Whole-Genome Sequencing and Metagenomic Surveillance for Antimicrobial Resistance in Low- and Middle-Income Countries: A Narrative Review

1. Introduction

Antimicrobial resistance (AMR) threatens the effective treatment of bacterial infections and places increasing pressure on health systems. The 2019 global burden analysis estimated that bacterial AMR was directly attributable to approximately 1.27 million deaths and associated with substantially more deaths worldwide, with particularly high burdens in several low-resource regions (Murray et al., 2022). More recent modelling has emphasized that a large share of the preventable burden occurs in low- and middle-income countries (LMICs), where access to effective antibiotics, diagnostics, infection prevention, and reliable surveillance can be uneven (Lewnard et al., 2024; Sulis et al., 2022).

AMR in LMICs is shaped by interacting clinical, socioeconomic, agricultural, and environmental pressures. Inappropriate antimicrobial use, limited access to microbiological diagnosis, weak infection-prevention infrastructure, inadequate water, sanitation, and hygiene (WASH), suboptimal governance, and antimicrobial contamination of the environment can all facilitate the selection and spread of resistant organisms (Collignon et al., 2018; Holmes et al., 2016; Iskandar et al., 2021; Nashwan et al., 2024; Otaigbe & Elikwu, 2023). These drivers vary substantially across and within countries, so LMICs should not be treated as a single homogeneous surveillance setting.

Routine culture and antimicrobial-susceptibility testing (AST) remain the foundation of clinically actionable AMR surveillance. They show whether an isolate is susceptible or resistant under standardized laboratory conditions and are indispensable for guiding treatment. Their limitations are different: routine phenotypic systems often provide less information about the genetic basis of resistance, the relatedness of isolates, horizontal gene transfer, and reservoirs outside clinical culture collections. Genomic tools can add this missing resolution, but they should complement rather than replace phenotypic surveillance (Lim et al., 2021; WHO, 2020).

This narrative review focuses on how whole-genome sequencing (WGS) and metagenomic surveillance can strengthen AMR monitoring in LMICs. It examines their distinct roles, selected applications in clinical and environmental surveillance, implementation barriers, and a practical framework for integrating genomic methods into resource-constrained surveillance systems. The review is intentionally narrative and uses illustrative studies from multiple LMIC settings rather than claiming exhaustive systematic coverage.

2. AMR Surveillance Gaps in LMICs

Surveillance quality depends on more than the availability of a sequencing platform. Many LMIC surveillance systems continue to face gaps in access to routine microbiology, standardized AST, representative sampling, laboratory quality assurance, electronic reporting, and linkage of laboratory data to clinical and epidemiological information (Iskandar et al., 2021; Sharma et al., 2022; Do et al., 2023). Consequently, surveillance datasets may over-represent tertiary hospitals or urban populations while under-representing community infections, rural settings, animal production systems, and environmental reservoirs.

The same structural conditions that increase infectious-disease transmission can also amplify AMR. Inadequate WASH and wastewater management facilitate the movement of organisms and resistance determinants between households, healthcare facilities, agriculture, and the wider environment. Governance, healthcare spending, private-sector regulation, and sanitation have all been associated with national variation in AMR (Collignon et al., 2018). Environmental studies likewise show that pharmaceutical residues, sewage, and resistant organisms can enter surface waters where wastewater treatment is insufficient (Kairigo et al., 2020; Nashwan et al., 2024).

A genomic surveillance strategy therefore has to solve two problems at once: it must generate higher-resolution biological information, and it must be embedded in a system capable of collecting representative samples, preserving isolates or nucleic acid, capturing metadata, analyzing results, and acting on them. Sequencing without these supporting functions risks producing technically sophisticated data with limited public-health value.

3. Whole-Genome Sequencing for AMR Surveillance

WGS determines the nucleotide sequence of nearly the entire genome of a cultured isolate. In bacterial surveillance, the usual workflow includes isolate selection, DNA extraction, library preparation, sequencing, quality control, genome assembly or read mapping, organism confirmation, and analysis of resistance determinants, virulence factors, mobile genetic elements, and genomic relatedness. The specific pipeline depends on the organism, platform, surveillance question, and available computational resources (WHO, 2020; Okeke et al., 2020; Purushothaman et al., 2022).

For AMR surveillance, one major advantage of WGS is that multiple questions can be addressed from a single sequence dataset. Known acquired resistance genes and resistance-associated mutations can be identified, while phylogenetic or genomic-distance analyses can determine whether isolates are closely related. This is especially useful in outbreak investigation, where phenotypically similar isolates may belong to unrelated lineages, or genetically related isolates may reveal transmission that routine epidemiology alone did not resolve (Gwinn et al., 2019; Hendriksen et al., 2019).

WGS also supports molecular surveillance over longer time scales. Genotyping frameworks such as GenoTyphi demonstrate how genomic data can be used to monitor the emergence and geographic spread of Salmonella Typhi lineages and resistance-associated clones (Dyson & Holt, 2021). In LMIC settings, WGS has similarly revealed resistance-associated mutations in Mycobacterium tuberculosis and characterized multidrug-resistant Enterobacterales and other priority pathogens (Mboowa et al., 2021; Medugu et al., 2025).

Genotypic prediction is not identical to phenotypic susceptibility testing. The performance of sequence-based resistance prediction varies by organism, antimicrobial agent, mechanism, database completeness, and analytical pipeline. Resistance genes may be present without corresponding phenotypic expression, while resistance can also arise through mechanisms that are poorly represented in reference databases. Reports of genotype-phenotype discordance illustrate why genomic results should be interpreted alongside validated phenotypic methods, particularly when decisions affect individual patient management (Urmi et al., 2020; Solanki et al., 2022).

4. Metagenomic Surveillance and the Environmental Resistome

Metagenomics analyzes genetic material directly from complex samples without requiring each organism to be cultured separately. Shotgun metagenomic sequencing can profile the taxonomic composition and functional gene content of a microbial community, including antimicrobial-resistance genes (ARGs) and, depending on sequencing depth and analysis, mobile genetic elements or genomic context (Nam et al., 2023; Purushothaman et al., 2022). Targeted amplicon approaches answer narrower taxonomic questions and should not be treated as equivalent to shotgun resistome profiling.

For AMR surveillance, metagenomics is particularly relevant to wastewater, sewage, animal, and environmental sampling. These matrices contain DNA from diverse microbial populations and may provide a community-level signal that is difficult to obtain from clinical cultures alone. Population-level faecal metagenomic profiling in Cambodia and Kenya, alongside the United Kingdom, showed associations between population resistome patterns and resistance in invasive Enterobacterales, illustrating the potential of metagenomics as a complementary population-surveillance tool (Auguet et al., 2021).

Wastewater studies are attractive because a single sample can integrate signals from many individuals, including people who do not seek healthcare. The global sewage analysis by Munk et al. (2022) demonstrated marked geographic variation in the abundance and composition of resistance genes across 101 countries. In Peru, Poterico et al. (2023) used metagenomic analysis of sewage samples to characterize the resistome of an urban setting, providing a locally specific example of how environmental surveillance can identify resistance patterns outside routine clinical laboratories.

Metagenomics also has important limitations. Detection of an ARG does not automatically identify the bacterial host, demonstrate expression, or establish phenotypic resistance. Low-abundance organisms or genes may be missed, and results are sensitive to sampling design, DNA extraction, sequencing depth, contamination control, reference databases, and bioinformatic methods. These constraints are particularly important when surveillance programs compare data across laboratories or countries.

5. WGS and Metagenomics Are Complementary, Not Interchangeable

WGS and metagenomics answer related but different surveillance questions. Isolate WGS is strongest when a laboratory needs high-resolution information about a specific cultured pathogen, including lineage, resistance mechanisms, and transmission. Metagenomics is strongest when the surveillance question concerns a microbial community or reservoir that cannot be represented adequately by selected cultured isolates. A resilient AMR surveillance system therefore uses each method selectively rather than assuming that one platform can replace routine culture, AST, and epidemiological investigation.

Feature	Isolate WGS	Shotgun metagenomics
Primary sample	Cultured isolate	Complex clinical, animal or environmental sample
Main surveillance strength	Strain-level characterization, resistance determinants, outbreak relatedness	Community composition and resistome profiling without culture
Phenotype linkage	Can be linked directly to AST from the same isolate	Usually indirect; ARG detection does not prove phenotypic resistance
Transmission analysis	Strong for clonal/lineage transmission when sampling is appropriate	Useful for reservoir trends; host attribution may be difficult
Environmental surveillance	Possible after culture of selected organisms	Particularly useful for wastewater and mixed microbial reservoirs
Bioinformatic complexity	High but increasingly standardized for priority pathogens	Very high; strongly affected by depth, databases and host/ARG assignment
Best LMIC use case	Targeted sequencing of priority pathogens, unusual phenotypes and outbreaks	Strategic wastewater/One Health surveillance at selected sites

6. Illustrative Evidence from LMICs

6.1. Nigeria: WGS of Multidrug-Resistant Escherichia coli

A recent Nigerian study provides a useful example of targeted isolate WGS in a hospital setting. Medugu et al. (2025) sequenced 107 multidrug-resistant E. coli isolates from a tertiary hospital and analyzed virulence genes, mobile genetic elements, phylogroups, sequence types, and AMR determinants. The study reported extensive linkage between virulence-associated genes and mobile genetic elements and demonstrated the genomic diversity of multidrug-resistant E. coli circulating in the clinical setting. The surveillance value of this approach lies not simply in confirming that isolates are resistant, but in showing how resistance and virulence determinants are distributed across lineages and potentially mobile genetic contexts.

Nigeria has also developed national experience with WGS-based AMR reference-laboratory capacity. Okeke et al. (2022) described establishment of a national reference laboratory using a WGS framework, highlighting that successful genomic surveillance requires quality systems, bioinformatics, data management, and networks that connect sequencing to routine microbiology rather than a stand-alone instrument.

6.2. Peru: Sewage Metagenomics and the Urban Resistome

Poterico et al. (2023) used metagenomic sequencing of sewage samples to characterize the resistome of a Peruvian city. This case illustrates a different surveillance objective from hospital isolate WGS: the aim is to observe resistance determinants circulating at the community-environment interface. Such surveillance can reveal ARG distributions that would be difficult to reconstruct from a collection of clinical isolates alone. The study should be interpreted as a local environmental resistome analysis, while Munk et al. (2022) provides a separate global comparison across sewage samples from many countries; the two sources should not be presented as the same dataset.

6.3. Pakistan: WGS of Extensively Drug-Resistant Salmonella Typhi

Extensively drug-resistant (XDR) Salmonella Typhi is a major genomic-surveillance priority in Pakistan. Mumtaz et al. (2024) reported WGS of XDR S. Typhi isolates from the Peshawar region and described resistance-associated genomic features and low genetic diversity consistent with continued circulation of closely related strains. This type of analysis complements phenotypic confirmation by placing resistant isolates into a transmission and evolutionary context. Earlier regional clinical reports, including Khan et al. (2022), provide epidemiological background but should not substitute for the genomic study when making WGS-specific claims.

Setting	Approach	Primary value	Interpretive caution
Nigeria - hospital MDR E. coli	WGS of 107 cultured isolates	Links resistance/virulence determinants with lineages and mobile elements	Hospital sample; findings should not be generalized to all Nigerian E. coli
Peru - urban sewage	Shotgun metagenomic resistome analysis	Community/environmental view of ARG diversity	ARG detection does not identify clinical burden or prove phenotype
Pakistan - XDR S. Typhi	WGS of selected clinical isolates	Characterizes resistance features and genomic relatedness	Small sequenced subset; requires linkage to broader epidemiology

8. A Tiered Framework for Genomic AMR Surveillance in LMICs

A realistic approach is to build genomic surveillance on top of progressively stronger microbiology systems. This avoids an all-or-nothing model in which countries are expected to move directly from limited culture capacity to universal sequencing.

Tier 1 - Strengthen sentinel microbiology. Core investments should include specimen quality, organism identification, standardized AST, laboratory quality assurance, isolate archiving, and standardized epidemiological metadata. These functions create the denominator and context that genomic data need.

Tier 2 - Introduce targeted WGS for priority questions. Sequencing should initially focus on high-consequence organisms, unusual or emerging resistance phenotypes, suspected outbreaks, treatment failures, and representative sentinel collections. Centralized or regional sequencing can be more sustainable than placing low-throughput instruments in every hospital.

Tier 3 - Develop regional sequencing and bioinformatics hubs. Hubs can provide library preparation, sequencing, quality assurance, analysis, training, and surge capacity for multiple hospitals or neighboring countries. The aim should be rapid return of interpretable results rather than accumulation of sequence files.

Tier 4 - Add strategic metagenomic surveillance. Wastewater or environmental metagenomics should be used where it answers a defined One Health or population-surveillance question and where sampling, sequencing depth, and analytical standards can be maintained. It is best viewed as a complement to clinical surveillance, not a proxy for clinical resistance prevalence.

Tier 5 - Integrate One Health genomic information into decision making. Mature systems can link human, animal, food, and environmental surveillance through interoperable metadata and shared analytic standards. Cross-sectoral capability frameworks are useful because One Health integration depends on governance and communication as much as sequencing technology (Ferdinand et al., 2024; Vallejo-Espin et al., 2025).

Tier	Core activity	Minimum requirement	Decision supported
1	Sentinel culture + AST	Quality microbiology, isolate archive, metadata	Reliable local/national resistance trends
2	Targeted WGS	Sequencing access + validated analysis	Outbreaks, unusual resistance, lineage monitoring
3	Regional genomic hubs	Workforce, QA, computing, referral network	Scalable national/regional genomic surveillance
4	Strategic metagenomics	Standardized sampling + deep bioinformatics	Wastewater/One Health resistome trends
5	Integrated One Health system	Governance + interoperable data sharing	Cross-sector risk assessment and policy

10. Limitations of This Narrative Review

This article is a narrative review and does not present a systematic review or meta-analysis. The studies discussed are illustrative rather than exhaustive, and the evidence base is heterogeneous in organism, setting, sample type, sequencing platform, and outcome. Several implementation claims are therefore context dependent. In particular, cost-effectiveness evidence for isolate WGS should not be generalized to metagenomic surveillance without direct economic evaluation, and environmental ARG abundance should not be interpreted as equivalent to clinical resistance burden.

11. Conclusion

WGS and metagenomics can substantially improve the resolution of AMR surveillance in LMICs, but their value depends on how they are positioned within broader microbiology and public-health systems. Isolate WGS is especially useful for identifying resistance mechanisms, characterizing lineages, and resolving transmission among priority pathogens. Metagenomics extends surveillance to complex microbial communities and environmental reservoirs, particularly wastewater, but presents greater challenges in interpretation and standardization. Neither approach removes the need for high-quality culture, AST, epidemiology, and infection-prevention infrastructure.

The most sustainable pathway is a tiered model that begins with reliable sentinel microbiology, introduces targeted WGS for defined public-health questions, develops regional sequencing and bioinformatics capacity, and adds metagenomic surveillance where it provides clear additional information. Nigeria, Peru, and Pakistan demonstrate different but complementary uses of genomic methods. The next phase of AMR genomics in LMICs should therefore focus less on demonstrating that sequencing is technically possible and more on building systems in which genomic evidence is representative, interpretable, sustainable, and directly connected to action.