Impact of antibiotics on gut microbiome composition and resistome in the first years of life in low- to middle-income countries: A systematic review

Introduction

Bacterial antimicrobial resistance (AMR) is an already present and growing threat to global health and economies [1,2]. Low- to middle-income countries (LMICs) are disproportionately burdened by AMR [3], with some of the highest attributable mortality rates [4] and the least resources to control AMR, resulting from limited laboratory capacity, surveillance systems, and health care infrastructure. The World Health Organization (WHO) has identified AMR as a critical threat to already strained health systems, food security, and development in LMIC regions [5].

Human antibiotic consumption is a key driver of AMR, which arises when microbes become resistant to antimicrobials [6–8]. Antibiotic consumption is rising and distributed inequitably across geographic scales [9]. Children in LMICs have high frequencies of infectious disease episodes and a great need for antibiotic access, yet are also the most vulnerable to AMR and the risks of appropriate and inappropriate antibiotic consumption [10,11]. There is insufficient knowledge on how antibiotic consumption may place children in LMICs at risk for adverse health impacts and AMR.

Antibiotics, when used appropriately, reduce all-cause mortality and are lifesaving for key infectious diseases among children in LMICs [12,13]. Yet, they are frequently used inappropriately [14] and have multiple off-target effects in human hosts, with specific, under-characterized risks for children in LMIC settings. Antibiotic exposures significantly alter the short- and long-term maturation of the intestinal microbiome in children [15–17]. How these alterations differ by antibiotic class and translate into clinical benefits or risks of disease in different populations is poorly understood. Antibiotic-induced alterations in the microbiota are known to reduce colonization resistance, increasing susceptibility to specific opportunistic (entero-) pathogens [18,19], such as Clostridium difficile and Salmonella [20]. Conversely, in LMICs, mass administration of azithromycin has been shown to significantly reduce childhood mortality, possibly through a reduction in enteropathogenic burden [12,13,21,22]. Antibiotic-driven alterations of early microbiome colonization in high-income settings have been associated with risks of multiple childhood diseases later in life, including obesity [23–27], asthma [28], and diabetes [29–31]. Studies in LMIC children have primarily demonstrated antibiotics’ growth-promoting capacity [32], with lack of reports on the long-term health effects of antibiotic microbiome modulation in these regions.

Foremost among the risks of antibiotics in LMICs is their potential to promote AMR through mutation and selection of bacteria with resistance to antibiotics while increasing the abundance of bacterial AMR genes (ARGs) (termed the resistome) [33]. Mobile genetic elements carrying ARGs can be spread horizontally to bystander bacteria which then acquire the ability to resist antibiotic treatment [34,35]. Antibiotics likely differ in their ability to drive the selection, spread, and persistence of AMR genes in the microbiome and in the degree to which they place children in LMIC communities at risk of infection from pathogens with AMR. Baseline differences in gut microbiome diversity and composition by geographical region and ethnicity may also modulate antibiotic effects [36]. There is paucity of information on the effects of antibiotic use on either the developing child microbiome or resistance gene carriage in LMIC children where antibiotic misuse is on the rise. This systematic review aims to address this knowledge gap by reviewing the literature on the impact of antibiotic exposures on both the faecal bacterial microbiome and resistome composition in children below the age of 2 from LMICs.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) checklist and guidelines were used in the design of the review and the protocol was registered with the International prospective register for systematic reviews (PROSPERO), with ID CRD42022307691.

Results

The literature search identified 4,369 articles and these were exported to Rayyan. Upon screening of titles and abstracts, 1,621 duplicate records were removed resulting in 2,748 articles, of these 2,666 did not meet the inclusion criteria and were removed (Fig 1). Full-text screening of the remaining 92 studies resulted in exclusion of 82 studies due to 21 being performed in HICs, 3 not mentioning the type of antibiotic the participants took, 27 involving animal experiments, 19 assessing the wrong outcomes, and 12 having the incorrect population group. This resulted in the inclusion of 10 studies. Included studies were assessed for bias, illustrated in Fig 2. All studies had a low risk of bias with the exception of D’Souza and colleagues [15] that had a high risk of bias due to randomization, as the caregivers were not blinded to treatment allocation. Due to the low number of the studies, risk of bias assessment results were not used to exclude studies.

A secondary analysis identified studies evaluating antibiotic impacts on the microbiome and resistome in HIC. The screening and exclusions identified 2 studies eligible for inclusion (see S1 Fig). Significant heterogeneity in infant health status, study design, antibiotic indication, and antibiotic class between studies identified in HIC versus LMIC precluded a comparison between HIC and LMIC studies. Data on these studies and their results are described in S1 Text, S2 Table, and S2 and S3 Figs.

Discussion

To our knowledge, this systematic review, which synthesizes the impacts of antibiotic use on both gut microbiome composition and resistance genes in infants, is the first to focus on infants in LMIC settings. The review found that antibiotics variably impacted microbiome diversity and composition, with effects likely mediated by antibiotic class, duration of administration, and follow-up time. Multiple studies included in the review also showed that antibiotic administration, namely AZI, can shift taxonomic composition of the microbiome, particularly by reducing enteropathogen burden. Finally, antibiotic administration expands both the total and antibiotic class-specific resistome. Even short antibiotic exposures are able to increase AMR genes for months. The magnitude and persistence of these change is likely dictated by duration of antibiotic exposure and class.

The range of antibiotics in the included studies was limited, assessing only AZI, CTX, and AMX (1 study), with AZI being the most widely reported antibiotic. While AZI is not the most commonly used antibiotic in the treatment of childhood infectious syndromes, its capacity to reduce childhood mortality through prophylactic mass administration in high-risk pediatric populations [45], does make it relevant for study in LMIC. No studies conducted in LMICs were found evaluating some of the most commonly used antibiotics of childhood with capacity for anaerobic depletion such as clindamycin and metronidazole.

Antibiotic duration was also found to be a key determinant of the impact on microbiome structure. Even short courses of antibiotics like AZI were able to immediately reduce microbiota diversity [51] with persistent reductions in richness of the microbiota for up to 6 months following administration [44]. A short course of CTX diminished alpha diversity indexes [48], whereas chronic use increased diversity indexes over time [15]. AZI demonstrated a consistent and persistent alteration in taxonomic composition in line with experimental data in adults [53,54].

The review also showed that antibiotic administration can alter the taxonomic composition of the microbiome in LMIC children. A key question is whether antibiotics reduce overall pathogen burden and risk of infections or render children more susceptible to infections—possibly through depletion of the commensal microbiota and loss of colonization resistance. Mouse studies have suggested that early life exposures to antibiotics increase susceptibility to bacterial enteropathogens in adulthood [55]. In the few pediatric studies where enteropathogens were considered, antibiotics tended to reduce enteropathogenic burden in the microbiome. Targeted enteropathoge AZI administration reduced Campylobacter spp. relative abundance in Niger and Burkina Faso [45,47] and reduced Proteobacteria with pathogenic potential in India [51]. One study showed an increase in the relative abundance of 1 putative enteropathogen (E. albertii) in Malawi [49]. These results suggest that 16S and metagenomic analyses may be insufficient to identify clinically-relevant enteropathogens, and complementary targeted enteropathogen testing may be required to characterize antibiotic impacts on enteropathogen burden. In contrast, antibiotic studies in adults have shown increases in enteropathogens post-antibiotics, primarily C. difficile [56] and vancomycin-resistant enterococci [57], underscoring the complexity of these positive and negative associations with antimicrobial use across age spectra.

Finally, our systematic review found that antibiotics drove total and antibiotic class-specific increases in AMR gene determinants in all included studies. A second key question is to what degree antibiotic usage places a child at risk for infections with antimicrobial resistant pathogens, with AMR gene carriage being a surrogate endpoint in this pathway. AZI consistently increased macrolide resistance that persisted up to 6 months following administration. Longer biannual use notably also increased resistance determinants to non-macrolide antibiotics. In parallel, CTX increased total resistance gene diversity, sulfonamide and trimethoprim resistance determinates. Children in LMICs are subject to multiple rounds of appropriate and inappropriate antibiotics, and therefore, these studies are likely a gross underestimation of the impact of antibiotics across a child’s early life [11,58].

A remaining question, that this systematic review did not address, is how antibiotic-induced reductions in diversity and disruption of the microbiome may impact the biologic functions of the infant microbiome and how these alterations translate into health and illness outcomes for infants in LMICs. The microbiome plays multiple roles in hosts, including immune maturation, immune regulation, and host metabolism [59]. Antibiotic-induced losses in diversity may result in reduction of organisms that play key roles in these biologic functions, potentially altering infants’ immunity to exogenous pathogens, alongside nutrition and growth outcomes. Mouse studies have shown that antibiotic administration in early life increases subsequent risk of atopic disease—such as asthma and adiposity [60,61]. Similarly, unknown is how alterations in AMR gene abundance translate into a child’s short- and long-term risk of infection from an AMR-bearing pathogen. Although metagenomic sequencing can identify changes in AMR genes, it is difficult to deduce from which bacteria those genes are derived. A key limitation of genomic resistome analyses is therefore bacterial host assignment and whether increases in genomic AMR genes translate to clinically relevant phenotypic resistance. Research is urgently needed to understand how antibiotics may impact the functional microbiome diversity [62], phenotypic resistance in enteropathogens, and risk of disease, in order to better inform antibiotic prescribing practices and health policy around AMR and population-level antibiotic administration.

Limitations of this systematic review were the paucity of studies identified and included, the limited number of countries studies were performed in, and the lack of diversity in antibiotic classes. The available data was dominated by 2 large clinical trials in Africa, MORDOR [63] and ARMCA [64]. Other included studies were highly heterogeneous with differing antibiotic dosing schemes and testing time points even within antibiotic class. Thus, both the generalizability and comparability of results is constrained and precluded pooling data and permitting a meta-analysis. While most individual studies detected significant impacts of antimicrobial usage on microbiome composition and resistome, a few studies did not report any significant treatment effects [15,45,49]. This may have been due to a true lack of effect, timing of sampling, or insufficient sample size. Lastly, few studies included long-term sampling, limiting the understanding of how long antibiotic-induced alterations in the microbiome and resistome persist and the resilience of the pediatric LMIC microbiome following antibiotic challenge.

Notable strength of this review is that it provides a comprehensive overview and needed synthesis of the available data on antibiotic usage and microbiome in pediatric LMIC populations with high incidences of antibiotic use. This review thereby maps the current state of knowledge, highlighting available data alongside persistent research gaps that can guide both policy and research agendas. The included studies clearly show that antibiotic administration results in large-scale alterations of the microbiome and resistome with enough evidence to alert policy makers to the adverse and indirect impacts of (mass) antibiotic administration. In parallel, the review highlights key future research needs, namely moving from how alterations in the microbiome and resistome translate to tangible health benefits or risks for pediatric populations, including risk of bacterial infections with AMR.

Our systematic review also provides a mapping to move this field forward. It can be used to address the urgent need for a harmonized methodology to permit comparison of the impacts of antibiotic administration across antibiotic class, geographic setting, and time. Future studies should focus on pediatric populations in LMIC and evaluate commonly used pediatric antibiotics, incorporating sampling at the start, during, and following antibiotic administration, including long-term follow-up. Studies should be adequately powered and strive to employ metagenomic sequencing techniques with sufficient sequencing depth to describe genus and species level antibiotic perturbations alongside alterations in the resistome and functional metagenome. Harmonized annotation pipelines and uniform outcome measures will permit comparability across studies. When possible, sequencing should be combined with bacterial culture to permit host annotation for key resistance genes and host enteropathogens. Open-source data repositories will permit researchers to then re-analyze data across studies, permitting policy makers and clinicians to draw actionable conclusions from this research.

In summary, antibiotics in children in LMIC disrupt the diversity and taxonomy of the gut microbiome and drive long-term expansion of ARG determinants. There is a paucity of research describing the impacts of class-specific antibiotics on microbiome diversity and ARG development in infants in LMIC. Given the vulnerability of children in LMIC to infections, the frequency of antibiotic consumption in their early lives, and the growing global threat of infections with AMR, harmonized research initiatives are urgently needed to understand how antibiotic perturbations on the microbiome translate into health benefits or risk of disease.

Supporting information

Data Availability

All relevant data are within the manuscript and its Supporting Information files.

Funding Statement

This research was supported by The Netherlands Organization for Health Research and Development (ZonMw, zonmwl.nl) VENI 09150161810022 The Netherlands Organization for Health Research and Development (ZonMw, zonmwl.nl) 10430022010019 to VCH, Wellcome Trust (wellcome.org) 219775/Z/19/Z to VCH, Health Holland TKI (https://www.health-holland.com) GLORIA fund to VCH. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.