Effect of long-term azithromycin treatment on gut microbial diversity in children and adolescents with HIV-associated chronic lung disease

Abstract

Background: HIV-associated chronic lung disease (HCLD) is common in children and adolescents growing up with HIV. The use of azithromycin (AZM) reduces the rate of acute respiratory exacerbations in this population, however, impact of this treatment on the gut microbiota and associations with blood-derived inflammatory markers have not been studied.

Methods: Children and adolescents with HCLD in Harare, Zimbabwe and Blantyre, Malawi were recruited in a double-blind, placebo-controlled trial of once-weekly AZM or placebo for 48 weeks (BREATHE trial, NCT02426112). Rectal swabs were collected at inclusion (N = 346), 48 weeks (treatment end, N = 313), and 72 weeks (six months after treatment cessation, N = 244). The bacterial composition of fecal swabs was determined using 16S rRNA gene sequencing. Plasma biomarkers at inclusion and 48 weeks were measured using Luminex multiplex bead assay.

Findings: At 48 weeks, bacterial α-diversity was significantly lower in the AZM group, with 27 bacterial genera showing differential abundance between the study groups. The placebo group exhibited higher interconnectivity between bacterial genera at 48 weeks compared to the AZM group. Correlations between the top seven differentially abundant genera and biomarkers observed at inclusion were no longer significant at 48 weeks in both groups. Depletion of Campylobacter persisted for six months after cessation of AZM treatment.

Interpretation: Long-term AZM treatment in HCLD patients affects their gut bacterial composition at least 6 months after its cessation. The consequences of reduced bacterial diversity, such as altered interaction with the immune system and risk of resistance, need further investigation to understand how to optimise gut health during long-term antibiotic treatments.

Funding: The study was funded by the Norwegian Research Council and Helse Nord (HNF 1387-17).

Keywords: Azithromycin; Chronic lung disease; Gastrointestinal microbiome; HIV; Southern Africa.

Fig. 1

Study flow chart. Abbreviations: ART, antiretroviral therapy; LTFU, lost to follow-up.

Fig. 2

Associations between systemic biomarkers and various covariates at (a) baseline and (b) after 48 weeks, utilising a linear model (LM). In panel b, all genera have interacted with the antibiotic in the LM model.

Fig. 3

Principal Coordinate Analysis (PCoA) describing the microbiota composition in the placebo and azithromycin (AZM) groups at (a) inclusion (baseline), (b) 12 months (48 weeks), and (c) 18 months (72 weeks) using Bray–Curtis distance. The p-value was calculated by adonis2 with 999 permutations.

Fig. 4

Spearman correlation matrix plot of relative abundance at the genera level for (a) virally suppressed group (viral load < 1000 copies/ml) at baseline, (b) high viral load group (viral load > 1000 copies/ml) at baseline, (c) Zimbabwean cohort at baseline, (d) Malawian cohort at baseline, (e) placebo group after 48 weeks, and (f) azithromycin-treated group after 48 weeks. The colour scale represents the Spearman correlation coefficient, and the square size indicates the correlation strength, meaning the absolute value of Spearman coefficient, reflecting the magnitude of the association. All results shown were statistically significant (adjusted p < 0.05).