Background: Previously, we demonstrated that plasma and vaginal viral loads are strongly correlated, and release of pro-inflammatory cytokines associated with cervical ulceration markedly enhances vaginal HIV shedding. However, vaginal viral load can increase and decrease significantly after controlling for plasma viral load and genital inflammation. We designed this study to determine if subclinical inflammation, defined as elevated IL-1β or TNF-α level, in the female genital tract is associated with vaginal viral load independent of plasma viral load.

Methods: During 1996–2000, blood and vaginal lavages were collected at 987 visits of 135 HIV-infected women enrolled in the Emory Vaginal Ecology Study. At each visit, women were tested for gonorrhea, HSV, CMV, chlamydia, bacterial vaginosis, and candida vaginitis. For this study, 60 lavage samples were selected based on a linear regression analysis of plasma and vaginal viral loads, whereby 1) 20 samples had viral loads that were strongly correlated with that in plasma; 2) 20 samples had viral loads that were below the lower limit of the 95% confidence interval around the linear regression line; and 3) 20 samples had viral loads that were above the upper limit of the 95% confidence interval. Each sample was analyzed for IL-1β and TNF-α by commercially available ELISAs.

Results: At these visits, genital tract infections were infrequent and equally distributed among the 3 groups of samples. The following table shows the median values of virus loads in plasma (copies/mL) and lavage (total copies) and IL-1β and TNF-α (pg/mL) in lavage for the 3 groups of samples.

Sample group	Vaginal virus load	Plasma virus load	IL-1β level	TNF-α level
Below 95% CI	< 100	200,000	7.49	4.00
Correlated	4,200	60,000	79	3.14
Above 95% CI	470,000	14,000	210	6.98

 

The levels of IL-1β and TNF-α were associated with the vaginal viral loads in the groups of samples (p < 0.01 and p = 0.01, respectively, using analysis of variance of ranks).

Conclusions: Commonly diagnosed genital tract infections did not explain higher or lower vaginal viral levels than would be expected based on plasma viral loads. In contrast, IL-1β and TNF-α levels in vaginal secretions were significantly associated with the vaginal viral loads independent of plasma viral loads. These results suggest that subclinical inflammation, as measured by increased cytokine levels, is an important determinant of vaginal viral shedding.