A patient with a bloodstream infection caused by extensively drug-resistant Escherichia coli was failing to respond to treatment. The organism carried multiple resistance mechanisms, including the New Delhi metallo-β-lactamase NDM-5 and an insertion in penicillin-binding protein 3 (PBP3), a cellular target of several β-lactam antibiotics. As the infection persisted, clinicians at the University of Texas MD Anderson Cancer Center used genomic information about the pathogen to identify a different treatment strategy.
At IDWeek 2026, clinical microbiologist Micah M. Bhatti described how his team used the organism’s resistance profile to guide the addition of an agent with complementary antibacterial activity. The revised regimen was followed by clearance of the patient’s blood cultures and clinical improvement. Although the case cannot establish the independent contribution of the added agent, it illustrates how mechanistic knowledge can inform treatment when conventional options are failing.
Antibiotic developers face a more difficult version of this problem. Clinicians can analyze resistance mechanisms already present in a patient’s infection, but developers must consider how bacteria might adapt to a drug that has not yet reached clinical use. An antibiotic designed to overcome today’s resistance mechanisms may still be vulnerable to mutations or physiological changes that emerge under the selective pressure of treatment.
Recent studies have demonstrated ways to investigate some of these vulnerabilities before clinical deployment. Environmental resistance genes can reveal weaknesses in investigational compounds, experimental evolution can identify potential escape routes, and systematic susceptibility mapping can inform the selection of complementary agents. These approaches raise the possibility of using anticipated resistance to guide molecular optimization and combination development while researchers still have opportunities to change the therapies they are designing.
From Protecting Antibiotics to Targeting Complementary Mechanisms
The development of β-lactam combinations provides an established example of how understanding resistance can influence therapeutic design. Many Gram-negative bacteria produce β-lactamases, enzymes that degrade β-lactam antibiotics before they can reach their cellular targets. Pairing an antibiotic with an appropriate β-lactamase inhibitor can protect it from these enzymes and restore activity against otherwise resistant organisms.
Researchers have also developed compounds that contribute antibacterial activity through complementary target engagement. β-lactam enhancers bind PBPs, enzymes involved in bacterial cell-wall synthesis, allowing combinations to interfere with more than one essential component of that process.
Cefepime/zidebactam illustrates this strategy. Cefepime primarily targets PBP3, while zidebactam binds PBP2 and inhibits certain β-lactamases. In a 2019 study published in Antimicrobial Agents and Chemotherapy, researchers at the University of Wisconsin–Madison investigated the combination against metallo-β-lactamase-producing Enterobacteriaceae.1 They found substantial antibacterial activity despite zidebactam’s inability to inhibit metallo-β-lactamases directly. The combination reduced cefepime minimum inhibitory concentrations (MICs) approximately 128-fold across eight tested isolates, and experiments in a mouse pneumonia model supported the contribution of complementary PBP engagement to bacterial killing.
Targeting complementary PBPs can preserve antibacterial activity when an established resistance mechanism compromises one component of a combination. However, bacteria may still evolve ways to evade the resulting therapy, including adaptations that affect drug accumulation or the binding of either agent to its intended target.
How Bacteria Escape Multitarget Antibiotic Combinations
The emergence of resistance to cefepime/zidebactam illustrates why complementary mechanisms do not necessarily eliminate evolutionary escape routes. Although the combination can overcome several established resistance mechanisms, its activity still depends on both agents reaching their targets and retaining sufficient antibacterial effects.
In a report published online in 2025 and appearing in the January 2026 issue of Clinical Infectious Diseases, researchers at the University of Pittsburgh and collaborating institutions described a patient with Pseudomonas aeruginosa pneumonia who developed resistance during cefepime/zidebactam treatment.2 The combination’s MIC increased from 8/8 to 32/32 mg/L, accompanied by clinical failure. Genomic analysis identified new mutations in the MexAB-OprM efflux system, which helps bacteria expel antibacterial compounds. These mutations were associated with cross-resistance to other β-lactam combinations, demonstrating how altered drug accumulation can compromise treatment.
Other adaptations can directly interfere with target engagement. Researchers at Hospital Universitari Son Espases and collaborating institutions in Spain investigated a PBP2 mutation designated V516M in P. aeruginosa. In their 2026 study published in Antimicrobial Agents and Chemotherapy, the mutation substantially reduced zidebactam binding and increased resistance. It also affected related agents differently, with cross-resistance to nacubactam but not durlobactam.3 The investigators found no measurable reduction in growth rate or virulence associated with the mutation in their experimental models, identifying a potential escape route without an apparent fitness penalty under the conditions tested.
A third study demonstrates how identifying a resistance mechanism can inform the selection of an alternative combination partner. In research published in npj Antimicrobials and Resistance in 2026, researchers at the University of Illinois Chicago, Revagenix, and the University at Buffalo examined NDM-producing E. coli carrying insertions in PBP3.4 These modifications compromised the synergistic activity of cefepime/zidebactam in the three tested isolates with PBP3 insertions. However, pairing zidebactam with aztreonam restored synergistic killing experimentally, demonstrating that changing the PBP3-active component could overcome a particular resistance-associated vulnerability.
The three studies reveal different weaknesses that developers might investigate before selecting a candidate combination. Efflux-associated resistance highlights the importance of drug accumulation, target mutations expose vulnerabilities in molecular binding, and the restoration of synergy through an alternative partner demonstrates the value of evaluating combinations against specific resistance genotypes.
These findings also suggest that resistance assessment need not depend on predicting a single mutation. Developers can investigate multiple potential escape routes and use that information to modify a molecule, select a complementary agent, or explore biological processes that facilitate resistance evolution.
Using Resistance Forecasting to Design Better Antibiotic Molecules
What if researchers could identify an investigational antibiotic’s vulnerabilities before resistant infections exposed them? Environmental microorganisms offer one place to look. They harbor genes that protect them from antibacterial compounds, including resistance determinants that may not yet be prominent among clinical pathogens. Testing these genes against new antibiotic candidates can reveal potential weaknesses while medicinal chemists still have opportunities to modify the molecules.
Researchers at The Rockefeller University applied this strategy to albicidin, an antibacterial natural product that inhibits bacterial DNA gyrase. In a 2025 study published in Proceedings of the National Academy of Sciences, the team screened environmental DNA for genes capable of conferring albicidin resistance.5 They then investigated how those resistance determinants affected different albicidin structures, using the results to guide compound selection and chemical optimization.
The investigators identified resistance determinants through metagenomic screening and assessed their effects on albicidin activity. They then compared naturally occurring albicidin congeners, closely related molecules with structural differences that influenced their susceptibility to particular resistance mechanisms. This allowed the researchers to consider resistance profiles alongside antibacterial potency when evaluating candidate compounds.
One approach involved combining albicidin congeners with complementary susceptibility profiles. The researchers identified mixtures, including congeners 6, 7, and 10 and congeners 10 and 11, that collectively overcame the resistance determinants represented in their experimental panel. Rather than depending on a single structure to evade every mechanism, the mixtures exploited differences in how resistance determinants affected the component molecules.
The investigators also used the screening results to guide chemical modification of an individual molecule. Their work produced albicidin analog 18, which retained antibacterial potency while evading a high-priority resistance mechanism identified during screening. The analog remained susceptible to other determinants, illustrating the importance of evaluating a candidate against multiple potential vulnerabilities.
These experiments demonstrate two ways resistance information can influence molecular development. Researchers might modify a candidate to reduce its susceptibility to a particular mechanism, or they might select related compounds with complementary resistance profiles. Both approaches introduce resistance assessment into decisions that would otherwise emphasize antibacterial potency, physicochemical properties, and other characteristics important for development.
The Rockefeller study also illustrates an important limitation of resistance forecasting. Environmental screening identifies mechanisms that can compromise an antibiotic under experimental conditions, but it does not establish which determinants are most likely to spread into clinical pathogens or emerge during treatment. The value of the approach lies in exposing plausible vulnerabilities early enough to investigate them, rather than predicting the precise course of future resistance evolution.
This distinction has practical consequences. Developers may identify resistance mechanisms that warrant molecular optimization even without knowing whether those mechanisms will become clinically prevalent. Experimental evolution provides another source of information by exposing bacterial populations to antibiotics under controlled conditions and identifying adaptations that arise during selection. Combining these approaches could broaden the range of resistance pathways considered before a candidate advances.
Choosing Combination Partners With Resistance Evolution in Mind
Resistance to one antibiotic can also change susceptibility to another. Sometimes the same adaptation reduces susceptibility to multiple drugs, a phenomenon known as cross-resistance. In other cases, resistance to one compound increases susceptibility to another, creating collateral sensitivity. These relationships offer a possible basis for selecting combinations that remain active against particular resistant variants or constrain some pathways through which resistance develops.
In a study published in Nature Microbiology in December 2024, researchers at the European Molecular Biology Laboratory and collaborating institutions systematically mapped these relationships in E. coli.6 Using chemical genetics data from bacterial strains carrying individual gene deletions, the researchers analyzed responses to 40 antibiotics. They inferred 404 cross-resistance relationships and 267 collateral-sensitivity relationships, subsequently validating 64 of 70 tested interactions through experimental evolution. Selected combinations involving collateral-sensitive drugs also reduced resistance development in vitro.
The study revealed that relationships between antibiotics can depend on the underlying resistance mechanism. A drug pair might exhibit cross-resistance in one genetic context and collateral sensitivity in another. Consequently, the evolutionary behavior of a combination cannot necessarily be inferred from the identities of its component drugs alone.
For a new antibacterial candidate, such mapping could inform the selection of companion agents during development. Researchers might expose bacterial populations to the lead compound, characterize the resistance mechanisms that emerge, and determine whether another antibiotic retains activity against those variants. The findings could help prioritize combinations for further investigation before substantial development resources are committed.
A more exploratory strategy would target biological factors that facilitate resistance evolution itself. In a 2026 essay published in PLOS Biology, researchers at the University of Oxford and the University of Bristol examined resistance potentiators, bacterial genes, genetic elements, and pathways that accelerate the evolution of resistance to particular antibiotics.7 Drawing on experimental evolution, pathogen genomics, and molecular microbiology, the authors described how these factors can influence the acquisition or effects of resistance determinants.
The concept raises the possibility of developing companion agents that interfere with the processes enabling resistance, rather than directly inhibiting an established resistance mechanism. Experimental findings provide a rationale for investigating certain potentiators, but clinical validation is lacking, and bacteria may retain alternative evolutionary routes.
The approaches therefore differ in their immediate objectives. Resistance-guided partner selection may preserve activity against resistant variants, while targeting potentiators aims to influence the process through which resistance arises. Both require mechanism-specific evidence, and neither can be assumed to produce greater clinical durability simply because it involves complementary agents.
Translating Complementary Mechanisms Into Effective Drug Exposure
Even a well-chosen combination can fail if its components do not achieve sufficient concentrations at the infection site. Laboratory experiments typically evaluate antibacterial activity under controlled conditions, but drug concentrations change throughout treatment. When both agents contribute to bacterial killing, the concentration of one may influence the activity of the other.
This creates challenges for pharmacokinetic/pharmacodynamic (PK/PD) evaluation, which relates drug exposure to antibacterial effects. The cefepime/zidebactam experiments conducted by researchers at the University of Wisconsin–Madison demonstrated that complementary activity could substantially alter the cefepime exposure associated with bacterial killing.1 However, exposure requirements varied among the organisms studied, complicating the selection of a single pharmacodynamic target for resistant pathogens.
Researchers are investigating methods that account more directly for changing concentrations of both agents. In a study published in the January 2026 issue of the Journal of Infection and Chemotherapy, researchers at Keio University evaluated cefepime/nacubactam in a mouse pneumonia model using an instantaneous MIC, a measure that reflects changing susceptibility as the concentration of the companion agent varies.8 The investigators incorporated drug concentrations in epithelial lining fluid, the fluid covering the lung’s air-exchange surfaces, to examine how infection-site exposure relates to antibacterial activity.
This approach addresses a practical challenge for combination development. Two agents may demonstrate complementary activity in susceptibility testing, but their interaction during treatment depends on the concentrations each achieves over time. During her presentation in the same session at IDWeek 2026, Ashlan Kunz-Coyne of the University of Kentucky College of Pharmacy highlighted related questions about infection-site exposure, the degree of target engagement required for bacterial killing, and the limits of increasing drug exposure to overcome resistance.
For developers, resistance-informed combination selection must therefore be accompanied by pharmacodynamic evaluation of the agents together. A promising mechanistic pairing must also support a feasible dosing strategy capable of sustaining its intended antibacterial effects.
What Would Resistance-Informed Development Look Like in Practice?
The studies discussed here suggest several points at which resistance information could influence antibiotic development. During lead optimization, investigators could screen candidate molecules against environmental resistance determinants and experimentally evolved mutants, using the findings to prioritize structures with fewer identified vulnerabilities. For compounds that remain susceptible to important resistance mechanisms, researchers could investigate whether structural modification or a complementary agent offers a more promising route to preserving activity. Combination candidates could then undergo resistance-evolution and pharmacodynamic testing before advancing into more resource-intensive development.
Developing reliable methods for these decisions is an active research objective. The Combination Therapy Development Toolbox against Antimicrobial Resistance (COMBAT-AMR), an international project coordinated at Leiden University, aims to create standardized experimental and computational approaches for designing antibacterial combinations.9 Funded for 2025–2028, the initiative plans to integrate experimental models that simulate human dosing schedules with computational methods for evaluating combination effects on bacterial pathogens. Its objectives include improving combination selection and dosing while minimizing resistance selection. The researchers also intend to use their findings to support future regulatory qualification of the proposed workflows.
The initiative addresses a gap between identifying complementary antibacterial mechanisms and systematically evaluating candidate combinations. Researchers need methods that account for the contributions of both agents, compare potential resistance pathways, and assess how those relationships change under clinically relevant exposure conditions. COMBAT-AMR intends to develop and evaluate such tools, although its proposed outcomes have not yet been established.
For developers considering a more prospective approach, the central questions concern when and how resistance information should influence investment decisions. Screening during early discovery might reveal a vulnerability that medicinal chemistry could address before candidate selection. Other findings might justify investigating a complementary agent or reconsidering whether a particular lead merits further development. Experimental resistance profiles could also help determine which combinations deserve more extensive pharmacodynamic evaluation.
These decisions would involve trade-offs. Additional resistance screening requires resources, and the discovery of a plausible escape mechanism does not establish its future clinical importance. Developers would need to weigh the relevance of identified vulnerabilities against antibacterial potency, safety, manufacturability, and other properties that determine whether a candidate can become a useful medicine.
The potential value of resistance forecasting lies in informing these choices before a development program becomes committed to a particular molecule or pairing. Whether incorporating such assessments improves the likelihood of producing clinically durable antibiotics remains to be demonstrated.
What Would Demonstrate That Resistance-Informed Design Produces More Durable Therapies?
The evidence supporting resistance-informed development spans several distinct outcomes. Complementary mechanisms can improve antibacterial killing, laboratory experiments can identify potential escape routes, and resistance-guided optimization can produce compounds with improved activity against defined resistance determinants. Maintaining activity against resistant variants and reducing the probability that resistance emerges are related but distinct objectives, and each requires appropriate experimental evidence.
Experience with antibiotic combinations more broadly illustrates the importance of this distinction. Researchers at ETH Zürich, University Hospital Zürich, and collaborating institutions examined the effects of combining antibiotics on within-patient resistance development in a systematic review and meta-analysis released as a medRxiv preprint in 2024.10 The investigators identified 42 eligible randomized trials, including 29 trials involving 5,054 patients in the quantitative analysis. They found no clear overall evidence that treatment with more antibiotics reduced resistance emergence. The studies were heterogeneous, resistance was frequently not the primary outcome, and many trials lacked sufficient statistical power to evaluate it reliably.
The review does not specifically evaluate combinations prospectively designed around predicted resistance mechanisms, and it did not count β-lactamase inhibitors as separate antibiotics. Its findings therefore cannot determine whether resistance-informed combinations would outperform conventional approaches. They do underscore the importance of evaluating resistance outcomes directly rather than assuming that combining agents will suppress resistance.
A rigorous assessment of prospective resistance-informed development would begin by determining whether mechanisms identified during screening are relevant to the bacterial populations and treatment conditions a candidate is likely to encounter. Experimental evolution could reveal whether resistant variants emerge through anticipated pathways or through adaptations that the original screening failed to identify. Testing across diverse bacterial backgrounds would help establish how broadly the findings apply.
Subsequent studies would need to determine whether the selected compounds or combinations retain their advantages under clinically relevant exposure conditions and in infection models. These investigations could also clarify whether resistance suppression requires exposures that are feasible and tolerable in patients.
Ultimately, the strategy’s value will depend on whether incorporating resistance forecasting into development produces therapies that maintain efficacy or reduce resistance emergence compared with alternatives. The available evidence supports investigating that possibility, but improved activity against selected resistant variants and reduced resistance evolution in laboratory systems cannot yet be equated with greater clinical durability.
Making Anticipated Resistance Part of Antibiotic Design
This and other clinical cases described at IDWeek illustrates how understanding resistance mechanisms can inform treatment when established options fail. For antibiotic developers, the more ambitious possibility is to investigate those vulnerabilities before they become obstacles to clinical use.
Recent studies demonstrate several ways to pursue that objective. Environmental resistance screening can identify weaknesses in candidate molecules and guide structural optimization. Experimental evolution can reveal adaptations that compromise complementary antibacterial mechanisms. Systematic susceptibility mapping can identify relationships among drugs that inform combination selection, while pharmacodynamic models can help determine whether promising pairings are likely to function under relevant exposure conditions.
These approaches remain at different stages of development, and none eliminates the possibility that bacteria will evolve through unanticipated pathways. Their importance lies in expanding the information available when developers choose which compounds and combinations to advance. The value of resistance forecasting may ultimately lie in the decisions it changes: which molecules advance, which combinations warrant investment, and which vulnerabilities developers choose to address before clinical resistance reveals them.
