Is human papillomavirus an infectious cause of non-cervical anogenital tract cancers?

BMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7109.620 (Published 13 September 1997) Cite this as: BMJ 1997;315:620
  1. Howard D Strickler, Senior clinical investigatora,
  2. Mark H Schiffman, Chief interdisciplinary studies sectionb
  1. a Viral Epidemiology Branch
  2. b Environmental Epidemiology Branch, National Cancer Institute, Bethesda, Md 20892, USA

    Results from a large study provide strong epidemiological evidence

    “Temporality” is a central tenet in the epidemiological assessment of causality, requiring that exposure to a putative cause must precede development of the disease.1 However, temporality has been difficult to demonstrate in the study of human papillomavirus infection as a cause of non-cervical anogenital tract tumours.

    In cervical cancers human papillomavirus DNA can be detected in over 90% of lesions. Moreover, cervical human papillomavirus has been detected before the development of cervical intraepithelial neoplasia, which is well established to be a precursor of cancer.2 In non-cervical anogenital tumours a common causal relation with human papillomavirus is suggested by the raised risk of anal, vulvar, and vaginal tumours after cervical cancer3; a high prevalence of penile intraepithelial neoplasia in the sexual partners of women with cervical intraepithelial neoplasia4; and the high prevalence of human papillomavirus DNA in non-cervical anogenital cancer tissues.5 Infection of normal non-cervical anogenital epithelium may be as common as in the cervix.6

    Thus, the anogenital tract can be considered to be a generally susceptible region for epithelial human papillomavirus infections, with each local area a potential viral reservoir and source of autoinoculations. Human papillomavirus infection of the cervical transformation zone, however, produces the major burden of anogenital cancers. Indeed, non-cervical anogenital cancers are rare, and tissue specimens from before the development of non-cervical anogenital neoplasia are not generally available from prospective studies of human papillomavirus infection. Using recently validated serological methods, Bj⊘rge et al, in this edition of the BMJ, present the first major study to examine the relation of exposure to human papillomavirus to the later development of non-cervical anogenital cancers.7

    The researchers conducted a nested case-control investigation of the seroprevalence of human papillomavirus antibodies and subsequent anogenital tumours in Finland and Norway by matching data in the cancer registries in the two countries to nearly 700 000 subjects who had provided stored blood samples at various earlier times. The pre-disease specimens were tested for IgG to virus-like particles of human papillomavirus 16, 18, and 33—three types commonly detected in specimens from anogenital tumours. They tested 81 patients with non-cervical anogenital cancer and 240 matched controls. Human papillomavirus antibodies, but not antibodies to another sexually transmitted infection—Chlamydia trachomatis, were more common in people who later developed anogenital cancers, including vulvar, vaginal, and penile cancers. Anal cancer, however, was not related to seroprevalence of these three types of human papillomavirus.

    The findings suggest that human papillomavirus infection is associated with later diagnosis of vulvar, vaginal, and penile cancers. This inference is further reinforced by the consistency of these findings with results from prospective human papillomavirus serological studies of patients who later developed cervical cancer.8 9 There are caveats, however. In the investigation by Bj⊘rge et al only the results in vulvar and vaginal cancer were statistically significant, and the absence of an anti-human papillomavirus IgG association with anal cancer was not explained—although it may have been due to a chance increase in seroprevalence among the small group of matched controls. It would also have been useful to know whether the cases positive for human papillomavirus antibody developed tumours containing DNA of the same types as implicated by serology. In addition, consideration of pathological subtypes of cancers in an investigation with longer follow up is needed. For example, warty and basaloid vulvar tumours in younger women are generally associated with human papillomavirus, but not the typical keratinising carcinomas of older women.10 Lastly, repeated observations over time, rather than a single serological result for each patient, would further confirm these causal relations. Nevertheless, given the rarity of non-cervical anogenital tumours and the general unavailability of large collections of prospectively stored specimens, it is unlikely that many other research groups could expand considerably on these results. Thus, Bj⊘rge et al have provided an important and difficult to obtain piece of evidence demonstrating the probable relation between human papillomavirus infection and later development of non-cervical anogenital cancers.

    Prospective and large population based studies like this show that human papillomavirus serology is an increasingly useful biomarker. Its sensitivity and specificity are not optimal, however. For example, only about half the women positive for cervical human papillomavirus 16 DNA are antibody positive.11 Therefore, human papillomavirus seroassays remain primarily epidemiological research tools. Overall, the major public health importance of the findings by Bj⊘rge et al is that a successful human papillomavirus vaccine could have benefits beyond the primary goal of preventing cervical cancer.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    View Abstract