Therapeutic beta-lactam dosages and broad-spectrum antibiotics are associated with reductions in microbial richness and diversity in persons with cystic fibrosis

Abstract

Persons with cystic fibrosis (PwCF) suffer from pulmonary exacerbations (PEx) related in part to lung infection. While higher microbial diversity is associated with higher lung function, the data on the impact of short-term antibiotics on changes in microbial diversity is conflicting. Further, Prevotella secretes beta-lactamases, which may influence recovery of lung function. We hypothesize that sub-therapeutic and broad spectrum antibiotic exposure leads to decreasing microbial diversity. Our secondary aim was to evaluate the concerted association of beta-lactam pharmacokinetics (PK), antibiotic spectrum, microbial diversity, and antibiotic resistance on lung function recovery using a pathway analysis. This was a retrospective observational study of persons with CF treated with IV antibiotics for PEx between 2016 and 2020 at Children's National Hospital; respiratory samples and clinical information were collected at hospital admission for PEx (E), end of antibiotic treatment (T), and follow-up (F). Metagenomic sequencing was performed; PathoScope 2.0 and AmrPlusPlus were used for taxonomic assignment of sequences to bacteria and antibiotic resistance genes (ARGs). M/W Pharm was used for PK modeling. Comparison of categorical and continuous variables and pathway analysis were performed in STATA. Twenty-two PwCF experienced 43 PEx. The study cohort had a mean age of 14.6 years. Only 12/43 beta-lactam courses had therapeutic PK, and 18/43 were broad spectrum. A larger decrease in richness between E and T was seen in the therapeutic PK group (sufficient - 20.1 vs. insufficient - 1.59, p = 0.025) and those receiving broad spectrum antibiotics (broad - 14.5 vs. narrow - 2.8, p = 0.030). We did not detect differences in the increase in percent predicted forced expiratory volume in one second (ppFEV1) at end of treatment compared to PEx based on beta-lactam PK (sufficient 13.6% vs. insufficient 15.1%) or antibiotic spectrum (broad 11.5% vs. narrow 16.6%). While both therapeutic beta-lactam PK and broad-spectrum antibiotics decreased richness between PEx and the end of treatment, we did not detect longstanding changes in alpha diversity or an association with superior recovery of lung function compared with subtherapeutic PK and narrow spectrum antimicrobials.

Figure 1

Change in alpha diversity and bacterial load by beta-lactam pharmacokinetic exposure from pulmonary exacerbation to end of antibiotic treatment. Panel (A) Change in Species Observed. Panel (B) Change in Shannon Diversity. Panel (C) Change in Inverse Simpson Index. Panel (D) Log Change in Bacterial Load. PK, pharmacokinetics; PEx, pulmonary exacerbation; E, pulmonary exacerbation; T, end of treatment; ET, change between pulmonary exacerbation and end of treatment. Plus signs represent the mean.

Figure 2

Change in alpha diversity and bacterial load by antibiotic spectrum exposure from pulmonary exacerbation to end of antibiotic treatment. Panel (A) Change in Species Observed. Panel (B) Change in Shannon Diversity. Panel (C) Change in Inverse Simpson Index. Panel (D) Log Change in Bacterial Load. PEx, pulmonary exacerbation; E, pulmonary exacerbation; T, end of treatment; ET, change between pulmonary exacerbation and end of treatment. Plus signs represent the mean.

Figure 3

Changes in beta diversity. Panel (A) Bray–Curtis Non-metric multidimensional scaling (NMDS) Plot, Pharmacokinetic Sufficient. Panel (B) Bray–Curtis NMDS Plot, Antibiotic Spectrum. Panel (C) Morisita–Horn Index, Pharmacokinetic Sufficient. Panel (D) Morisita–Horn Index, Antibiotic Spectrum. NMDS, non-metric multidimensional scaling; PK, pharmacokinetic; E, pulmonary exacerbation; T, end of treatment; F, follow up; ET, change between pulmonary exacerbation and end of treatment; EF, change between pulmonary exacerbation and follow-up. On the NMDS plots, the ellipses represent the t distribution. The permutational multivariate analysis of variance (PERMANOVA) result for beta-lactam PK was p = 0.120 and for antibiotic spectrum was p = 0.133. On the box and whiskers plots, the center horizontal line corresponds to the median and the colored boxes represent the interquartile ranges. Small circles represent individual values that were outliers.

Figure 4

Recovery of pulmonary function. Panel (A) Increase in ppFEV1 from PEx to end of antibiotic treatment by beta-lactam pharmacokinetic exposure. Panel (B) Percent recovery of baseline ppFEV1 at end of treatment compared to the best ppFEV1 in the 6 months prior to PEx by beta-lactam pharmacokinetic exposure. Panel (C) Increase in ppFEV1 from PEx to end of antibiotic treatment by antibiotic spectrum exposure. Panel (D) Percent recovery of baseline ppFEV1 at end of treatment compared to the best ppFEV1 in the 6 months prior to PEx by antibiotic spectrum exposure. ppFEV1, percent predicted forced expiratory volume in one second; PEx, pulmonary exacerbation. Plus signs represent the mean.