Antibiotics are a foundation of modern medicine and have saved hundreds of millions of lives since their discovery. Every year, roughly one in three people worldwide rely on them to treat common infections such as urinary tract infections, ear infections, and pneumonia, and they are essential for the hundreds of millions who undergo surgery, receive chemotherapy, or need an organ transplant each year. The emergence of antibiotic-resistant infections is therefore a major and growing global public health threat.

The aim of Resistance Networks is to measure the extent to which resistance moves through bacterial networks and to design interventions that prevent or slow that progression. Doing so could extend the efficacy of existing antibiotics and protect new antibiotics under development.

Antibiotic resistance is typically detected only after it has become a clinical problem — after an infection is found untreatable, a resistant bacterium isolated and tested, and in rare cases sequenced. Recent shifts in our understanding of microbial evolution suggest resistance emerges much earlier, moving from the bacterial networks that exist in and around us into pathogens that ultimately cause disease.1–3

Genomic analysis has revealed the evolutionary power of small, transmissible circular pieces of DNA called plasmids, which collect and concentrate antibiotic resistance genes (ARGs) and move them, cell to cell, through bacterial networks.4,5 Some specialize in shuttling these genes — referred to here as pARGs — between cells through conjugation,6–8 letting bacteria acquire complex traits like resistance, virulence, and fitness by acquisition rather than evolving them from scratch.9,10

The H30-Rx sub-lineage of Escherichia coli — the dominant strain within ST131, which accounts for a large share of ESBL-producing E. coli infections worldwide — shows the stakes. It did not evolve its resistance gradually; it did so in two moments. Between 1979 and 1984, a cell already resistant to fluoroquinolones acquired a single plasmid conferring resistance to four more antibiotic classes at once; around 1992 it acquired a second. Within a decade, H30-Rx had been detected on every inhabited continent.11–15 This was not a single unlucky sequence: the same pattern recurred with ST69 — globally endemic within a decade of a plasmid acquisition around 199013 — and in Klebsiella, Salmonella, and other highly virulent E. coli.16–18

Antibiotic resistance does not spread only when resistant bacteria spread. It spreads when the genetic instructions for resistance move between bacteria — carried on plasmids that cross strain boundaries, species boundaries, and communities.19–21

The rate at which these lineages acquire resistance, the conditions that drive it, whether it transmits, and how we might intervene have never been directly measured. That is the gap this program is designed to close. The need is acute: in 2021, bacterial infections were associated with 4.71 million deaths worldwide, including 1.14 million directly attributable to antibiotic resistance,22 and on current trends AMR-attributable deaths across all bacterial pathogens are projected to exceed 1.91 million annually by 2050.22 New antibiotics are necessary but slow (up to 10 years) and expensive (roughly $1.5 billion per drug),23,24 and because resistance inevitably evolves they are often held in reserve; public health and stewardship strategies are helping but cannot keep pace.25

What if we could use epidemiological methods to detect, model, design and test interventions that prevent or slow the movement and exchange of resistance in bacterial networks?

If antibiotic resistance were only a trait that accumulates under selection pressure, the sole lever would be to reduce that pressure by changing prescribing across entire health systems. But if plasmid-mediated resistance is transmitted within the bacterial networks of the treated gut,5,26 spreads to the bacterial network of household contacts, and propagates through community or global transmission networks, the intervention calculus changes entirely.27–40 A local action — protecting one individual’s microbiome after treatment, blocking shedding at the point of amplification, or interrupting household transmission — could have compounding benefits across the entire network, and the same logic that made vaccines and infection control transformative would apply, if resistance is treated in bacterial networks as something that transmits, crosses thresholds, and can be reined in.

Resistance Networks asks whether we can reframe the problem of antibiotic resistance from targeting resistant bacteria to an epidemiological problem within bacterial networks and, in so doing, develop novel strategies that slow the spread of resistance itself.

The key epidemiological parameter is R₀, the basic reproductive number — the average number of new cases one index case generates in a susceptible, well-mixed population; above 1, spread amplifies, and below 1 it dies out. For SARS-CoV-2, the January 2020 estimate placed R₀ at 2.2 from symptomatic cases alone;41 counting asymptomatic cases revised it to 5.742,43 — the scale that produced a pandemic. R₀ cannot be observed directly: transmission events are invisible, so epidemiologists count new cases and reconstruct it.44 The central hypothesis of Resistance Networks is that plasmid-mediated antibiotic resistance spreads within bacterial networks to pathogenic bacteria like a contagion — and can be controlled like one.

For antibiotic resistance, no equivalent measurement yet exists: the events that would let us count “cases” — new plasmid acquisitions — occur inside the gut, in bacterial populations and are too complex and too infrequent to observe directly.45,46 What this program seeks to measure is a resistance gene moving into a new bacterial host, within the gut or from one individual’s microbiome to another’s; counting those events against the population of bacteria at risk is the plasmid equivalent of counting new transmissions — the empirical input from which R₀ is reconstructed.15,30