Serologically diagnosed infection with human papillomavirus type 16 and risk for subsequent development of cervical carcinoma: nested case-control study
BMJ 1996; 312 doi: https://doi.org/10.1136/bmj.312.7030.537 (Published 02 March 1996) Cite this as: BMJ 1996;312:537- Matti Lehtinen, senior research fellowa,
- Joakim Dillner, senior investigatorb,
- Paul Knekt, senior scientistc,
- Tapio Luostarinen, research fellowd,
- Arpo Aromaa, director of medical researchc,
- Reinhard Kirnbauer, senior research fellowe,
- Pentti Koskela, senior scientista,
- Jorma Paavonen, associate professor of obstetrics and gynaecologyf,
- Richard Peto, professor of medical statisticsg,
- John T Schiller, senior investigatorh,
- Matti Hakama, professor of epidemiologyi
- a National Public Health Institute, Helsinki and Oulu, Finland
- b Microbiology and Tumorbiology Centre, Karolinska Institute, Sweden
- c Research and Development Centre, Social Insurance Institution, Helsinki, Finland
- d Finnish Cancer Registry, Helsinki, Finland
- e Department of Immunodermatology, University of Vienna Medical School, Vienna, Austria
- f University of Helsinki, Finland
- g Clinical Trial Service Unit and ICRF Cancer Studies Unit, University of Oxford, Oxford
- h Laboratory of Cellular Oncology, National Cancer Institute, Bethesda, maryland, United States
- i Tampere School of Public Health, University of Tampere, Tampere, Finland
- Correspondence to: Dr Lehtinen, Department of Chronic Viral Diseases, NPHI, Mannerheimintie 166, FIN-00300 Helsinki, Finland.
- Accepted 30 November 1995
Abstract
Objective: To study human papillomavirus type 16 in the aetiology of cervical carcinoma.
Design: Within a cohort of 18814 Finnish women followed for up to 23 years a nested case-control study was conducted based on serological diagnosis of past infection with human papillomavirus type 16.
Subjects: 72 women (27 with invasive carcinoma and 45 with in situ carcinoma) and 143 matched controls were identified during the follow up.
Main outcome measure: Relative risk of cervical carcinoma in presence of IgG antibodies to human papillomavirus type 16.
Results: After adjustment for smoking and for antibodies to various other agents of sexually transmitted disease, such as herpes simplex virus type 2 and Chlamydia trachomatis, the only significant association was with infection with human papillomavirus type 16 (odds ratio 12.5; 95% confidence interval 2.7 to 57, 2P<0.001).
Conclusion: This prospective study provides epidemiological evidence that infection with human papillomavirus type 16 confers an excess risk for subsequent development of cervical carcinoma.
Key messages
Key messages
Human papillomavirus type 16 is the main micro-organism linked to the development of cervical cancer
Prospective studies of infection with this virus and cervical cancer have not been reported because of ethical and clinical difficulties and because diagnosis of past infections with the virus has not been possible
In this nested case-control study in over 18000 Finnish women who donated blood to a serum bank 25 years ago we were able to measure past infection with human papillomavirus type 16 with new serological tools
The results show that infection with the virus confers an increased risk of developing cervical cancer
Introduction
Infection with human papillomavirus type 16 (HPV16) is the major factor that has been linked to cervical neoplasia.1 2 But no prospective studies of infection with human papillomavirus and cervical carcinoma have yet been reported. We have previously reported the risks of cervical carcinoma associated with various other sexually transmitted diseases in a cohort of 18814 Finnish women followed for 12 years.3 Only Chlamydia trachomatis infection was associated with an increased risk. A recently developed serological assay provides a type restricted measure of infection with human papillomavirus type 16.4 This and extension of the maximum follow up time to 23 years enabled us (a) to determine whether infection is a particularly strong risk factor for subsequent development of cervical carcinoma and (b) to determine whether any risk associated with other sexually transmitted infections is independent of the risk associated with exposure to this papillomavirus.
Subjects and methods
Cases and controls were identified as follows. The Finnish Social Insurance Institution carried out a mobile health examination survey among 30 different population groups in various parts of Finland during 1966-72. More than 30000 women (aged 15 years or more) were invited to a health examination, which included asking about medical history and smoking habits and taking a blood sample. The serum samples of 18814 women were stored at -20°C.
The population based Finnish cancer registry receives reports of cancer cases from hospitals, pathology laboratories, and physicians throughout Finland. Fewer than 200 cases each of carcinoma in situ (excluding cervical intraepithelial neoplasia) and invasive cervical carcinoma are reported annually. All women who had given blood in the mobile health examination survey during 1968-72 and were free of cancer at the baseline were followed. Those who had cervical carcinoma diagnosed after the baseline examination were identified by linking the data files of the mobile health examination survey and the Finnish cancer registry. Until 1991, 72 cases of cervical carcinoma (27 invasive cervical carcinoma and 45 carcinoma in situ) were diagnosed. Altogether 143 women individually matched for sex, age, and municipality were identified to act as controls. Age was matched by using nearest available matching: in 61 sets the age was exactly matched, in nine sets at least one of the controls differed by one to two years and in two sets by three to four years. Mean (range) age at the baseline was 39.1 (15-83) years and at the diagnosis 49.2 (22-95) years. Mean (range) time (follow up time) between withdrawal of serum and diagnosis of cervical carcinoma was 10.1 (0.7-22.8) years.
IgG antibody analyses were performed by standard enzyme linked immunosorbent assay (ELISA). For human papillomavirus type 16 analysis baculovirus expressed capsids purified by ultracentrifugation and comprising both the L1 and L2 proteins were used with bovine papillomavirus capsids as controls.4 5 6 We used C trachomatis elementary body to detect chlamydia infection, antigen to lysate from cells infected with herpes simplex virus type 1 for herpes simplex virus, and glycoprotein gG-2 from herpes simplex virus type 2 for herpes simplex virus type 2.3 7 The same standardised reference serum samples, antihuman IgG enzyme conjugates, and cut off levels were used as in previous studies.3 4 5 6 7 The specificity of the chlamydia antibodies for C trachomatis was confirmed by identifying ELISA positive cases who were microimmunofluorescence positive (>1:32) for the C trachomatis serovars B,E,D/C,J,H,I/G,F,K/(Washington Research Foundation, Seattle)8 and for the solely genital C trachomatis serovar G,F,K,9 respectively.
Relative risks of cervical carcinoma (odds ratios, 95% confidence intervals, and two tailed P values) were estimated by conditional logistic regression for matched sets of case-control triplets (one case, two controls)10 or by the exact inference methods for contingency tables11 with EGRET and EPIXACT software (Statistics and Epidemiology Research Corporation, Seattle).
Results
In univariate analyses we found that type restricted antibodies to the papillomavirus and C trachomatis were both associated with an increased risk of cervical carcinoma (odds ratio 13.2, P<0.001; 2.1, P<0.05, respectively), but only for papillomavirus was the association highly significant. The risk associated with the presence of antibodies specific for the Chlamydia genus was modest and only just conventionally significant (1.8, 2P=0.05), and although the risk associated with the presence of antibodies to the genital C trachomatis serovars G,F,K was higher, it too was only just conventionally significant (3.4, P=0.05). Adjustment for smoking had only minor effects on the estimates, but it made all but the papillomavirus association cease to be conventionally significant (each P>0.05; table 1). The papillomavirus association was seen both for patients with a short time from sampling to diagnosis (n=21) and for those with a long time (n=51) (odds ratio for <5 years 8.6, 95% confidence interval 1.0 to 75; for >5 years 18, 2.3 to 142) as well as for patients with invasive cervical carcinoma (n=27) and for those with carcinoma in situ (n=45) ((infinity), 2.0 to (infinity); 6.0, 1.2 to 29.7). Finally, we used a multivariate model to evaluate the relative risk associated with papillomavirus and herpes simplex virus type 2 infections and infections with genital C trachomatis serovars G,F,K (table 1). Again only the papillomavirus association remained significant (12.5; P<0.001).
Odds ratios (95% confidence interval) of cervical carcinoma according to presence of IgG antibodies to different sexually transmitted diseases
Discussion
Between 65% and 95% of cervical carcinomas contain human papillomavirus DNA, and about half carry type 16.12 Also about half of patients with cervical carcinoma are positive for antibodies to papillomavirus type 16.5 13 In our study nearly a quarter of the women who subsequently developed cervical carcinoma had detectable antibodies to papillomavirus type 16 at the baseline. Possibly additional women may have become infected14 during the average lag of 10 years (and up to 23 years) between withdrawal of serum and diagnosis of cervical carcinoma.
In line with most cross sectional seroepidemiological studies on cervical carcinoma12 the multivariate analyses found papillomavirus type 16 to be a far stronger risk factor than any other. Although the analysis for the genital C trachomatis serovars G,F,K may have suffered from small numbers, our results suggest that the increased risk associated with other sexually transmitted disease agents may be secondary to papillomavirus type 16, reflecting high risk sexual behaviour.
Presence of antibodies to papillomavirus type 16 increased the risk for subsequent development of both carcinoma in situ and invasive cervical carcinoma. Furthermore, the highest risk was found for cervical carcinoma diagnosed more than five years after serum withdrawal, which is the expected finding if persistent infection with papillomavirus type 16 is causally involved in the aetiology of this disease.
Footnotes
-
Funding ML is supported by research grants from the Nordic Cancer Union and Finnish Cancer Organisations.
-
Conflict of interest None.