Independent Effects of HIV and Antiretroviral Therapy on the Oral Microbiome Identified by Multivariate Analyses

Abstract

The oral microbiome is an important predictor of health and disease. We recently reported significant yet modest effects of HIV under highly active antiretroviral therapy (ART) on the oral microbiome (bacterial and fungal) in a large cohort of HIV-positive (HIV⁺) and matched HIV-negative (HIV^-) individuals. As it was unclear whether ART added to or masked further effects of HIV on the oral microbiome, the present study aimed to analyze the effects of HIV and ART independently, which also included HIV^- subjects on preexposure prophylaxis (PrEP) therapy. Cross-sectional analyses of the effect of HIV devoid of ART (HIV⁺ ART^- versus matched HIV^- subjects) showed a significant effect on both the bacteriome and mycobiome (P < 0.024) after controlling for other clinical variables (permutational multivariate analysis of variance [PERMANOVA] of Bray-Curtis dissimilarity). Cross-sectional analyses evaluating the effects of ART (HIV⁺ ART⁺ versus HIV⁺ ART^- subjects) revealed a significant effect on the mycobiome (P < 0.007) but not the bacteriome. In parallel longitudinal analyses, ART (before versus after the initiation of ART) had a significant effect on the bacteriome, but not the mycobiome, of HIV⁺ and HIV^- PrEP subjects (P < 0.005 and P < 0.016, respectively). These analyses also revealed significant differences in the oral microbiome and several clinical variables between HIV^- PrEP subjects (pre-PrEP) and the HIV-matched HIV^- group (P < 0.001). At the species level, a small number of differences in both bacterial and fungal taxa were identified within the effects of HIV and/or ART. We conclude that the effects of HIV and ART on the oral microbiome are similar to those of the clinical variables but collectively are modest overall. IMPORTANCE The oral microbiome can be an important predictor of health and disease. For persons living with HIV (PLWH), HIV and highly active antiretroviral therapy (ART) may have a significant influence on their oral microbiome. We previously reported a significant effect of HIV with ART on both the bacteriome and mycobiome. It was unclear whether ART added to or masked further effects of HIV on the oral microbiome. Hence, it was important to evaluate the effects of HIV and ART independently. For this, multivariate cross-sectional and longitudinal oral microbiome analyses (bacteriome and mycobiome) were conducted within the cohort, including HIV⁺ ART⁺ subjects and HIV⁺ and HIV^- (preexposure prophylaxis [PrEP]) subjects before and after the initiation of ART. While we report independent significant effects of HIV and ART on the oral microbiome, we conclude that their influence is similar to that of the clinical variables but collectively modest overall.

Keywords: ART; HIV; bacteriome; mycobiome; oral microbiome.

FIG 1

Multivariate distance-based analysis of the effects of HIV on the oral bacterial community composition. Nonmetric multidimensional scaling ordination of Bray-Curtis dissimilarities between oral bacterial (A) and fungal (B) communities is shown for HIV⁻ subjects (high risk [HR]), HIV⁺ subjects prior to the initiation of ART, and HIV⁻ pre-PrEP subjects. The spiders connect the sample points to the centroid of each group. The accompanying table indicates PERMANOVA of the significant differences between HIV⁻ and HIV⁺ pre-ART groups in multivariate analyses. *, P < 0.05; **, P < 0.01; ***, P < 0.001.

FIG 2

Cross-sectional analysis of the effect of ART on the microbiome in HIV⁺ individuals. HIV⁺ subjects enrolled under ART were compared to HIV⁺ subjects prior to the initiation of ART. Nonmetric multidimensional scaling ordination of Bray-Curtis dissimilarities between oral bacterial (A) and fungal (B) communities is shown. The spiders connect the sample points to the centroid of each group. The accompanying table indicates PERMANOVA of the significant differences between groups in multivariate analyses. HAART, highly active antiretroviral therapy; *, P < 0.05, **, P < 0.01, ***, P < 0.001; NS, not significant.

FIG 3

Comparison of ordinal clinical variables between HIV⁻ high-risk subjects and HIV⁻ pre-PrEP subjects. Box-and-whisker plots are shown for the indicated variables, with the boxes showing the 25th, 50th, and 75th quartiles; the whiskers showing 1.5 times the interquartile difference or the limit of the data if lower; and outliers shown as points. Significance is indicated with asterisks (**, P < 0.01; ***, P < 0.001).

FIG 4

Longitudinal analysis of the effect of ART on the oral bacteriome of HIV⁺ individuals. ART was initiated following HIV diagnosis. Samples were taken preinitiation and 1 to 6 months and 6 to 18 months post-ART. Nonmetric multidimensional scaling ordination of the Bray-Curtis dissimilarity between oral bacterial communities is shown. The ellipses represent the 95% confidence region for the centroid of the group. Line segments connect the three NMDS ordination points from each subject that were collected over time. The accompanying table shows the PERMANOVA comparisons overall and post hoc pairwise comparisons between time points.

FIG 5

Longitudinal analysis of the effect of ART on the oral bacteriome in HIV⁻ PrEP subjects. ART was initiated upon request for HIV-negative subjects under the PrEP program. Samples were taken prior to the initiation of ART (Pre-PrEP) and 1 to 6 months and 6 to 18 months after the initiation of ART. Nonmetric multidimensional scaling ordination of Bray-Curtis dissimilarities between oral bacterial communities is shown. Ellipses represent the 95% confidence regions of the group centroids. Line segments connect the three NMDS ordination points from each subject that were collected over time. The accompanying table shows the PERMANOVA comparisons between all time points.

FIG 6

Longitudinal analysis of the effect of ART on the oral mycobiome in HIV⁺ and HIV⁻ PrEP individuals. ART was initiated following HIV diagnosis for HIV⁺ subjects and upon request for HIV⁻ subjects under the PrEP program. Samples were taken prior to the initiation of ART/tenofovir-emtricitabine (prePrEP) and 1 to 6 months and 6 to 18 months after the initiation of therapy. Nonmetric multidimensional scaling ordination of Bray-Curtis dissimilarities between oral fungal communities is shown. The dissimilarities for fungal communities were calculated using qPCR-adjusted abundances. The ellipses represent the 95% confidence intervals for the position of the group centroid. Line segments connect the three NMDS ordination points from each subject that were collected over time. The overall PERMANOVA P values for each longitudinal experiment are shown.