Chlamydia and Cervical Cancer: Title and subTitle BreakA Real Association?

Worldwide, cervical cancer is among the 3 most prevalent malignancies in women, with more than 400 000 cases reported in 1998, of which 12 800 were in the United States.¹ The causal relationship between subtypes of human papillomavirus (HPV) and cervical cancer has been firmly established. The epidemiological data are compelling and consistent, most recently from nested case-control studies in Scandinavia that demonstrated HPV in specimens collected before incident cervical intraepithelial neoplasia (CIN).^{2 - 3}

However, there are still important unanswered questions related to pathogenesis. Human papillomavirus infection is common. In adolescent populations, infection rates as high as 44% to 90% have been observed.^{4 - 6} Recent data strongly suggest that the majority of persons with HPV spontaneously clear the infection without any specific medical intervention. Only a small minority of these infected persons develop cervical dysplasia, and of these, a minority develop invasive carcinomas. Therefore, the involvement of cofactors in the pathogenesis of HPV-related complications has been suspected.

From a methodological standpoint, identifying cofactors for cervical cancer is difficult. As a sexually transmitted infection (STI), HPV has numerous potential behavioral and clinical confounders. For example, sexual behavior variables such as number of partners also are related to infection with other STIs. Since the progression from HPV infection to CIN takes years, cofactors could be operative at different points in the natural history of HPV infection, either during incident infection, having an impact on potential clearance, or during the prolonged period of latent infection. Study designs that can tease out these numerous variables are critical. Because of the relatively low incidence of cervical carcinoma, large populations are required to sort out these issues. Smoking, to which people are typically exposed long term, has been consistently identified as a risk for progression of HPV infection to CIN.^{5 - 8} In serological studies, smoking at the time of serum collection can be validated by measurement of cotinine levels.

The literature on the causes of cervical cancer is replete with classic tales of epidemiologic confounders. Before the discovery of HPV, carefully performed studies attempted to link cervical cancer with STIs. In 1970, Nahmias et al⁹ demonstrated in a case-control study of Atlanta maternity patients that herpes simplex virus 2 was associated with cervical cancer. Schachter et al¹⁰ similarly attempted to determine the association between Chlamydia trachomatis infection and cervical cancer. These studies did not control for sexual behavior, nor did they account for the yet-undiscovered HPV, and are in retrospect inconclusive.

Later associations of cervical cancer with C trachomatis were identified in several of the initial investigations that linked HPV to cervical carcinoma. In an STI clinic study conducted between 1984 and 1986, Koutsky et al¹¹ evaluated the association of CIN with HPV. After adjustment for HPV status, CIN was also associated independently with serological evidence of either infection with chlamydia, gonorrhea, or cytomegalovirus. Lehtinen et al¹² used data from a Finnish national mobile health examination survey conducted between 1966 and 1972 to link results with subsequent cervical cancers reported to the national registry. They found that prior chlamydial infection was significantly associated with cervical carcinoma, with higher levels of significance for serotype pool GFK, but the study was limited by small numbers of cancers and the lack of multivariate analysis. In contrast, cross-sectional studies in southwestern US populations¹³ and a series of case-control studies in Honduras,¹⁴ Colombia, and Spain¹⁵ failed to demonstrate an association between chlamydial infection (or any other STI) and cervical cancer.

The study by Anttila and colleagues¹⁶ in this issue of THE JOURNAL is 1 of a series of investigations that have used data from large serum banks (in this case, collected from more than 530 000 individuals in 3 Nordic countries) and matched these data with subsequent cervical cancer cases collected by national cancer registries.¹⁷ Since the baseline specimens were collected an average of 56 months before the cytological diagnosis of cervical cancer, the causal inference is substantial.

The Nordic studies strongly suggest that there is an independent relationship between C trachomatis infection and cervical squamous cell carcinoma (SCC), after controlling for HPV infection and smoking. In the study by Anttila et al, multiple tests for C trachomatis serology were used with an emphasis on specificity. Highest point estimates involved longer lag times between donating the serum and developing cervical cancer. The authors note that specific serotypes (G, I, and D) were more likely to be associated with SCC than the others.

These data, as the authors acknowledge, parallel a previous report from Finland that suggests that the serotype pool GFK is associated with the development of CIN.¹² However, the findings of Anttila et al, while provocative, are limited by the small sample sizes used for these estimates and by the consequent wide confidence limits (as demonstrated in Table 1 of their article). Confirmation of this finding is therefore important. Anttila et al also report that C trachomatis DNA was found in 5% of the cases of CIN. Since C trachomatis causes a curable bacterial infection, DNA is not likely to be present in situations where the infection has already been treated. Also, the DNA may represent new, recent chlamydial infections (ie, those occurring after malignant transformation had already occurred).

Despite its elegant nested case-control design, the study has limitations. There are limited behavioral data associated with either the baseline serum measurement or the subsequent CIN diagnosis. Therefore, the possibility of an additional unmeasured behavioral or biological confounder cannot be excluded.

Nevertheless, the results lead to a number of intriguing questions. How does C trachomatis enhance the progression of HPV to invasive cervical cancer? Since HPV viral load has been associated with subsequent development of CIN,¹⁸ does bacterial infection, or the consequent mucosal inflammation, induce a transactivation-like response resulting in an increased viral load? Does chlamydial infection or other STI modify the host inflammatory or immunological response to HPV? Do these infections modify the spontaneous clearance rate of HPV? Chlamydia can present as an acute infection or as one that is asymptomatic and persistent. Is the cancer-promoter effect related to duration of infection? Is chlamydia an independent risk factor?

The data also provide potential clinical relevance to the chlamydia typing system. The chlamydia serotypes have been used mostly as epidemiologic research tools. Specific serotypes have been linked to clinical syndromes: serotype D with pelvic inflammatory disease,¹⁹ serotypes F and G with symptomatic rectal infection,²⁰ serotype F with asymptomatic cervical infection,²¹ and more recently, serotype C with persistent cervical infection.²² All of these studies, however, had small sample sizes and the results are therefore not generalizable.

Until recently, the accumulation of these serotype data would have been little more than a curiosity to nonepidemiologists. If the results of Anttila and colleagues (that serotypes D, G, and I are related to increased rates of CIN) are confirmed, a series of public health responses could follow. The obvious research question to evaluate is what outer membrane protein components are involved in the potential promoter effect. The sequencing of the chlamydia genome²³ facilitates further research into determining the membrane and genetic correlates of these findings, all of which should allow more insight into the pathogenesis of chlamydia and its complications.

Moreover, the findings of Anttila et al suggest that cervical malignancy should be added to the complications and costs associated with genital chlamydial infections. These complications include pelvic inflammatory disease, infertility, and facilitated transmission of human immunodeficiency virus. These data provide additional justification for expanding chlamydial infection screening, not only to protect against pelvic inflammatory disease and infertility, but potentially to prevent cervical cancer.