The effect of sexually transmitted co-infections on HIV viral load amongst individuals on antiretroviral therapy: a systematic review and meta-analysis

Abstract

Background: Antiretroviral therapy (ART) markedly reduces HIV transmission, and testing and treatment programs have been advocated as a method for decreasing transmission at the population level. Little is known, however, about the extent to which sexually transmitted infections (STIs), which increase the HIV infectiousness of untreated individuals, may decrease the effectiveness of treatment as prevention.

Methods: We searched major bibliographic databases to August 12(th), 2014 and identified studies reporting differences in HIV transmission rate or in viral load between individuals on ART who either were or were not co-infected with another STI. We used hierarchical Bayesian models to estimate viral load differences between individuals with and without STI co-infections.

Results: The search strategy retrieved 1630 unique citations of which 14 studies (reporting on 4607 HIV viral load measurements from 2835 unique individuals) met the inclusion criteria. We did not find any suitable studies that estimated transmission rates directly in both groups. Our meta-analysis of HIV viral load measurements among treated individuals did not find a statistically significant effect of STI co-infection; viral loads were, on average, 0.11 log10 (95% CI -0.62 to 0.83) higher among co-infected versus non-co-infected individuals.

Conclusions: Direct evidence about the effects of STI co-infection on transmission from individuals on ART is very limited. Available data suggests that the average effect of STI co-infection on HIV viral load in individuals on ART is less than 1 log10 difference, and thus unlikely to decrease the effectiveness of treatment as prevention. However, there is not enough data to rule out the possibility that particular STIs pose a larger threat.

Fig. 1

Flow chart of the selection process

Fig. 2

Posterior means and 95 % credible intervals of the effect size for each study included in the meta-analysis. The effect size is expressed as the difference of HIV viral load (log10) between an individual HIV positive, co-infected with any other STI and an individual only infected with HIV. The black square represents the posterior mean, with its area proportional to study sample size. The red diamond and arrows reflect estimates of the pooled effect (i.e., across all studies, STIs and anatomical sites)

Fig. 3

Forest plot for a given STI/anatomical site pair. Posterior means and 95 % credible intervals of the effect size for each study included in the meta-analysis. The effect size is expressed as the difference of HIV viral load (log10) between an individual HIV positive, co-infected with an STI and an individual only infected with HIV. The black square represents the mean of the distribution, its area is proportional to the number of data points associated with the specific STI/anatomical site pair; segments represent the 95 % credible intervals (CI). The red diamond shows the mean of the pooled effect (across all studies, STIs and anatomical sites) and the segment its 95 % CI. Bv: Bacterial vaginosis; Ct: Chlamydia trachomatis; Cv: Candidal vaginitis; Bacterial vaginosis; HPV: human papillomavirus; HSV: human simplex virus type 2; Ng: Neisseria gonorrhoeae; Tp: Treponema pallidum; Tv: Trichomonas vaginalis; Ur: urethritis

Fig. 4

Funnel plot. The horizontal axis represents the mean posterior effect size for each study (the log10 HIV viral load between HIV positive individuals with and without STI co-infection), the vertical axis is the effect size precision (inverse of the standard deviation) of the associated study. Each point represents a study. The solid vertical line represents the mean effect size and the solid curved lines represent its 95 % CI. The dashed vertical line represents a null effect size (no HIV viral load difference)