Oral and Stool Microbiome Coalescence and Its Association With Antibiotic Exposure in Acute Leukemia Patients

Failure to maintain segregation of oral and gut microbial communities has been linked to several diseases. We sought to characterize oral-fecal microbiome community coalescence, ectopic extension of oral bacteria, clinical variables contributing to this phenomenon, and associated infectious consequences by analyzing the 16S rRNA V4 sequences of longitudinal fecal (n=551) and oral (n=737) samples from 97 patients with acute myeloid leukemia (AML) receiving induction chemotherapy (IC). Clustering observed in permutation based multivariate analysis of variance (PERMANOVA) of Bray-Curtis dissimilarity and PCoA plot of UniFrac distances between intra-patient longitudinal oral-stool sample pairs suggested potential oral-stool microbial community coalescence. Bray-Curtis dissimilarities and UniFrac distances were used to create an objective definition of microbial community coalescence. We determined that only 23 of the 92 patients exhibited oral-stool community coalescence. This was validated through a linear mixed model which determined that patients who experienced coalescence had an increased proportion of shared to unique OTUs between their oral-stool sample pairs over time compared to non-coalesced patients. Evaluation of longitudinal microbial characteristics revealed that patients who experienced coalescence had increased stool abundance of Streptococcus and Stenotrophomonas compared to non-coalesced patients. When treated as a time-varying covariate, each additional day of linezolid (HR 1.15, 95% CI 1.06 - 1.24, P <0.001), meropenem (HR 1.13, 95% CI 1.05 - 1.21, P = 0.001), metronidazole (HR 1.13, 95% CI 1.05 - 1.21, P = 0.001), and cefepime (HR 1.10, 95% CI 1.01 - 1.18, P = 0.021) increased the hazard of oral-stool microbial community coalescence. Levofloxacin receipt was associated with a lower risk of microbiome community coalescence (HR 0.75, 95% CI 0.61 - 0.93, P = 0.009). By the time of neutrophil recovery, the relative abundance of Bacteroidia (P<0.001), Fusobacteria (P=0.012), and Clostridia (P=0.013) in the stool were significantly lower in patients with oral-gut community coalescence. Exhibiting oral-stool community coalescence was associated with the occurrence of infections prior to neutrophil recovery (P=0.002), as well as infections during the 90 days post neutrophil recovery (P=0.027). This work elucidates specific antimicrobial effects on microbial ecology and furthers the understanding of oral/intestinal microbial biogeography and its implications for adverse clinical outcomes.