Last-resort antibiotics are drugs held back from routine use and deployed only when bacteria resist everything else. They include polymyxins (colistin and polymyxin B), certain carbapenems, and a handful of newer agents like ceftazidime-avibactam. In 2019 alone, an estimated roughly 392,000 deaths worldwide were linked to bloodstream infections caused by carbapenem-resistant bacteria, the very pathogens these drugs exist to fight.1PubMed. Global and regional burden of bloodstream infections caused by carbapenem-resistant Gram-negative bacteria in 2019: A systematic analysis from the MICROBE database These are not wonder drugs. Most are old, many are toxic, and the bacteria they target are evolving fast enough that even the last line of defense is fraying.
What Makes an Antibiotic “Last Resort”
The term has no single legal definition, but it maps closely onto what the World Health Organization calls its “Reserve” group. The WHO’s AWaRe framework sorts antibiotics into three buckets: Access drugs for common infections, Watch drugs used more cautiously, and Reserve drugs that should be used only when all other options have failed. Reserve-category antibiotics include colistin, polymyxin B, some advanced carbapenems, and newer combinations such as ceftazidime-avibactam. The Reserve label is not about potency in the traditional sense. Some of these drugs are less effective than standard antibiotics against ordinary bacteria. They earn the designation because they still work against organisms that have learned to shrug off almost everything else.
Colistin is the most recognizable last-resort drug and also the most paradoxical. It was discovered in the 1940s, fell out of favor by the 1970s because of kidney damage, and was revived decades later when multidrug-resistant gram-negative bacteria left clinicians with no alternatives. It was never designed to be a modern frontline drug. Its return was born of desperation.
How Last-Resort Drugs Kill Bacteria
Most last-resort antibiotics target bacterial structures that are difficult to modify without crippling the organism. Polymyxins, the class that includes colistin, attack the outer membrane of gram-negative bacteria. The drug carries a positive charge that binds to the negatively charged phosphate groups on lipid A, a molecule that forms the backbone of the outer membrane. This binding displaces calcium and magnesium ions that normally stabilize the membrane, disrupting its structure and increasing its permeability.2PubMed. Dissecting the Molecular Mechanism of Colistin Resistance in mcr-1 Bacteria As the membrane falls apart, the bacterial cell’s contents leak out and more drug molecules push inward. Molecular simulations show that polymyxin molecules fold into a compact shape as they penetrate toward the hydrophobic center of the membrane, causing local disorganization that amplifies the damage.3Journal of Antimicrobial Chemotherapy. Molecular dynamics simulations informed by membrane lipidomics reveal the structure–interaction relationship of polymyxins with the lipid A-based outer membrane of Acinetobacter baumannii
Carbapenems work differently. They belong to the beta-lactam family and kill bacteria by blocking penicillin-binding proteins, the enzymes bacteria need to build and maintain their cell walls. What sets carbapenems apart from ordinary penicillins and cephalosporins is their broad reach: a carbapenem like imipenem can inactivate nearly all of a bacterium’s penicillin-binding protein targets at once, making it harder for the bug to find a workaround.4PubMed Central. PBP Target Profiling by β-Lactam and β-Lactamase Inhibitors in Intact Pseudomonas aeruginosa That breadth of action is exactly what makes carbapenems valuable against resistant gram-negative infections, and exactly what makes carbapenem resistance so alarming.
The Global Scale of the Problem
Antimicrobial resistance is not a future threat. A landmark 2019 analysis estimated that bacterial AMR was directly responsible for about 1.27 million deaths globally that year, with nearly five million more deaths associated with resistant infections.5PubMed Central. Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis Among the deadliest pathogen-drug combinations were carbapenem-resistant Acinetobacter baumannii and carbapenem-resistant Klebsiella pneumoniae, each responsible for between 50,000 and 100,000 attributable deaths.5PubMed Central. Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis
The burden falls unevenly. A systematic analysis of carbapenem-resistant gram-negative infections found that nearly 480,000 related deaths occurred across just eight selected countries in 2019, with the overwhelming majority concentrated in South Asia. India alone accounted for over a million estimated infections requiring WHO Reserve antibiotics that year.6The Lancet Infectious Diseases. Antibiotics of Last Resort: A Final Line of Defense The organisms driving these numbers, A. baumannii, K. pneumoniae, and Pseudomonas aeruginosa, are the same pathogens that sit at the top of the WHO’s priority list for new antibiotic development.7PubMed. Global and regional burden of bloodstream infections caused by carbapenem-resistant Gram-negative bacteria in 2019
Why Last-Resort Drugs Are Hard on Patients
There is a reason these drugs were shelved for decades. Colistin, the most widely used polymyxin, causes kidney damage in a substantial share of patients who receive it. For years, clinicians assumed this nephrotoxicity was simply the drug’s detergent-like action stripping cell membranes inside the kidney tubules. More recent research paints a more complicated picture: the drug accumulates inside kidney cells and appears to damage mitochondria as a primary site of injury.8PubMed Central. Molecular Mechanisms of Colistin-Induced Nephrotoxicity
Animal studies have confirmed that colistin triggers kidney injury through multiple overlapping pathways, including mitochondrial dysfunction, activation of cell-death signaling cascades, and stress in the endoplasmic reticulum. After seven days of colistin treatment in mice, markers of kidney damage like blood urea nitrogen and serum creatinine climbed significantly, while protective antioxidant enzymes in the kidneys declined.9PubMed Central. Colistin-induced nephrotoxicity in mice involves the mitochondrial, death receptor, and endoplasmic reticulum pathways A newer line of research has identified another mechanism: colistin causes an overload of zinc ions inside kidney cells, which triggers a form of inflammatory cell death called pyroptosis. In mouse experiments, chelating the excess zinc reduced the kidney damage.10PubMed. Colistin-induced acute kidney injury via zinc Dyshomeostasis-triggered GSDMD-executed Pyroptosis That finding hints at possible ways to protect the kidneys during colistin therapy, but none have reached routine clinical practice yet.
The toxicity of polymyxins creates a grim calculus at the bedside. Doctors must weigh certain kidney damage against the near-certainty that without colistin, a pan-resistant infection will kill the patient. This is not a comfortable trade-off, and it underscores why the medical community is desperate for better alternatives.
How Bacteria Are Breaching the Last Defenses
Resistance to last-resort antibiotics spreads through two broad channels. Bacteria can mutate their own chromosomes to modify the drug’s target, or they can acquire resistance genes from other bacteria through mobile genetic elements like plasmids. The second route is far more alarming, because a single resistance gene on a transferable plasmid can jump between species, turning previously treatable bacteria into untreatable ones overnight.
For colistin, the game-changer was the discovery in 2015 of the mcr-1 gene, which encodes an enzyme that modifies lipid A so that polymyxins can no longer bind to it. Before mcr-1, colistin resistance had to evolve independently inside each bacterial lineage. The mcr-1 gene, sitting on a plasmid, can be passed horizontally between species, and it has already been found in Escherichia coli, Salmonella, and other members of the Enterobacteriaceae family.11PubMed Central. Characteristics of the plasmid-mediated colistin-resistance gene mcr-1 in Escherichia coli isolated from a veterinary hospital in Shanghai Research on companion-animal E. coli isolates has demonstrated that plasmid transfer occurred in all tested strains, raising the possibility that colistin resistance could pass from pets to people.11PubMed Central. Characteristics of the plasmid-mediated colistin-resistance gene mcr-1 in Escherichia coli isolated from a veterinary hospital in Shanghai
For carbapenems, the threat comes from carbapenemases, enzymes that chew apart the drug before it can reach its target. These enzymes fall into several classes depending on whether they use zinc to function. They are among the most versatile beta-lactamases known, capable of destroying not just carbapenems but many other beta-lactam antibiotics as well.12PubMed Central. Structural basis for carbapenem-hydrolyzing mechanisms of carbapenemases conferring antibiotic resistance Carbapenemase genes also travel on plasmids, and healthcare settings have become hotspots for their spread.
When Resistance Genes Stick Around
One question researchers grapple with is whether resistance genes carry a fitness cost heavy enough that bacteria will shed them once the antibiotic pressure disappears. The answer, frustratingly, is often no. Carrying mcr-1 or mcr-3 does impose some burden on the bacterium, but compensatory mutations can evolve to offset that cost. Studies tracking the evolution of mcr-3 variants in E. coli have found that later versions of the gene have already accumulated compensatory changes, making resistant bacteria nearly as fit as their sensitive counterparts.13The ISME Journal. Compensatory mutations modulate the competitiveness and dynamics of plasmid-mediated colistin resistance in Escherichia coli clones
The picture is more nuanced for different mcr variants. Experimental comparisons using identical bacterial backgrounds show that mcr-1 and mcr-3 impose relatively low fitness costs, while mcr-9, a related gene, only provides a survival advantage when some colistin is present. Without the drug, mcr-9 actually reduces bacterial fitness significantly.14PubMed Central. Site-selective modifications by lipid A phosphoethanolamine transferases linked to colistin resistance and bacterial fitness This means that some resistance genes are more “sticky” than others. The ones with the lowest fitness costs are the hardest to purge, and unfortunately, those include the most clinically relevant variants.
Newer Drug Combinations and Timing
The most significant recent addition to the last-resort toolbox is ceftazidime-avibactam, a combination of a cephalosporin antibiotic with a beta-lactamase inhibitor that blocks many of the enzymes bacteria use to destroy beta-lactam drugs. A systematic review of its use against serious gram-negative infections found clinical success rates ranging from 45% to 100% across published studies, with the wide spread reflecting differences in patient severity and infection types.15PubMed Central. Ceftazidime-Avibactam for the Treatment of Serious Gram-Negative Infections with Limited Treatment Options: A Systematic Literature Review
A meta-analysis comparing ceftazidime-avibactam to other treatments for gram-negative infections found no statistically significant difference in clinical success overall, though there was a trend toward better microbiological clearance.16PubMed Central. Efficacy of Ceftazidime-avibactam in treating Gram-negative infections: a systematic review and meta-analysis Where the drug seems to matter most is in patients who have no other active antibiotic to fall back on. A large real-world study across 22 U.S. medical centers found that in patients who had not received any other active antibiotic, starting ceftazidime-avibactam within about 42 hours of diagnosis was linked to higher clinical success (roughly 73% vs. 63%) and lower 30-day mortality (about 11% vs. 19%).17PubMed Central. Ceftazidime-avibactam for multidrug-resistant gram-negative infections: outcomes and timing of initiation across 22 U.S. medical centers The practical takeaway: when the bug is resistant to everything else, getting the right last-resort drug started fast is the single biggest lever clinicians have.
Detecting Resistance Before It Is Too Late
One of the underappreciated bottlenecks in treating resistant infections is simply identifying them. Traditional culture-and-sensitivity testing can take 48 to 72 hours. During that window, a patient with a carbapenem-resistant bacterium may be receiving a carbapenem that does nothing, wasting precious time. Rapid diagnostic tests are starting to close this gap. An evaluation of two approaches, a lateral flow assay and a PCR-based molecular test, demonstrated that molecular testing can detect carbapenemase genes directly from primary patient samples without waiting for a cultured isolate.18PubMed Central. Rapid detection of carbapenemases in multiresistant Gram-negative strains: evaluation of two tests Knowing on day one whether you are dealing with a carbapenemase-producing organism lets the clinical team skip ineffective drugs and reach for the right reserve antibiotic immediately.
This speed advantage feeds directly into the timing data for ceftazidime-avibactam: if a rapid test can identify carbapenem resistance within hours rather than days, clinicians can start appropriate treatment well within that 42-hour window where outcomes are meaningfully better. The diagnostic and the drug work as a system, and neither is as useful without the other.
Beyond Antibiotics Altogether
When bacteria evolve past every available antibiotic, some clinicians have turned to phage therapy, using viruses that naturally infect and kill bacteria. Phages have a century of history in Eastern European medicine but remain largely experimental in the West. Their appeal for pan-resistant infections is straightforward: phages target bacteria through mechanisms completely unrelated to antibiotic chemistry, so antibiotic resistance genes are irrelevant.
Recent compassionate-use case series offer early encouraging data. At the Israeli Phage Therapy Center, 18 patients with persistent infections who had failed conventional antibiotics were treated with phages; 14 achieved clinical remission, a success rate of 78%.19Open Forum Infectious Diseases. Compassionate Use of Bacteriophages for Failed Persistent Infections During the First 5 Years of the Israeli Phage Therapy Center A separate series using a phage called PASA16 against drug-resistant Pseudomonas aeruginosa reported cure or remission in 13 of 15 patients, with the phage well tolerated throughout treatment.20Med. Refractory Pseudomonas aeruginosa infections treated with phage PASA16: A compassionate use case series Two patients with cystic fibrosis and pan-drug-resistant Pseudomonas have also been treated with phages as a novel therapeutic strategy when no antibiotic options remained.21PubMed Central. Bacteriophage Therapy for Pan-Drug-Resistant Pseudomonas aeruginosa in Two Persons With Cystic Fibrosis
These numbers are promising, but the sample sizes are tiny and the patients are highly selected. There are no large randomized trials, and phage therapy faces practical hurdles including the need to match each phage to the patient’s specific bacterial strain. It is not yet a reliable replacement for last-resort antibiotics, but it is the closest thing to a genuinely novel approach the field has produced in years.
Other experimental strategies aim to disarm bacteria rather than kill them. Anti-virulence therapies block the molecular tools bacteria use to cause disease, such as adhesins that attach to host cells, toxins that destroy tissue, and communication systems that coordinate attacks. Because these therapies do not directly kill bacteria, they may generate less selection pressure for resistance.22PubMed Central. Anti-virulence therapeutic strategies against bacterial infections: recent advances Antimicrobial peptides, short proteins that punch holes in bacterial membranes, are also being explored. Some show therapeutic indexes between 50 and 500, meaning they are far more toxic to bacteria than to human cells, though translating that selectivity from the lab to the clinic has been slow.
A Broken Pipeline and Experimental Fixes
The antibiotics pipeline is in trouble because the economics work against it. A last-resort antibiotic is, by design, one you use as little as possible. That means low sales volume, which means pharmaceutical companies cannot recoup the investment of bringing a drug to market. Several new antibiotics have won regulatory approval in recent years, but almost none feature new chemical classes or genuinely novel mechanisms.23PubMed Central. Challenges and opportunities for incentivising antibiotic research and development in Europe The pipeline is producing incremental modifications of existing drug classes, not the breakthroughs needed to stay ahead of evolving resistance.
To address this market failure, several countries are experimenting with new payment models. The United Kingdom launched what has been called a “Netflix” model for antibiotics in July 2022, paying manufacturers a fixed annual fee for access to a drug regardless of how many doses are actually used. This delinks payment from sales volume, removing the perverse incentive to promote heavy use of the very drugs that should be conserved.24PubMed Central. Can the UK ‘Netflix’ Payment Model Boost the Antibacterial Pipeline? Other proposed incentives include lump-sum market entry rewards, transferable patent exclusivity extensions, and milestone payments tied to development progress.23PubMed Central. Challenges and opportunities for incentivising antibiotic research and development in Europe None of these have yet been proven to substantially reinvigorate the pipeline, but the UK model is the farthest along and is being watched internationally as a template.
Fighting Resistance Genes Directly
One of the more audacious research directions is using gene-editing technology to strip resistance genes out of bacteria. A proof-of-concept study used CRISPR-Cas9 to target and eliminate mcr-1-carrying plasmids in E. coli, restoring the bacteria’s sensitivity to colistin. The engineered system also prevented the transfer of mcr-1 via conjugation, the process by which bacteria pass plasmids to their neighbors.25PubMed Central. Reversal of mcr-1-Mediated Colistin Resistance in Escherichia coli by CRISPR-Cas9 System The idea of deploying CRISPR against resistance genes in a clinical setting remains speculative. Delivering the editing machinery to bacteria inside a living patient is an unsolved problem. But as a concept, it represents a fundamentally different way of thinking about resistance: instead of finding new drugs to outrun evolution, you edit evolution’s toolkit directly.
Whether this kind of approach can scale is an open question. Bacteria evolve fast, and a CRISPR system targeting one resistance gene does nothing about the dozens of others circulating on plasmids worldwide. Still, in a field where incremental improvements to existing drug classes have been the norm, the idea of actively removing resistance has a certain appeal. It reframes the arms race between medicine and microbes, shifting from purely defensive to something closer to offense.