Evaluation of Oxford Nanopore Sequencing for Antimicrobial Resistance Surveillance in Salmonella: Comparison with Phenotypic Antimicrobial Susceptibility in a Large-Scale Study

This large-scale study evaluating Oxford Nanopore sequencing on 1,490 *Salmonella* isolates from Taiwan demonstrates that while genotypic resistance generally correlates with phenotypic results, specific discrepancies driven by breakpoint definitions, gene expression, and unknown determinants highlight the need for careful interpretation to leverage whole-genome sequencing as a superior alternative to conventional antimicrobial susceptibility testing for surveillance and treatment guidance.

The Gist

Imagine Salmonella as a notorious burglar that breaks into our food supply, making people sick. For a long time, doctors and scientists have tried to catch this burglar by testing how well different "locks" (antibiotics) keep it out. This traditional method, called phenotypic testing, is like trying every single key in a giant keyring to see which one fits the lock. It works, but it takes time and can sometimes give confusing results.

This paper describes a study where scientists in Taiwan tried a new, high-tech tool: Oxford Nanopore Sequencing (ONT). Think of this as a super-fast, high-definition camera that doesn't just try keys; it takes a perfect photograph of the burglar's blueprints (their DNA) to see exactly which tools they are carrying.

Here is what the study found, broken down simply:

1. The High-Tech Camera vs. The Old Keyring
The researchers took 1,490 samples of Salmonella and used the new DNA camera to predict which antibiotics would work. They then compared these predictions to the old "try-every-key" method.

• The Good News: For most antibiotics, the DNA camera and the old keyring agreed perfectly. The new method is fast and sees the whole picture.

2. When the Two Methods Disagreed
Sometimes, the DNA camera said, "This burglar has a master key!" but the old keyring said, "No, the lock still works." Or vice versa. The study found four main reasons for these mix-ups:

• The Ruler Problem: Sometimes the difference depends on how strictly you measure the "lock." The DNA sees the tool, but the old test has a specific line (a breakpoint) it uses to decide if the tool is dangerous enough to count.
• The Sleeping Giant: Sometimes the burglar has the master key in their pocket (the gene), but they aren't using it right now. The DNA sees the key, but the old test doesn't see the burglar trying to pick the lock.
• The Volume Knob: There's a specific switch in the bacteria called ramAp that acts like a volume knob. It can turn the bacteria's resistance up or down. The DNA sees the switch, but the old test might not realize how loud the resistance is actually getting.
• The Missing Blueprint: Sometimes the old test says the lock is broken, but the DNA camera can't find the master key in the blueprints. This happened often with colistin (a strong antibiotic) and nalidixic acid. The bacteria were resistant, but the scientists couldn't find the specific gene responsible for it yet.

3. The "False Sense of Security" Trap
One of the most important findings was about ESBL and AmpC bacteria (a type of super-burglar). The old keyring method sometimes labeled these dangerous bacteria as "safe" (susceptible) or "maybe safe" (intermediate) against certain antibiotics like cefotaxime.

• The Metaphor: It's like a security guard saying, "This door is locked," when in reality, the burglar has a tool that can pick that specific lock easily. The study warns that relying only on the old method might lead to picking the wrong "key" (antibiotic) to treat the patient, causing the treatment to fail.

The Bottom Line
The study concludes that this new DNA camera (ONT-WGS) is a powerful tool. It can see the burglar's tools directly, rather than guessing based on how the burglar behaves. While it needs to be read carefully to understand those tricky "volume knobs" and "missing blueprints," it offers a clearer, more accurate way to figure out which antibiotics will actually stop Salmonella, potentially avoiding the mistakes that happen with the older, slower testing methods.

Technical Summary

Technical Summary: Evaluation of Oxford Nanopore Sequencing for Antimicrobial Resistance Surveillance in Salmonella

Problem Statement
Salmonella remains a critical zoonotic foodborne pathogen, with antimicrobial resistance (AMR) posing a substantial public health threat. Conventional antimicrobial susceptibility testing (AST) is often slow and lacks the granularity required for comprehensive surveillance. While Whole-Genome Sequencing (WGS) offers a promising alternative for rapid AMR characterization, the specific performance of Oxford Nanopore Technology (ONT)-based WGS in large-scale surveillance contexts requires rigorous validation against phenotypic standards to ensure reliability in clinical and public health decision-making.

Methodology
This study conducted a large-scale comparative analysis involving 1,490 Salmonella isolates collected through nationwide surveillance in Taiwan in 2025. The researchers employed ONT-based WGS to generate genomic data for all isolates. Genotypic resistance profiles were inferred directly from the sequencing data and systematically compared against phenotypic AST results. The primary objective was to assess the concordance between WGS-inferred resistance and phenotypic susceptibility, specifically investigating the sources and nature of any discrepancies observed.

Key Results
The study found that, overall, WGS-inferred resistance demonstrated high concordance with phenotypic resistance across most antimicrobials. However, significant genotype-phenotype discordance was identified and categorized into four distinct mechanisms:

1. Breakpoint-dependent classification: Discrepancies arising from the specific thresholds used in phenotypic testing.
2. Reduced or absent phenotypic expression: Cases where resistance genes were present but not expressed phenotypically.
3. MIC modulation by ramA: Specifically noted in the context of nalidixic acid resistance, where ramA mediated elevation of Minimum Inhibitory Concentrations (MIC).
4. Absence of known AMR determinants: Instances where resistance was observed phenotypically without identifiable genetic markers.

Specific notable findings included:

• Tigecycline: Resistance was detected without known genetic determinants.
• Nalidixic Acid: Resistance was linked to ramA-mediated MIC elevation.
• Colistin: A high prevalence of resistance (35.4%) was observed in S. Enteritidis isolates despite the absence of identifiable AMR determinants.
• Cephalosporins: A significant proportion of isolates producing Extended-Spectrum Beta-Lactamases (ESBLs) and AmpC enzymes were classified as susceptible or intermediate to cefotaxime and ceftazidime under Clinical and Laboratory Standards Institute (CLSI) criteria. This highlights a risk of misclassification that could lead to treatment failure.

Key Contributions
The study validates ONT-WGS as a tool capable of providing accurate and comprehensive AMR characterization. It specifically demonstrates the utility of WGS in directly identifying AMR determinants, thereby offering a mechanism to minimize misclassification errors that can occur with breakpoint-based phenotypic interpretations. The research elucidates specific biological and technical reasons for discordance, such as the limitations of current genetic databases in explaining certain resistance phenotypes (e.g., colistin and tigecycline) and the complexities of gene expression.

Significance and Claims
The paper asserts that when interpreted with appropriate caution regarding known discordance mechanisms, ONT-WGS serves as a valuable alternative to conventional susceptibility testing. The authors claim that this approach supports better antimicrobial selection by providing direct identification of resistance determinants. The study positions WGS not merely as a research tool but as a viable method for supporting antimicrobial therapy and enhancing surveillance efforts, provided that the limitations regarding specific resistance mechanisms (such as those lacking known genetic markers) are acknowledged.