405. Searching for the Path of Least Resistance: Characterizing Longitudinal Changes in the Gut Resistomes of Healthy Infants and NICU Infants

TL;DR

This study tracks antibiotic resistance genes in infants' guts over three years, comparing those with NICU exposure and antibiotic use to healthy infants.

Contribution

The study reveals longitudinal differences in gut resistomes of infants with NICU exposure and antibiotic use versus healthy infants.

Findings

Infants with NICU exposure had higher vanB resistance gene abundance regardless of antibiotic use.

Unexpectedly, some antibiotic resistance genes were more abundant in healthy infants without NICU or antibiotic exposure at age three.

Resistome diversity significantly differed between the three study groups over time.

Abstract

214,000 infants die annually from antimicrobial resistance worldwide. Clinically-important antibiotic resistance genes (CI-ARG) drive these poor outcomes and impact antibiotic selection. The infant gut harbors many CI-ARG, and the pool of all gut resistance genes comprises the gut resistome. While antibiotics have been shown to increase CI-ARG abundance in older patients, longitudinal CI-ARG development of infants with neonatal intensive care unit (NICU) exposure is not well characterized.Patient DemographicsThe three study groups are quite different from one another in terms of vaginal deliveries, gestational age, and hospital length of stay. IQR = interquartile range.Principal Coordinates Analysis Plot of Resistome Beta Diversity with Bray-Curtis Dissimilarity DistancesBeta diversity of ARG was significantly different between the three study groups over time, even after adjustment for inter-personal variation (p = 0.003). The resistance genes that drive these differences are noted. Older postmenstrual age is correlated with movement of sample resistome composition in the direction of the red arrow.

Patient Demographics

The three study groups are quite different from one another in terms of vaginal deliveries, gestational age, and hospital length of stay. IQR = interquartile range.

Principal Coordinates Analysis Plot of Resistome Beta Diversity with Bray-Curtis Dissimilarity Distances

Beta diversity of ARG was significantly different between the three study groups over time, even after adjustment for inter-personal variation (p = 0.003). The resistance genes that drive these differences are noted. Older postmenstrual age is correlated with movement of sample resistome composition in the direction of the red arrow.

We compared CI-ARG longitudinally up to 3 years of age in 3 groups, categorized by initial NICU admission and antibiotic (abx) exposure: +NICU/+abx (74 infants), +NICU/-abx (54 infants), -NICU/-abx (51 infants) (Figure 1). Metagenomic shotgun sequencing was performed on 1141 serial stool samples. To account for inter-group gestational age differences, linear mixed effects models were used to analyze CI-ARG abundance over postmenstrual age. CI-ARG included specific genes like vanB as well as groups of genes conferring resistance to antimicrobials such as carbapenems and cephalosporins. We also collected clinical metadata such as mode of delivery, breastfeeding, and later antibiotic use.

Hypothesis: The +NICU/+abx group will have more abundant CI-ARG longitudinally than the +NICU/-abx or -NICU/-abx groups.Linear Mixed Effects Models of Relative CI-ARG Abundance versus Postmenstrual AgeEach dot in the plots shows relative abundance and postmenstrual age of each sample for 3 representative CI-ARG: vanB, carbapenems, and cephalosporins. Linear mixed effects models were then applied to assess trends in the data.Heat Map of vanB Abundance Over TimeA plot of each cohort is shown; each horizontal line represents one patient, with the patient’s stool samples represented by diamond shapes on that line. Abundance of vanB for each sample is shown on a color spectrum, with dark blue indicating the lowest abundance and bright yellow indicating the highest abundance. Antibiotic courses are shown with red lines (for post-discharge samples, antibiotics were given at any point in the prior 3 months). Samples with no vanB detected are shown in gray. The vertical dotted blue line represents 40 weeks postmenstrual age for reference.

Linear Mixed Effects Models of Relative CI-ARG Abundance versus Postmenstrual Age

Each dot in the plots shows relative abundance and postmenstrual age of each sample for 3 representative CI-ARG: vanB, carbapenems, and cephalosporins. Linear mixed effects models were then applied to assess trends in the data.

Heat Map of vanB Abundance Over Time

A plot of each cohort is shown; each horizontal line represents one patient, with the patient’s stool samples represented by diamond shapes on that line. Abundance of vanB for each sample is shown on a color spectrum, with dark blue indicating the lowest abundance and bright yellow indicating the highest abundance. Antibiotic courses are shown with red lines (for post-discharge samples, antibiotics were given at any point in the prior 3 months). Samples with no vanB detected are shown in gray. The vertical dotted blue line represents 40 weeks postmenstrual age for reference.

The 3 study groups had significantly different resistomes longitudinally (Figure 2, p = 0.003). Some CI-ARG such as vanB were more abundant in children with any NICU stay, regardless of antibiotic use (Figure 3a, p = 0.0003). Unexpectedly, most CI-ARG were more abundant in the -NICU/-abx group at 3 years, including genes associated with resistance to carbapenems and cephalosporins (Figure 3b, p = 0.019; Figure 3c, p = 0.008, respectively). To improve visualization of longitudinal CI-ARG abundance at a patient level, we piloted a novel heat map, with an example of the map for vanB shown in Figure 4.

This study showed long-term CI-ARG differences among children with and without initial NICU or antibiotic exposure. While vanB abundance increased in children exposed to antibiotics in the NICU, several CI-ARG were more abundant in the healthy infants. Next steps include analyzing taxonomic and clinical data to clarify the basis of these counterintuitive results.

Rana F. Hamdy, MD, MPH, MSCE, FPIDS, StrepApp: "StrepApp"; patent application pending

Figures