Impact of Antibiotic Use for Febrile Neutropenia on Gut Microbiome Phyla in Children with Leukemia and Lymphoma

Introduction: Approximately 40% of children with leukemia/lymphoma experience one or more episodes of febrile neutropenia. These episodes are treated with empiric intravenous antibiotics that include broad spectrum β-lactams and vancomycin. Although studies suggest antibiotics change the gut microbiome, the effect is unclear. This study reports the impact of repeated courses and duration of antibiotic use on microbiome.

Methods: We examined 39 pediatric patients diagnosed and treated at the IWK Health Centre, Nova Scotia Canada for acute lymphoblastic leukemia (28), acute myeloid leukemia (5), non-Hodgkin lymphoma (4) or Hodgkin lymphoma (2) after informed consent. DNA was extracted from 127 stool samples from timepoints before and during treatment were used for whole shotgun sequencing (Nextera XT). Sequences were cleaned of low quality, and human, and PhiX174 contaminate sequences (kneaddata pipeline). MetaPhlAn3 was used to profile the taxonomic composition of the microbial community. Samples were grouped by number of courses, none, 1-2, or ≥3, and by duration of courses, none, short (≤10 days), long/mixed (>10 days or mix of long and short) of β-lactams, vancomycin, or general antibiotic use. The abundance of phyla within these groups were compared using the Kruskal Wallis test (α<0.05) and post hoc Dunn test for pairwise comparison, corrected for multiple tests. This study was approved by the IWK ethics board.

Results: The 39 patients include 22 males and 17 females, with 29 patients over the age of 3 and 10 patients under 3. Bacteroidetes phylum abundant in the gut were decreased when exposed to repeated courses of all antibiotic groups with none vs ≥3 courses (β-lactam p=0.0012; vancomycin p=0.0018; multiple antibiotic p=0.0021) and 1-2 courses vs ≥3 courses (β-lactam p=3.1E-6; vancomycin p=0.024; multiple antibiotic p=1.6E-8). We also noted an increase in Firmicutes in the children who received 1-2 courses vs ≥3 courses of β-lactam antibiotics (p=0.034). The phylum Proteobacteria was increased across those children who received 1-2 courses vs ≥3 courses (β-lactam p=8.1E-6; vancomycin p=0.0018; multiple antibiotic p=1.6E-5) and also increased in none vs ≥3 courses for vancomycin (p=3.8E-4).

We noted differences in abundance based on duration of antibiotic treatment. For Bacteroidetes the abundance is higher in children who receive short course antibiotics vs those that receive longer course (>10 days) for β-lactam and multiple antibiotic (p=1.3E-6 and p=4.4E-7 respectively) and lower in children who receive short course vancomycin vs longer course (p=0.024). The Proteobacteria in the samples of children treated with β-lactam and multiple antibiotics showed higher abundance in those who received longer duration antibiotics (p=2.7E-5 and p=8.7E-6 respectively) whereas the children who received short courses of vancomycin showed higher counts (p=0.0018). No difference in Firmicute abundance was detected.

Discussion: Concordant with the spectrum of β-lactam antibiotics, we saw a decrease in abundance of Bacteroidetes a predominately gram-negative gut bacterium. In contrast the Proteobacteria, a gram-negative phylum, was increased in children exposed to repeated courses of antibiotics. When we consider duration of antibiotic treatment, we see that the Bacteroidetes have higher prevalence in children who do not receive antibiotics or receive shorter course antibiotics indicating that this phylum is controlled by antibiotic treatment. In contrast, the Proteobacteria are more prevalent in children who receive longer duration of β-lactam. One limitation of this study was non-uniformity of stool sample collection due to logistical reasons in pediatric population.

Conclusion: Repeated courses of antibiotics administered empirically to pediatric cancer patients with leukemia/lymphoma increases the relative abundance of certain gram-negative bacteria in the microbiome. Longer courses of the β-lactams and multi-antibiotic samples result in higher relative abundance of Proteobacteria potentially increasing infections in immunocompromised children. Further studies to determine optimum antibiotic duration and therapy are needed.

No relevant conflicts of interest to declare.