From April 1997 to October 1999, 141 women with putative invasive cervical cancer seen at the hospitals during the study period were asked to participate. Of these, one woman declined and eight were excluded because of lack of histological confirmation, leaving 132 case participants with histologically confirmed cervical cancer. Of these 132, 114 had been recruited from the first hospital and 18 from the second hospital. No significant differences were observed between these two groups with respect to demographics, economic status, the number of lifetime sexual partners, parity or HIV infection status. A total of 135 consenting women were enrolled as potential controls, representing approximately 70% of those eligible from those recruited. Of these 135 women, 120 with normal cervical cytology results were retained as controls; 14 women with cervical dysplasia (including Atypical Squamous Cells of Undetermined Significance (ASCUS) and SIL) and one with missing cytology results were excluded.
Characteristics of the study participants
Table shows the demographic characteristics and distribution of risk factors for cervical cancer among cases and control participants. Case participants were slightly older than controls and had higher percentages with low economic status and with no formal education. However, cases and controls had similar percentages who reported having had first sex before 16 years and having had > 6 sexual partners during their lifetime. Cigarette smoking was rare both among cases and among control participants. Twenty-two (16.7%) women with cervical cancer and 10 (8.3%) control women were HIV-infected (p = 0.048). None of the women with HIV infection were on antiretroviral therapy (ART); during the study period, ART was not routinely available in Côte d'Ivoire.
Characteristics of case and control participants
The five most prevalent types of HPV found in cervical cancer specimens were: HPV16 (45%), HPV18 (21%), HPV45 (10%), HPV35 (8%), and HPV31 (3%). Other high-risk HPV types found were: 33, 39, 51, 52, 53, 56, 58, 59, 66, and 68. The HPV types in cases and control specimens were similar. However, the prevalence of high-risk HPV infection was higher among women with cervical cancer than among controls (89.4 vs. 31.1; p < 0.0001).
Table presents risk factors (covariates) of HIV infection among control participants. The percentage of women with low economic status, or women with no formal education did not significantly differ between HIV-positive and HIV-negative controls. However, a higher percentage of HIV-positive controls were from low socioeconomic status. HIV infection was significantly associated with age < 40 years, parity > 2, lifetime number of sex partners > 6, and having high-risk HPV infection.
Covariates of HIV infection among control participants
High-risk HPV types were detected in all (100%) of the HIV-positive women with cervical cancer and in 98/110 (89%) of HIV-negative women with cervical cancer (p = 0.1). In multivariate analysis, only HIV infection was associated with HPV infection among controls (OR = 3.0; 95% CI 0.7-13.5). A stronger association was observed among younger women, but the number of samples were small and the difference was not statistical significant (OR = 6.8; 95% CI 0.5-98.0).
Risk factors for cervical cancer
Table shows risk factors associated with cervical cancer in univariate and logistic regression analyses. In univariate analyses, women with cervical cancer were more likely to be either HIV-1-infected or HIV-2-infected than were control women; however, these differences were not statistically significant. In logistic regression analysis, when controlling for age, low socioeconomic status, and lifetime number of sex partners > 6, the variables that remained associated with cervical cancer were having high-risk HPV infection (OR 23.0; 95% CI 10.5-50.2), parity > 2 (OR 5.5; 95% CI 2.3-13.4), and low economic status (OR 2.4; 95% CI 1.2-4.9). However, when high-risk HPV infection was removed from the model, HIV-1 infection or dual HIV infection was associated with cervical cancer (OR 3.4; 95% CI 1.4-8.3). Although HIV-2 infection was also associated with having cervical cancer, the association did not achieve statistical significance, perhaps due to the small number of HIV-2-positive women (OR 3.9; 95% CI 0.4-35.6). The positive association of HIV infection and cervical cancer remained unchanged when the analysis was restricted to case and control participants with evidence of high-risk HPV DNA (OR 3.4; 95% CI 1.1-10.8).
Risk factors for cervical cancer in univariate and logistic regression analyses
Of women aged ≤ 40 years, those with cervical cancer were similar to controls with respect to economic status and number of lifetime sexual partners. Table shows the risk factors for cervical cancer in separate logistic regression analyses by age group. The risk factors associated with having cervical cancer were having high-risk HPV infection and parity > 2, both in women aged ≤ 40 years and in older women. However, HIV infection was associated with cervical cancer only in younger women (OR 4.5; 95% CI 1.5-13.6). There was no confounding of the relationship by age, socioeconomic status, or lifetime number of sexual partners. However, a potential interaction between high-risk HPV and HIV infection could not be assessed.
Risk factors for cervical cancer by age group in logistic regression analysis
Table shows the characteristics of HIV-positive and HIV-negative women with cervical cancer. All HIV-positive case-participants had evidence of high-risk HPV infection. Among HIV-positive women, cervical cancer was detected at a younger age (median age 39 years (range 24-64)) than among HIV-negative women (median age 47 years (range 28-69)). Nevertheless, a similar percentage (50% each) of HIV-positive and HIV-negative women had advanced stage cervical cancer (Clinical Stage III or IV) at diagnosis. Only 18% of HIV-positive women with cervical cancer had CD4 counts below 200 cells/μl.
Characteristics of HIV-positive and HIV-negative women with cervical cancer
Although HIV-positive women with cervical cancer had lower median CD4 counts than did HIV-positive controls, the difference was not statistically significant (376 cells/μl vs. 572 cells/μl: p = 0.6).