNA in cervical lesions/tumours (Cullen et al., 1991, Schiffman et al., 1993, IARC, 1995, Vernon et al., 1997, Liaw et al., 1999), have made it hard to quantify the impact of HPV exposure in, and its prevention to cervical carcinogenesis. Estimates at the population level are better based on longitudinal studies with multiple DNA tests or serology, but both have their pitfalls.
Prospective cohort studies of women diagnosed with HPV infection can use only surrogate end-points for cervical carcinoma such as squamous intraepithelial lesion (SIL) or cervical intraepithelial neoplasia (CIN) (Koutsky et al., 1992, Rozendaal et al., 1996, Ho et al., 1998, Moscicki et al., 1998, Liaw et al., 1999). Thus, estimates from such studies cannot directly be extrapolated to HPV associated relative risk (RR) of invasive cervical carcinoma (ICC). Recent archival studies, and longitudinal nested case–control studies have reported on the association of HPV infection and subsequent development of ICC and cervical carcinoma in situ (CIS) (Lehtinen et al., 1996, Dillner et al., 1997, Shah et al., 1997, Vonka et al., 1999, Wallin et al., 1999, Josefsson et al., 2000). Studies using archival cytological smears provide RR estimates of CIS and/or ICC associated with HPV exposure at the time of sample withdrawal but have applied different PCR methodology (Wallin et al., 1999, Josefsson et al., 2000). Longitudinal serological studies measure the cumulative HPV exposure up to the time of serum sample withdrawal but have limited sensitivity.
Previous estimates of the population attributable fraction or the population attributable risk percent (PAR%) of HPV in cervical carcinoma have been based on case–control studies (Lorincz et al., 1992, Pisani et al., 1997) and/or surrogate end-points (Schiffman et al., 1993, Winther et al., 1997). We reviewed the longitudinal studies to assess the PAR% of HPV16 in cervical carcinoma.
Section snippets
Criteria for selecting studies
We followed the BMJ recommendations (Egger et al., 1997, Egger and Davey-Smith, 1998, Egger et al., 1998) of selecting studies for review. All longitudinal studies on HPV and cervical neoplasia available until June 2000 were eligible. The studies had to fullfill the following criteria: follow-up with cohort, nested case–control approach among initially healthy women, or to be an archival (retrospective) study; definition of exposure (present HPV infection, measured by acceptable
Results
Overall 16 out of the 97 identified studies were found eligible for the review (Table 1). Five cohort studies that used PCR for the definition of exposure to oncogenic HPV types (including HPV16) or all genital HPV types, and one of the surrogate end-points (SIL/CIN1 or HSIL/CIN2-3), reported different RRs ranging from 6.7 to 116 (Table 1). On the contrary, corresponding three studies using DNA-hybridization for the identification of the genital HPV types revealed similar RRs ranging between 11
Discussion
In the longitudinal studies magnitude of the HPV associated RR of cervical carcinoma appeared to be largely determined by the method of assessing exposure to the virus (5–10 fold difference between PCR and serological studies). However, comparison of our weighted point estimate for the longitudinal PCR studies (RR=17) with previous systematic reviews of case–control PCR studies with RRs ranging from 50 to 100 (IARC, 1995, Pisani et al., 1997) suggests that both the design and the method of
Acknowledgements
The authors wish to thank Prof. Vladimir Vonka for providing individual exposure data for the meta-analysis. This study was supported by the Nordic Cancer Union and the Academy of Finland
References (40)
- et al.
Meta-analysis in controlled trials
Contr. Clin. Trials
(1986) - et al.
Viral load of human papillomavirus 16 as a determinant for development of cervical carcinoma in situ: a nested case control study
Lancet
(2000) - et al.
The natural history of human papillomavirus infection as measured by repeated DNA testing in adolescent and young women
J. Pediatr.
(1998) Human papillomavirus and cancer. The epidemiological evidence
J. Clin. Virol.
(2000)Disrupted dichotomous intracellular control of human papillomavirus infection in the cancer of the cervix
Lancet
(1994)- et al.
Persistence of human papillomavirus (HPV) infections preceding cervical carcinoma
Cancer
(1996) - et al.
A prospective study of the risk of cervical intra-epithelial neoplasia among healthy subjects with serum antibodies to HPV, comapred with HPV DNA in cervical smears
Int. J. Cancer
(1996) - et al.
Attributable risk percent in case–control studies
Br. J. Prev. Soc. Med.
(1971) - et al.
Analysis of physical state of different HPV DNAs in intraepithelial and invasive cervical neoplasias
J. Virol.
(1991) - et al.
Immunoglobulin G responses against HPV type 16 virus-like-particles in a prospective non-intervention study cohort of women with cervical intraepithelial neoplasia
J. Natl. Cancer Inst.
(1997)