Cervical cancer is frequent in Sub-Saharan Africa. Moreover, as the access to antiretroviral treatment improves, it is expected that chronic HIV-related diseases such as cervical cancer will increasingly come to the foreground. In the Democratic Republic of the Congo, little information is available about the frequency of cervical cancer in women with and without HIV. The study aim was to estimate the strength of the association between HIV infection and the presence of (pre)cancerous lesions of the uterine cervix in women from Kinshasa.
We conducted a cross-sectional study and enrolled HIV-positive and HIV-negative women in two HIV screening and treatment centres in Kinshasa in 2006-2007. Cervical smear samples were examined using liquid-based cytology and classified according to the Bethesda 2001 classification of cervical pathology. Women with cytology results indicating low-grade squamous intraepithelial lesions or higher-grade lesions (LSIL+) were considered to have (pre)cancerous lesions. We collected information about sociodemographic and behaviour-related factors that could act as confounders in the relation between HIV infection and LSIL+, i.e. age, marital status, socio-economic status, smoking, age at first sexual intercourse, number of lifetime sex partners, childbirths and habit of vaginal cleansing. Multiple logistic regression was used to check for confounding and effect modification and to calculate an adjusted odds ratio (OR) for the association between HIV and LSIL+. In addition, among HIV-positive women, we described the relation between CD4 cell counts and LSIL+ and used the Mann-Whitney U test to compare groups.
One hundred twenty-eight HIV-positive and 132 HIV-negative women were included. Their mean age was 34 years (standard deviation 10). Five HIV-negative (4%) and 41 HIV-positive women (31%) were diagnosed with LSIL+. The final logistic regression model included five variables: HIV, number of childbirths, marital status, number of lifetime sex partners, and age at first sexual intercourse. HIV was the only variable for which the association with LSIL+ remained significant on multivariable analysis. The adjusted OR for the association between HIV infection and presence of LSIL+ was 7.5 (95% confidence interval 2.6-21.4). CD4 cell counts were available for 87 HIV-infected women: those with LSIL+ had lower CD4 counts (median 162 cells/µl; interquartile range 93-327) than those without LSIL+ (336 cells/µl; 184-475; P=0.002).
Among women of Kinshasa, there is a strong association between HIV infection and the presence of precancerous or cancerous lesions of the uterine cervix.