Objective: To assess the reproducibility of and factors associated wit
h HIV detection in cervicovaginal secretions (CVS). Design: Longitudin
al study oi 43 HIV-1-infected pregnant women in Paris. Methods: HIV DN
A was detected in peripheral blood mononuclear cells (PBMC) by Amplico
r and gag nested polymerase chain reaction (PCR) assays. The HIV genot
ype was determined by heteroduplex mobility assay. Amplicor and gag ne
sted PCR assays were performed on serial CVS samples for HIV DNA detec
tion, and the HIV Monitor test was used for HIV RNA detection in plasm
a and CVS. Results: A total of 144 CVS samples were collected from the
women included in the study. HIV-1 DNA was detected in 36 (25%) of th
e 144 samples, from 16 (37.2%) of the 43 women. Results of HIV-1 DNA d
etection were concordant in the first two samples in 27 (84.4%) of the
32 women with at least two CVS samples. The last CVS sample collected
in each woman was HIV-1 DNA-positive in 13 (30.2%) of the 43 women. T
hree factors were found to be independently associated with HIV-1 DNA
detection in CVS: HIV-1 subtype B, absence of zidovudine therapy, and
microbial cervicovaginal infection. HIV RNA was detected in CVS from 1
0 (23.3%) out of 43 women and correlated with DNA detection in the sam
e sample and HIV RNA detection in plasma. Conclusions: DNA and RNA PCR
can be used to detect HIV in cells and supernatants of CVS. These tec
hniques may be useful in cohort studies to investigate HIV transmissio
n and to evaluate the efficacy of antiretroviral drugs to reduce HIV e
xcretion.