ESKAPE pathogens: antimicrobial resistance, epidemiology, clinical impact and therapeutics

Abstract

The rise of antibiotic resistance and a dwindling antimicrobial pipeline have been recognized as emerging threats to public health. The ESKAPE pathogens — Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp. — were initially identified as critical multidrug-resistant bacteria for which effective therapies were rapidly needed. Now, entering the third decade of the twenty-first century, and despite the introduction of several new antibiotics and antibiotic adjuvants, such as novel β-lactamase inhibitors, these organisms continue to represent major therapeutic challenges. These bacteria share several key biological features, including adaptations for survival in the modern health-care setting, diverse methods for acquiring resistance determinants and the dissemination of successful high-risk clones around the world. With the advent of next-generation sequencing, novel tools to track and combat the spread of these organisms have rapidly evolved, as well as renewed interest in non-traditional antibiotic approaches. In this Review, we explore the current epidemiology and clinical impact of this important group of bacterial pathogens and discuss relevant mechanisms of resistance to recently introduced antibiotics that affect their use in clinical settings. Furthermore, we discuss emerging therapeutic strategies needed for effective patient care in the era of widespread antimicrobial resistance.

Introduction

Antibiotics have changed the history of medicine and made possible advances in surgery, organ transplantation and cancer chemotherapy, amongst many other fields. However, the emergence of resistance to these life-saving drugs has been a major concern¹. During the golden age of antibiotic discovery, the relative proliferation of new classes of antibiotics in a relatively short span of time seemed to promise an unending number of new antimicrobial compounds². Over time, however, the number of truly new classes of antibiotics began to dwindle, while rates of antimicrobial resistance steadily climbed. As a result, antimicrobial resistance is now a tangible threat to the day-to-day practice of medicine across the globe³.

A large proportion of the resistant infections identified in clinical medicine can be attributed to a specific group of pathogens that includes Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and other members of the Enterobacterales group, called the ‘ESKAPE’ organisms⁴ (Table 1). These organisms share a common set of characteristics, namely the ability to thrive in the modern health-care environment, and possess a repertoire of intrinsic or acquired resistance determinants that have enabled them to be a prominent cause of resistant infections over time (Fig. 1). Recent reviews have provided an in-depth discussion of the mechanisms of resistance among ESKAPE pathogens to most traditional antimicrobial classes⁵. In this Review, we summarize the numerous recent advances in our understanding of the molecular epidemiology, clinical impact, emerging mechanisms of resistance and novel therapeutic approaches targeting ESKAPE organisms.

Table 1 |.

Overview of ESKAPE pathogens

Organism	Major clinical syndromes	High-risk clonal lineages	Possible treatment options for serious infections	Major resistance determinants
Vancomycin-resistant Enterococcus faecium	Bloodstream infection, infective endocarditis, intra-abdominal infection, UTI	Hospital-associated lineages (ST17, ST80, ST117, ST412, ST584, ST664, ST736)	Daptomycin, linezolid, tigecycline, eravacycline, omadacycline, oritavancin	β-Lactams (low-affinity PBP5) Vancomycin (vanA, Tn1546; vanB, Tn1549) Daptomycin (liaFSR, cls) Linezolid (23S rDNA mutations; cfr, optrA, poxtA)
Methicillin-resistant Staphylococcus aureus	Bloodstream infection, infective endocarditis, ABSSSI, CAP, HAP/VAP, bone and joint infection	CC5, CC8, CC22, CC30, CC45	Vancomycin, daptomycin, linezolid, ceftaroline, ceftobiprole, trimethoprim–sulfamethoxazole, delafloxacin, oritavancin, dalbavancin, telavancin	β-Lactams (mecA, SCCmec) Vancomycin (vanA, Tn1546 (exceptional)); VISA: vraRS, graRS, walKR) Daptomycin (vraRS, graRS, walKR, mprF, dlt, rpoB) Linezolid (23S rDNA gene mutations, ribosomal protein substitutions, cfr, optrA, poxtA) Ceftaroline (PBP2a mutations) Delafloxacin (changes in ParC)
Carbapenem-resistant or ESBL-producing Klebsiella pneumoniae	Bloodstream infection, UTI, CAP, HAP/VAP, intra-abdominal infection	CRE (CG258, CG15, CG20, CG29, CG37) ESBL (CG101, CG147, CG307)	Ceftazidime–avibactam, meropenem–vaborbactam, imipenem–cilastatin–relebactam, cefiderocol, tigecycline, eravacycline, polymyxin B	β-Lactamases (ESBL: TEM, SHV, CTX-M; AmpC: CMY; carbapenemase: KPC, OXA-48-like, NDM, VIM, IMP) Porin loss (ompK35, ompK36) Efflux pumps (AcrAB–TolC, OqxAB) Aminoglycoside modifying enzymes and 16S rRNA methyltransferases (AAC(6′)-Ib, APH(3′)-Ia, AAC(3)-IV, ANT(2″)-Ia) RmtB (16S rRNA methyltransferase)
Carbapenem-resistant Acinetobacter baumannii	HAP/VAP, bloodstream infection, UTI	CC1, CC2, CC3	Sulbactam–durlobactam, cefiderocol, minocycline, tigecycline, eravacycline, polymyxin B	β-Lactamases (AmpC: ADC (intrinsic); ESBL: PER, GES, VEB; carbapenemase: OXA-23, OXA-24/40, OXA-58, OXA-72, NDM) Porin loss (carO, occAB1, omp33–36) Efflux pumps (AdeABC) Aminoglycoside modifying enzymes and 16S rRNA methyltransferases (AadA2, AadB, AphA6, AAC(6′)-Ib, ArmA (16S rRNA methyltransferase))
Carbapenem-resistant Pseudomonas aeruginosa	HAP/VAP, bloodstream infection, UTI	ST235, ST111, ST175	Ceftolozane–tazobactam, ceftazidime–avibactam, imipenem–cilastatin–relebactam, cefiderocol, polymyxin B	β-Lactamases (AmpC: PDC (intrinsic); ESBL: PER, GES, VEB; carbapenemase: KPC, GES, VIM, NDM, IMP) Porin loss (oprD) Efflux pumps (MexAB–OprM, MexCD–OprJ, MexEF–OprN, MexXY–OprM) Aminoglycoside modifying enzymes (AAC(6′)-Ib, ANT(2′)-I, APH(3′)-Ib)
Carbapenem-resistant or ESBL-producing Enterobacter cloacae complex	Bloodstream infection, UTI, HAP/VAP, intra-abdominal infection	Enterobacter xiangfengensis (ST114, ST171) Enterobacter hormaechei subsp. steigerwaltii (ST90, ST93) E. cloacae cluster III (ST78)	Ceftazidime–avibactam, meropenem–vaborbactam, imipenem–cilastatin–relebactam, cefiderocol, tigecycline, eravacycline, polymyxin B	β-Lactamases (ESBL: AmpC (intrinsic), TEM, SHV, CTX-M; AmpC: CMY; carbapenemase: KPC, OXA-48-like, NDM, VIM, IMP) Porin loss (ompC, ompF) Efflux pumps (AcrAB–TolC) Aminoglycoside modifying enzymes (AAC(6′)-I, ANT(2″), rmtB, armA)

ABSSSI, acute bacterial skin and skin structure infection; CAP, community-acquired pneumonia; CC, clonal complex; CG, clonal group; CRE, carbapenem-resistant Enterobacterales; ESBL, extended-spectrum β-lactamase; ESKAPE pathogens, Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp. HAP/VAP, health-care-associated pneumonia/ventilator-associated pneumonia; PBP, penicillin-binding protein; PDC, Pseudomonas-derived cephalosporinase; SCCmec, staphylococcal chromosomal cassette mec; ST, sequence type; Tn, transposon; UTI, urinary tract infection; VISA, vancomycin intermediate Staphylococcus aureus.

Fig. 1 |. Burden of antimicrobial resistance in ESKAPE pathogens.

a, Global incidence of antimicrobial resistance in ESKAPE organisms (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp.) from 1997 to 2022. b, Estimated global attributable mortality due to antimicrobial-resistant ESKAPE pathogens in 2019. Data compiled from refs. 24,25,45,109,210,249–251. CRAB, carbapenem-resistant Acinetobacter baumannii; CRE, carbapenem-resistant Enterobacterales; CR-PA, carbapenem-resistant Pseudomonas aeruginosa; MRSA, methicillin-resistant Staphylococcus aureus; VRE, vancomycin-resistant enterococci.

Gram-positive organisms

Staphylococcus aureus

S. aureus is a major human pathogen causing various infections, including skin and soft tissue, osteoarticular and bloodstream infections, pneumonia, infective endocarditis and device-related infections⁶. Although β-lactam antibiotics are the therapy of choice for methicillin-susceptible Staphylococcus aureus (MSSA) infections, methicillin-resistant Staphylococcus aureus (MRSA) exhibits resistance to most β-lactams, including anti-staphylococcal penicillins (for example, nafcillin, oxacillin and flucoxacillin) and cephalosporins (with the notable exception of ceftaroline and ceftobiprole). Methicillin resistance is mediated by acquisition of the staphylococcal chromosome cassette mec (SCCmec) element, likely from strains of coagulase-negative staphylococci⁷. Analysis of MRSA from across temporal and geographic locations suggests that methicillin resistance has emerged multiple times, with acquisition of SCCmec elements occurring independently both among different genetic lineages and within each sequence type (ST)⁸.

Molecular epidemiology.

The molecular epidemiology of S. aureus, and in particular MRSA, has been the subject of intensive investigation. The tools to track the emergence of evolution in these lineages have progressed in tandem with their spread (Box 1). S. aureus lineages can be broadly grouped into clonal clusters of related isolates, with nearly 90% of human derived isolates belonging to 1 of 11 clonal complexes (CCs), including the major lineages CC5, CC8, CC22, CC30 and CC45 (ref. 9). Importantly, geographic differences of circulating lineages are prominent with predominant circulating clones of both MSSA and MRSA, and shift over time, a phenomenon known as clonal replacement¹⁰.

Box 1. Tracking an epidemic of mobile resistance.

Bacterial whole-genome sequencing has been extremely useful to characterize the epidemiology of multidrug-resistant organisms, identify and investigate outbreaks and collect data for public health purposes. The standard approaches to dissect the population structure of multidrug-resistant organisms rely on defining the core (conserved) regions of the bacterial population to perform phylogenetic analyses. However, important areas of the bacterial genome are often overlooked with these analyses; in particular, the repertoire of acquired mobile genetic elements (MGEs), often referred as the ‘mobilome’²⁵². These MGEs include plasmids, bacteriophages, genomic islands, integrative and conjugative elements, insertion sequences, transposons (Tns), integrons and miniature inverted repeat transposable elements.

These MGEs play important roles in bacterial evolution through horizontal gene transfer, determining major antibiotic resistance phenotypes and contributing to the dissemination of resistance genes among bacterial communities, including those residing in the human microbiome. Therefore, tracking MGEs in the ‘accessory genome’ (defined as elements that are not encompassed by the core genome) becomes of paramount importance to understand the dynamics of colonization and infection caused by multidrug-resistant organisms. However, MGEs may be challenging to reconstruct using standard techniques, making it difficult to properly differentiate between temporary acquisition of resistance determinants and a more ‘stable’ incorporation of such traits into specific bacterial lineages. Furthermore, when analysing microbial communities, identifying the bacterial species that contain these MGEs and how these MGEs flow among the bacterial microbiota taxa is challenging. The development and refinement of long-read sequencing technologies, capable of generating sequencing reads that span the entirety of these complex genetic structures, has permitted a more precise reconstruction of complex MGEs, especially those carrying multiple antibiotic resistance determinants such as integrons (genetic elements that contain a site-specific recombination system able to integrate, express and exchange specific DNA)²⁵³. As the long-read sequencing technologies still lack the accuracy of short-read approaches to detect single-nucleotide polymorphisms (SNPs), hybrid assemblies are often generated using short-read sequencing to generate closed reference genomes²⁵⁴.

The above approaches are also important in the field of metagenomics in order to track MGEs in entire microbial communities present in diverse environments (both human and non-human)²⁵⁵. Both 16S rRNA gene sequencing (used for taxonomic classification) and whole-genome shotgun sequencing (where DNA molecules of bacterial communities are fragmented and sequenced) are used to identify the members of microbial communities and determine the exact location and, potentially, the flow and relative abundance of MGEs in members of the community²⁵⁶. Although refinements of these approaches are still developing, novel computational tools are becoming available to deal with contamination and gene linkage, making them powerful tools to track MGEs associated with antibiotic resistance.

The first described MRSA isolates were identified in 1961 and belonged to phage group III or 83A, marking a shift from the phage group 80/81 penicillinase-producing MSSA¹¹. These early MRSA lineages were members of ST250 in CC8, harboured SCCmecI and were limited to the hospital environment. In the United States, a sharp increase in the prevalence of MRSA was observed in the 1980s with nearly 30% of isolates resistant to methicillin by 1991 (ref. 12). This phenomenon coincided with the decrease of the early archaic clones in European hospitals and the emergence of hospital-associated MRSA lineages⁹. These hospital-associated MRSA lineages were more likely to carry SCCmecI–III, were associated with a multidrug-resistant phenotype due to the carriage of additional resistance determinants and were not linked to significant rates of community infection¹³.

In the late 1990s, there was a major shift in the epidemiology of MRSA from hospital to community settings reported initially in Western Australia and the United States^14,15. These community-associated MRSA strains were characterized by acquisition of SCCmecIV and, in some cases, the Panton Valentine leucocidin (PVL)¹⁶. In the Americas, community-associated MRSA infections came to be dominated by the ST8 USA300 clone (CC8). The likely ancestor for USA300 was a European ST8 MSSA that established itself in the United States in the early twentieth century¹⁷. This ancestral clone acquired the hallmarks of USA300, including the presence of the ΦSa2USA phage with the gene encoding PVL, the arginine catabolic mobile element (ACME) and SCCmecIV, as well as fluoroquinolone resistance as the strain migrated progressively south and west. In South America, a related clone, USA300-LV (for Latin American variant), emerged in the mid-2000s. The USA300-LV strains independently acquired a subvariant of SCCmecIV, and possess the copper and mercury resistance mobile element (COMER) instead of ACME, suggesting that these two epidemics of community-associated MRSA emerged in parallel from a common ancestor¹⁸. USA300 with spa type t121 has also established itself in Western Africa, where ongoing evolution and emergence of mutations unique to Africa may be aiding its spread¹⁷.

Outside the Americas, the population of community-associated MRSA is more diverse. ST80 (CC80) is predominant in Europe and likely emerged from a PVL-positive MSSA isolate from sub-Saharan Africa that spread to North Africa and across the Mediterranean^19,20. A livestock-associated lineage (designated CC398) has also been associated with an increasing number of MRSA infections²¹. Lineages of CC398 have spread globally, and severe MRSA infections due to CC398 without association with livestock have been reported, including a livestock-independent lineage within CC398 defined by spa type t571 (refs. 22,23). Nonetheless, a consistent decline in the rates of MRSA infections has been observed worldwide in both hospital-acquired and community-associated infections^24,25 (Fig. 1a).

Clinical impact of MRSA infections.

Infections due to MRSA are estimated to result in 10,600 deaths and a cost of more than US$1.7 billion annually in the United States²⁶. Although MRSA infections have generally been associated with higher hospital costs and increased mortality than those due to MSSA, the data since 2010 suggest this trend may be reversing in some areas²⁷. Rates of resistance to non-β-lactam antibiotics are higher in MRSA as compared with MSSA isolates, and some lineages, such as the ST5 Chilean-Cordobés clone, have also demonstrated decreased susceptibility to ceftaroline^25,28. Although β-lactam resistance mediated by the presence of penicillin-binding protein PBP2a, the product of the mecA gene, is the most common determinant encountered clinically, additional mec genes (mecB or mecC) and S. aureus isolates with the borderline oxacillin-resistant phenotype (BORSA) pose a challenge as they are not readily identified by many rapid diagnostics^29,30. High-level vancomycin resistance remains rare, and although some studies have suggested increases in heterogeneous vancomycin intermediate S. aureus (VISA), the difficulty in testing this phenotype outside research laboratories is likely to skew accurate assessment of any trends³¹.

Newer antimicrobial agents such as daptomycin, linezolid, tetracycline derivatives (for example, tigecyline and eravacycline), ceftaroline and ceftobiprole, and the latest generation of glycopeptides (for example, dalbavancin and oritavancin) generally retain robust activity²⁵. Some clinical data support using dalbavancin or oritavancin in bacteraemia, orthopaedic infections or infective endocarditis, including a successful clinical trial in catheter-associated bacteraemia, although emerging resistance to some of these agents is problematic^32,33 (Fig. 2). Daptomycin is a cationic lipopeptide antibiotic that binds to the membrane lipid phosphotidylglycerol (PG) and complexes with lipid II and its undecaprenyl intermediates to disrupt cell envelope biogenesis³⁴. Resistance to daptomycin arises from modification of the cell envelope, altering the cell surface charge and the availability of PG. At a molecular level, activation of the cell envelope signalling systems VraRS and GraRS upregulate expression of MprF (an enzyme that adds lysine to PG) and the dlt operon (leading to d-alanylation of cell wall teichoic acids), both of which add positive charge to the cell surface³⁵. In addition, rpoB gene mutations (encoding the β-subunit of RNA polymerase) have been associated with changes in daptomycin susceptibility, although the specific mechanism has not been elucidated³⁶. The oxazolidinones linezolid and tedizolid act by binding at the A site of the bacterial ribosome and altering the peptidyltransferase site of the ribosome. Chromosomal mutations in genes encoding 23S rRNA and ribosomal proteins, as well as several transmissible resistance determinants, including the methyltransferase cfr and the ribosomal protection factors optrA and poxtA, have been described in staphylococcal species, especially those of animal origin³⁷. Ceftaroline and ceftobiprole are cephalosporins with activity against MRSA, and ceftobiprole was non-inferior to daptomycin in a randomized controlled trial evaluating the treatment of complicated S. aureus bacteraemia³⁸. Resistance to these agents has been primarily associated with changes in PBP2a, but may also require additional mutations in pbp2, pbp4 and gdpP³⁹. Resistance to the fluoroquinolone delafloxacin, which has activity against MRSA, seems to arise through a combination of mutations in gyrA and gyrB (encoding DNA gyrase), or parC and parE (encoding topoisomerase IV), and intrinsic or acquired efflux pumps⁴⁰.

Fig. 2 |. Emerging mechanisms of resistance in Gram-positive pathogens.

A network of cell envelope stress response systems mediates resistance to lipopeptides (daptomycin) and glycopeptides (vancomycin). In Staphylococcus aureus, activation of the histidine kinase/response regulators VraSR and GraSR leads to upregulation of the dlt operon and mprF, each of which modulate the amount of positive charge of the cell wall and membrane. These pathways have been implicated in both the vancomycin intermediate phenotype (vancomycin intermediate Staphylococcus aureus (VISA)) and daptomycin resistance. In enterococci, vancomycin resistance is primarily due to the production of altered peptidoglycan precursors by the machinery encoded in the van operon. Daptomycin resistance in enterococci is associated with activation of the LiaFSR system, which along with changes in expression of and mutations in cls, the gene encoding cardiolipin synthase (Cls), leads to alterations in membrane phospholipids and protects the cell division machinery from disruption. Oxazolidinone (linezolid) resistance occurs through mutations in the genes encoding rRNA, ribosomal proteins or with the acquisition of a plasmid-encoded methyltransferase (cfr) or ribosomal protection factor (optrA and poxtA). Various drug efflux pumps are associated with resistance to tetracyclines and fluoroquinolones. Resistance to β-lactams occurs via production of altered penicillin-binding proteins (PBPs), which have a decreased affinity for nearly all β-lactams.

Despite the availability of agents with in vitro activity against MRSA, these infections remain challenging and the optimal therapeutic approach has not been established. Although vancomycin continues to be the likely first-choice therapy for most infections, several issues including resistance, pharmacological optimization and toxicities have prompted a re-evaluation of this antibiotic. Nonetheless, clinical data supporting the use of alternative agents have not provided robust and consistent evidence to fully replace vancomycin as the workhorse antibiotic for MRSA infections. Another ‘hot topic’ is the use of combination therapies against MRSA. In particular, the combination of vancomycin or daptomycin plus β-lactams has substantial in vitro and preclinical data. The rationale for the effectiveness of the combination seems to depend on the ‘see-saw effect’ phenomenon whereby the emergence of non-susceptibility to vancomycin or daptomycin appears to increase the susceptibility to β-lactams⁴¹. However, the clinical efficacy of combination regimens has proven difficult to ascertain in randomized clinical trials⁴².

Enterococci

Enterococci are commensals of the human gastrointestinal microbiota with decreased susceptibility to penicillin and intrinsic resistance to cephalosporins⁴³. Most Enterococcus faecalis isolates retain susceptibility to ampicillin with low rates of vancomycin resistance (5–10%)⁴⁴. By contrast, health-care-associated isolates of E. faecium are generally ampicillin-resistant, with the frequency of vancomycin resistance ranging from 0.3–3% in Western Europe and Scandinavia to 30–60% in parts of Southern and Eastern Europe, and to 50–80% in the United States and Latin America^24,45. Enterococci display a genomic plasticity that makes them very adaptable to antimicrobial challenge. Multidrug-resistant clinical isolates often lack a functional CRISPR–Cas system, which impairs bacterial defence against foreign DNA and leads to higher rates of assimilation of mobile elements harbouring resistance determinants⁴⁶.

Molecular epidemiology of Enterococcus faecium.

E. faecium constitutes the vast majority of clinical vancomycin-resistant enterococci (VRE) isolates and can be broadly divided into two clades, a health-care-associated clade, known as clade A, and a commensal clade B^47,48. Early genomic studies suggested a further division of clade A, with the A1 branch enriched for hospital isolates that displayed greater infectivity and higher rates of resistance⁴⁸. The A2 branch predominantly held isolates of animal origin, and a molecular clock analysis suggested a divergence at the start of the antibiotic era. Subsequent studies, with a more diverse set of isolates, including from the United Kingdom and Latin America, demonstrated that rather than being a distinct group, clade A2 was likely a series of paraphyletic branches on the evolutionary path of clade A and that divergence between these branches occurred some 300–500 years ago^49,50.

A large evaluation of hospital E. faecium from Europe and Australia supports the ancestral nature of clade A2 and suggests that clade A1 strains emerged from this group to dominate the health-care environment⁵¹. Evolution appears driven by major recombination events between clades including regions containing the stress response system liaFSR (involved in envelope defence and daptomycin resistance) and the capsular polysaccharide locus cps (important for immune evasion). The introduction of these and other alleles into the A1 population structure has fuelled their dominance in the health-care environment^51–53. The adaptation and emergence of resistance in clade A1 E. faecium have also been linked to plasmids and other mobile genetic elements (MGEs), which make up the accessory genome^54,55. An analysis of more than 1,500 E. faecium genomes demonstrated that hospital isolates are more likely to have a higher cumulative plasmid length (that is, larger plasmids or more plasmids per strain). Furthermore, these plasmids clustered by isolation source, such as hospital, animal or community origin, and were notable for differences in antibiotic resistance genes and metabolic genes, which help to give their host strain a competitive growth advantage⁵⁶.

Clinical impact.

VRE are a significant clinical challenge due to the patient populations who develop infection and the limited treatment options available. Risk factors for VRE bloodstream infections include cancer or haematologic malignancy, organ transplantation (especially liver transplant), prolonged hospitalization or admission from long-term care facilities, gastrointestinal tract surgeries or procedures and exposure to broad-spectrum antimicrobials^57,58. Once infection develops, there is substantial controversy regarding the outcomes of infections with vancomycin-susceptible versus vancomycin-resistant enterococcal isolates^59–62. Most of the studies are retrospective and subject to the influence of major confounders. In a recent cohort study of patients with enterococcal bacteraemia from the United States, infection with vancomycin-resistant E. faecium was associated with a 3.5-fold increase in mortality, and overall, VRE of any species were associated with recurrent bacteraemia and a failure to achieve clear blood cultures within 4 days⁶³.

Optimal therapy for invasive VRE infections remains uncertain. Clinical data comparing daptomycin with linezolid are largely observational with mixed results, although high-dose daptomycin (10–12 mg kg^–1) seems to improve clinical outcomes⁶⁴. Daptomycin is less potent against enterococci than staphylococci, with E. faecium displaying a several-fold increase in the minimum inhibitory concentration distribution. High-dose daptomycin increases the probability of target attainment in pharmacokinetic/pharmacodynamic modelling studies and may suppress the emergence of resistance⁶⁵. The emergence of daptomycin resistance remains an ongoing concern, especially in patients with haematologic malignancy and patients with liver transplant^58,66. Daptomycin resistance generally arises through activation of the LiaFSR cell envelope stress response system in conjunction with changes in cell membrane phospholipid metabolism⁵³ (Fig. 2). Several clade A1 lineages of E. faecium in the United States and Australia have LiaFSR polymorphisms associated with resistance, including ST80, ST664 and ST736, and there is evidence that these lineages have expanded in the decade since the introduction of this antibiotic^63,67. Similar to S. aureus, mutations in rpoB and dltC have been associated with daptomycin-resistant E. faecium belonging to ST203 (ref. 68).

Linezolid resistance in enterococci occurs mainly by two mechanisms. The first involves mutations in the intrinsic genes encoding the 23S rRNA, thus altering the oxazolidinone binding site, coupled with mutations of the L3 and L4 ribosomal proteins, which likely play a role in decreasing the fitness cost of rRNA mutations^69,70. As enterococci possess multiple copies of the 23S rRNA gene, recombination between mutated alleles can lead to rapid increases in the linezolid minimum inhibitory concentration⁷¹. As with S. aureus, plasmids and MGEs located in the chromosome carrying the cfr, optrA and poxtA genes can also lead to oxazolidinone resistance in enterococci. Newer generation tetracycline antibiotics, such as tigecycline, omadacycline and eravacycline, act by binding at the 16S rRNA of the 30S ribosomal subunit and retain activity against VRE^72,73. These agents may have a role in intra-abdominal infections or as a part of combination therapy. In vitro experimental evolution demonstrated the emergence of resistance associated with mutations in rpsJ (encoding the S10 ribosomal protein) and transposon (Tn)-mediated gene copy number expansion of the tet(L) efflux pump, with similar mutations in rpsJ and overexpression of both tet(L) and tet(M) described in resistant clinical isolates^74–76.

Gram-negative organisms

Enterobacterales

Several ESKAPE pathogens belong to the order Enterobacterales, including K. pneumoniae, Klebsiella aerogenes (formerly Enterobacter aerogenes) and Enterobacter cloacae⁷⁷. Escherichia coli, although not an original member of the ESKAPE pathogens, is also an emerging threat in the landscape of antimicrobial resistance⁷⁸. These organisms share multiple features that make them particularly problematic, including a natural reservoir in the gastrointestinal tract, a diverse accessory genome augmented by conjugative plasmids and their frequent involvement in urinary tract infections (UTIs), intra-abdominal infections, pneumonia and bacteraemia⁷⁹. The emergence of extended-spectrum β-lactamases (ESBLs; for example, CTX-M) and carbapenemases (for example, KPC and OXA-48-like serine carbapenemases; NDM, VIM and IMP metallo-β-lactamases) among several epidemic lineages across the globe has led to increasing difficulty in treating infections due to these organisms⁸⁰.

Molecular epidemiology of Klebsiella pneumoniae and Enterobacter spp.

Significant genomic diversity in bacteria once identified as K. pneumoniae or Enterobacter spp. by biochemical testing has led to the taxonomic reorganization of these organisms^81–83. K. pneumoniae sensu stricto comprises approximately 85% of clinical isolates, whereas other closely related species (Klebsiella quasipneumoniae, Klebsiella variicola, Klebsiella quasivaricola and Klebsiella africana) contribute variably to human disease^81,84. The defining features of the K. pneumoniae population structure are the emergence of internationally disseminated clones that harbour either multiple antimicrobial resistance determinants (including carbapenemases) or a hypervirulent phenotype.

The most prevalent lineage of carbapenem-resistant Klebsiella pneumoniae (CR-K. pneumoniae) is clonal group (CG) 258 (ST11/ST258/ST512), which makes up 71% of CR-K. pneumoniae in the United States⁸⁵. This CG likely emerged in the United States in the mid-1990s, with approximately 80% and 20% of the genome originating from ST11-like and ST442-like strains, respectively^86,87. Although CG258 is also the most prevalent carbapenem-resistant clone in Europe, there is a greater diversity of STs and carbapenemases that cluster geographically⁸⁸. An emerging clone in the United States, South America, Nigeria and South Africa is ST307 (refs. 89–91). These isolates are frequently ESBL producers and recovered from the urinary tract of hospitalized patients, but have also been reported to harbour KPC, NDM and OXA-48-like carbapenemases^92,93. Carbapenem-resistant isolates from South Asia and East Asia exhibit a much more diverse population structure, likely reflecting acquisition of plasmids carrying carbapenemases by the dominant local lineages⁸⁴. In China, ST11 is the predominant CR-K. pneumoniae strain⁹⁴. A worrisome trend in this region is the emergence of conjugative plasmids in ST11 associated with the hypervirulent phenotype also harbouring carbapenemases such as KPC-2, as well as acquisition of new capsule types and exchange of lipopolysaccharide loci by recombination in CR strains^95,96. ST101 and ST147 are emerging clones worldwide associated with various carbapenemases (KPC, VIM, NDM, OXA-48-like) and both may carry additional virulence factors^97,98. These genetic events suggest a blurring of historical divisions between the multidrug-resistant and hypervirulent phenotypes, with the potential risk for rapid spread of a multidrug-resistant hypervirulent clone worldwide^99,100.

The Enterobacter cloacae complex (ECC) group of organisms has gained increasing recognition as nosocomial pathogens. The ECC can be differentiated into two clusters, with an older clade characterized by a higher degree of heterogeneity and a younger, more closely related clade associated with the health-care environment¹⁰¹. A study of bloodstream isolates collected in the United Kingdom from 2001 to 2011 showed a divergence in the two groups approximately 50–250 years ago, and limited geographic clustering suggesting isolates were widespread amongst community microbiota, as opposed to hospital-acquired¹⁰².

The carbapenem-resistant Enterobacter cloacae complex (CREC) clusters into a smaller group of high-risk isolates: Enterobacter xiangfengensis (ST114 and ST171), Enterobacter hormaechei subsp. steigerwaltii (ST90 and ST93) and E. cloacae cluster III (ST78)¹⁰³. Interestingly, multiple STs in a geographic region often share the same MGEs and carbapenem resistance determinants. VIM enzymes are most common in Europe, KPC in the Americas, NDM in South America and Southeast Asia, OXA-48-like in the Middle East and North Africa, and IMP in the Pacific¹⁰³. The multifaceted evolution of CREC in the United States has been the most extensively studied. The predominant ST171 was first isolated in the Northeast United States and found to have acquired bla_KPC-3 or bla_KPC-4 on IncFIA or IncHI2/IncHI2A plasmids, respectively, prior to clonal dissemination across the American Midwest and Mid-Atlantic regions^104,105. ST78, on the other hand, was a widespread ESBL-producing clone that independently acquired KPC-bearing plasmids on multiple occasions¹⁰⁴. A large multicentre study of US CREC isolates found that a majority were non-carbapenemase producers despite a resistant phenotype, potentially related to alterations of porin channels required for antibiotic entry plus gene copy number amplification of ESBLs^85,106. This heterogeneity and diversity of resistance mechanisms make CREC a particular challenge for hospital-based infection prevention.

Clinical impact of carbapenem-resistant Enterobacterales.

Carbapenem resistance has a significant impact on the treatment and outcomes of Enterobacterales infections. In the United States, infections due to CR-K. pneumoniae had stabilized prior to the COVID-19 pandemic at around 2% of all K. pneumoniae infections, whereas there has been a steady increase in infections due to CREC over the past decade, up to 2.5% of isolates^107,108. As of 2020, most of Northern and Western Europe has CR-K. pneumoniae infection rates of <1%. However, there are significantly higher rates in Southern and Eastern Europe, with rates of CR-K. pneumoniae exceeding 50% in Belarus, Georgia, Greece, Moldova, Russia and Ukraine⁴⁵. From 2013 to 2016, the incidence of CR-K. pneumoniae in Latin America ranged from 0.8 to 12.7% and in the Asia-Pacific region ranged from 0 to 9.3%¹⁰⁹. Risk factors for acquiring carbapenem-resistant Enterobacterales (CRE) infection include prior CRE colonization, exposure to broad-spectrum antimicrobials, a stay in an intensive care unit, mechanical ventilation, prolonged hospital length of stay and indwelling urinary catheters¹¹⁰.

Fortunately, the treatment landscape for CRE is changing due to the availability of newer β-lactam–β-lactamase inhibitor (BL–BLI) combinations (ceftazidime–avibactam, meropenem–vaborbactam, imipenem–cilastatin–relebactam) that target class A serine β-lactamases (including ESBLs and carbapenemases), AmpC enzymes and also, in the case of avibactam, some class D OXA-48-like enzymes; the likely introduction of broad-spectrum BL–BLI combinations that inhibit or have activity against strains producing common class B metallo-β-lactamases (cefepime–taniborbactam, avibactam–aztreonam and cefepime–zidebactam, among others); and the introduction of the siderophore cephalosporin cefiderocol. Retrospective clinical data suggest that the newer BL–BLI combinations offer a significant therapeutic advantage over polymyxin-based regimens, with lower 30-day in-hospital mortality (9% versus 32%, when comparing ceftazidime–avibactam and colistin for CRE infection)¹¹¹. Similarly, 30-day mortality rates for bloodstream infections due to CRE were reported as 38–45% largely before the introduction of these agents, but more recent clinical data show lower rates of mortality (~34%)^89,112–114. Infections due to CRE organisms are associated with an increased risk of in-hospital mortality and length of stay as compared with carbapenem-susceptible isolates. However, one such study found there was no significant difference in mortality after adjusting for receipt of an antibiotic for which the isolate was susceptible¹¹⁵. This finding suggests that clinical outcomes for CRE infections may improve with the use of novel therapeutics.

Significant challenges remain in drug availability and emerging mechanisms of resistance (Fig. 3). Resistance to ceftazidime–avibactam in K. pneumoniae arises with changes in KPC-2 and KPC-3 that alter the ability of avibactam to inhibit the enzyme; of note, these changes may also impact hydrolytic activity, with the altered enzyme behaving similar to an ESBL rather than a carbapenemase¹¹⁶. Resistance to meropenem–vaborbactam and imipenem–cilastatin–relebactam is linked to loss of the OmpK36 outer membrane porin and an increased β-lactamase gene copy number^117,118. Decreases in susceptibility to cefiderocol are associated with multiple changes, including alteration of the target PBP3, loss of cirA, which encodes the iron-uptake receptor that cefiderocol utilizes to enter the cell, and the presence of certain exogenous β-lactamases (NDM, SHV-ESBLs) or mutations in the ampC gene of Enterobacter spp.^119,120. Taniborbactam shows promise as an inhibitor of metallo-β-lactamases; however, certain escape variants including IMP enzymes, NDM-9, NDM-30, VIM-83 and SIM-1 may circumvent this new inhibitor¹²¹.

Fig. 3 |. Emerging mechanisms of resistance in Gram-negative pathogens.

The outer membrane of Gram-negative organisms serves as a permeability barrier that restricts the access of antibiotics to their targets in the periplasmic space and cytoplasm. Loss of outer membrane porins decreases drug uptake and contributes to resistance of various antibiotics, most notably β-lactams. Once in the periplasmic space, β-lactam antibiotics face both serine β-lactamases and metallo-β-lactamases capable of hydrolysing the β-lactam ring and inactivating the antibiotic, and alterations in the target penicillin-binding proteins (PBPs) that decrease affinity for these drugs. Additionally, mutations in β-lactamases can alter the specificity of the enzyme, allowing hydrolysis of a wider range of β-lactam substrates, or confer resistance to inhibition by β-lactamase inhibitors. Polymyxin resistance occurs through activation of cell envelope stress response systems, ultimately leading to modification of the lipopolysaccharide of the outer membrane. This can occur via expression of intrinsic genes (eptA or arn locus) or through acquisition of the plasmid-associated mcr gene (not shown). Aminoglycoside resistance occurs through alterations of the ribosome that alter the antibiotic binding site or through aminoglycoside modifying enzymes, which transfer acetyl, phosphate or nucleotide groups to the amide or hydroxyl groups of the antibiotic. In addition, both acquired and intrinsic efflux pumps can lead to resistance to multiple classes of antibiotics, including β-lactams, fluoroquinolones, tetracyclines and aminoglycosides.

Pseudomonas aeruginosa

P. aeruginosa is an opportunistic pathogen which possesses various intrinsic mechanisms of resistance, including barriers to drug permeability, a wide array of multidrug efflux pumps and a chromosomally encoded AmpC enzyme (Pseudomonas-derived cephalosporinase (PDC))¹²². In addition, P. aeruginosa expresses multiple virulence factors, including pyoverdine, pyocyanin, up to six different secretion systems and associated secreted determinants, alginate, multiple quorum sensing pathways and exopolysaccharides, among others¹²³. Although carbapenem resistance in the United States is typically mediated by loss of the outer membrane porin OprD and overexpression of the MexAB–OprM efflux pump, the resistance repertoire of clinical isolates of P. aeruginosa worldwide are increasingly augmented by acquired resistance determinants including ESBLs and carbapenemases (VIM, IMP, GES, NDM and KPC, among others)¹²⁴.

Molecular epidemiology of Pseudomonas aeruginosa.

The population structure of P. aeruginosa is roughly divided into two major phyletic branches (clades A and B), with a smaller number of isolates split among an additional one to three groups (collectively called clade C)^125,126. An evaluation of the genus-level organization of P. aeruginosa based on whole-genome sequencing suggested that clade C was the first to branch, with a more recent divergence of clades A and B¹²⁷. There is significant diversity among P. aeruginosa strains from both single hospitals or cities, as well as a substantial presence of hospital-adapted high-risk clones worldwide^128,129. Interestingly, two of the genes encoding the major effectors of the type 3 secretion system, ExoS and ExoU, are almost mutually exclusive in the P. aeruginosa chromosome and possess a strong association with the clade. Indeed, 98% of clade A isolates are positive for exoS whereas 95% of clade B isolates are positive for exoU, and strains with both genes or neither gene are rare^125,129. This distinction has practical clinical implications, as exoU strains are associated with septic shock and poor outcomes^130–133. Similar to enterococci, loss of a functional CRISPR–Cas system in P. aeruginosa has been associated with a larger accessory genome and occurs in clade B isolates nearly twice as frequently as in clade A¹³⁴.

Antimicrobial resistance — in particular, that mediated by transferrable resistance determinants — is a feature of several high-risk clonal groups, including ST235 (clade B), ST111 (clade A) and ST175 (clade A), among others¹³⁵. Available evidence suggests that these lineages emerged in the mid to late 1980s, after the approval of ceftazidime, imipenem and fluoroquinolones for clinical use. Genomic features of ST235 are consistent with a hypothesis of increased antibiotic pressure leading to clonal emergence, with initial branching events associated with fluoroquinolone resistance mutations, subsequent substitutions in the OprD outer membrane porin and genes regulating expression of PDC¹³⁶. In addition, ST235 isolates are characterized by the presence of various genes with putative roles in bacterial transformation that may have facilitated the acquisition of exogenous antimicrobial resistance determinants¹³⁶. Similarly, the epidemic ST111 strain likely arose through a recombination event with a clade C strain that resulted in the acquisition of fluoroquinolone resistance and a capsular lipopolysaccharide O-antigen switch¹³⁷. The authors hypothesized that the antigenic switch allowed evasion of the host immune response enabling ST111 to spread globally. Other high-risk clonal groups identified in clinical isolates include ST277 in Brazil, CG298 in the United States, CG308 in Europe, South America and Southeast Asia, ST309 in Mexico and Texas, CG357 in the Middle East and China, CG453 in China and CG823 in Asia and Australia^138–142. Of note, the type of carbapenemase acquired by P. aeruginosa seems to vary across specific geographic areas, with KPC-containing isolates most frequently detected in South America and VIM-2 producing isolates in Europe^138,143.

Clinical impact of carbapenem-resistant Pseudomonas aeruginosa.

Antibiotic resistance in P. aeruginosa is a major problem as the diversity of resistance mechanisms can make selecting a reliable empiric regimen difficult. A retrospective observational study of bloodstream infections from a large US-based hospital database showed increased mortality among P. aeruginosa isolates with a ‘difficult to treat’ resistance phenotype (DTR; resistance to fluoroquinolones, cephalosporins and carbapenems), with a 40% increase in the adjusted mortality rate among patients infected with isolates resistant to all classes¹⁴⁴. Similar outcomes were seen in a large prospective observational study in Italy, where carbapenem-resistant Pseudomonas aeruginosa (CR-PA) infections were associated with higher odds of 30-day mortality as compared with carbapenem-susceptible infections¹⁴⁵.

Ceftolozane–tazobactam retains activity against CR-PA as it is unaffected by OprD porin loss, is a poor substrate for pseudomonal efflux pumps and is stable to hydrolysis by the PDC enzyme¹⁴⁶. Resistance to ceftolozane is primarily associated with changes in PDC, particularly those altering the binding pocket near the active site, as well as certain GES and metallo-β-lactamases¹⁴⁷. Although not a carbapenemase, PDC displays non-negligible rates of imipenem hydrolysis, and overexpression of this enzyme coupled with OprD loss can lead to imipenem resistance¹⁴⁸. The diazabicyclooctane (DBO) β-lactamase inhibitor relebactam can restore imipenem activity against some CR-PA, and PDC changes leading to ceftolozane resistance seem to decrease activity against imipenem¹⁴⁹. Genomic analysis of P. aeruginosa strains that developed resistance to imipenem–relebactam suggested that upregulation of the MexAB–OprM and MexEF–OprN efflux pumps may work in concert to extrude relebactam from the periplasmic space¹⁵⁰. Cefiderocol, a siderophore cephalosporin which commandeers the bacterial iron import pathway, has high rates of in vitro susceptibility against CR-PA. Resistance to cefiderocol has been associated with mutations that inactivate the expression of the TonB-dependent receptors needed for drug import, mutations in the PDC enzyme as well as PER-type and SPM-1 β-lactamases^151,152.

A significant factor that affects the outcomes of CR-PA infections is the mechanism of carbapenem resistance. A large international observational cohort of patients with CR-PA infections found that the 30-day mortality of patients infected with carbapenemase-producing isolates was significantly higher than those with non-carbapenemase-producing isolates (22% versus 12%)¹³⁸. Patients with a carbapenemase-producing CR-PA were more likely to have bloodstream infection and be immunocompromised, potentially accounting for some differences in mortality, although production of a carbapenemase remained statistically associated with mortality on inverse probability weighted analysis. The presence of metallo-β-lactamases also compromises the activity of newer agents such as ceftolozane–tazobactam, ceftazidime–avibactam and imipenem–cilastatin–relebactam¹⁵³.

Acinetobacter baumannii

The A. baumannii–calcoaceticus complex consists of four species that are difficult to differentiate on a biochemical basis, three of which cause clinical disease (A. baumannii, Acinetobacter pittii and Acinetobacter nosocomialis) and one (Acinetobacter calcoaceticus) that has a limited pathogenic role^154,155. The ability of A. baumannii to persist on environmental surfaces and resist killing by disinfectants favours bacterial survival in the health-care setting¹⁵⁶. Similar to other ESKAPE pathogens, an array of resistance determinants including chromosomal Ade efflux pumps, intrinsic ADC cephalosporinase, the OXA-51 β-lactamase and acquired OXA-type carbapenemases have made A. baumannii a formidable clinical challenge¹⁵⁷.

Molecular epidemiology of carbapenem-resistant Acinetobacter baumannii.

Molecular methods have been integral to understanding the population structure of A. baumannii. Two major multilocus sequence typing systems exist, one developed at the Pasteur Institute and the other known as the Oxford scheme (ST^Ox)^158,159. The Pasteur scheme (ST^Pas) seems to be more consistent for classification of A. baumannii into clonal groups, whereas the ST^Ox is more prone to artefacts due to recombination and the presence of two copies of the gdhB locus, although this scheme allows better discrimination between closely related isolates¹⁶⁰. Most clinical isolates can be found in one of three global lineages, CC1 (made of ST1^Pas and related isolates), CC2 (ST2^Pas) and CC3 (ST3^Pas). Currently, CC2 is the dominant carbapenem-resistant Acinetobacter baumannii (CRAB) lineage, making up 77% of US isolates and 59% of genomes worldwide^161,162. The CC2 lineage can be further subdivided into several sub-lineages, with ST208^Ox being widely distributed in the United States, although a recent multicentre study suggested an ongoing process of clonal replacement with ST281^Ox and the presence of a non-CC2 lineage ST499^Pas (refs. 161,163). A similar structure is seen in China, with most strains belonging to CC2 and individual hospitals containing a dominant circulating clone¹⁶⁴. One complicating factor in describing the global population structure of A. baumannii is that a relatively limited number of countries represent a large proportion of the sequenced genomes. Small studies from South America, Japan and nations in Africa report STs not associated with the larger CC2 lineage for the majority of isolates. However, it is uncertain whether these isolates represent the population at a single hospital or broader regional trends^165–168.

The emergence of CRAB was reconstructed from a collection of CC1 A. baumannii isolates spanning nearly three decades¹⁶⁹. The ancestors of this group divided into two lineages in the 1960s, with the dominant antibiotic-resistant lineage acquiring the AbaR0 or AbaR3-like genomic islands (carrying sulfonamide, aminoglycoside and β-lactam resistance genes) in the 1970s¹⁷⁰. This resistance cassette likely originated from an IncM plasmid R1215, identified from a contemporaneous isolate of Serratia marcescens¹⁷¹. Emergence of fluoroquinolone resistance mutations, and ISAba1-mediated hyperexpression of the intrinsic gene coding for the ADC enzyme, roughly mirrored the introduction and clinical use of fluoroquinolone and expanded-generation cephalosporins, respectively¹⁷². Entering the new millennium, Tn-mediated acquisition of OXA carbapenemases (such as OXA-23, OXA-24, OXA-58 and OXA-72) led to the widespread dissemination of CRAB¹⁷³. In addition to OXA enzymes, various other ESBLs (GES, PER, VEB) and carbapenemases (KPC, IMP, NDM and VIM) have been described in A. baumannii clinical isolates from each of the clonal groups^174,175.

Clinical impact of CRAB.

There are several aspects of CRAB infections that complicate treatment. Patients often have long-term exposure to the health-care environment, a significant risk for exposure to antibiotics and colonization with resistant isolates¹⁷⁶. Surveillance data from the United States show that the prevalence of carbapenem resistance in line-associated bloodstream infections is 35% in general hospital wards, 47% in intensive care units and nearly 77% in long-term care settings¹⁷⁷. Global rates of carbapenem resistance vary substantially; in 2021, rates ranged from 0–5% in Western Europe and Scandinavia to greater than 80–90% in some countries of Southern and Eastern Europe, Latin America, Africa and the Middle East^24,45. The high frequency of resistance coupled with the fact that CRAB tends to infect patients who are critically ill or with multiple comorbidities leads to poor outcomes¹⁷⁸.

Polymyxin-based regimens, both monotherapy and combination therapy, have been used for CRAB infections given the relatively high rates of susceptibility. Outcomes, however, have been generally poor and, despite in vitro synergy, randomized clinical data for combinations of colistin with meropenem, levofloxacin or rifampin have not shown a significant mortality benefit¹⁷⁹. Ampicillin–sulbactam (for which sulbactam is the active component against A. baumannii PBP3) is the current preferred backbone of therapy, with evidence of improved outcomes when used at a high dose (≥9 g per day of sulbactam) in combination with another active agent in mostly observational studies¹⁸⁰. Minocycline, tigecycline and eravacycline may be candidates for combination therapy. However, optimized dosing should be used, and tigecycline and eravacycline have no established break points.

The OXA carbapenemases typically found in CRAB are poorly inhibited by the newer β-lactamase inhibitors avibactam, relebactam and vaborbactam. Cefiderocol retains high levels of in vitro activity against CRAB¹⁸¹. However, emergence of resistance is a challenge, with changes in the TonB iron import pathways, the PBP3 target and the presence of PER and NDM β-lactamases associated with elevated minimum inhibitory concentrations^182,183. Randomized clinical data comparing cefiderocol with the best available therapy (mostly colistin combinations) for treatment of CRAB infections showed no difference in outcomes and a numerically higher mortality in the cefiderocol arm¹⁸⁴. Thus, additional clinical data are needed to determine the optimal use of cefiderocol in these infections. Sulbactam–durlobactam was recently approved by the US Food and Drug Administration (FDA) for the treatment of CRAB infections. Durlobactam is a potent inhibitor of the OXA enzymes and serves to protect sulbactam from hydrolysis¹⁸⁵. In the phase III registrational trial, sulbactam–durlobactam was non-inferior to colistin (both given in conjunction with imipenem–cilastatin) with an absolute all-cause mortality difference of 13.2% at 28 days¹⁸⁶. Although a promising development for the treatment of CRAB infections, resistance may still arise through alterations of PBP3 or the presence of metallo-β-lactamases^185,187.

Novel treatments and emerging approaches targeting ESKAPE pathogens

Various compounds are currently under clinical development for treatment of infections caused by ESKAPE organisms (Table 2 and Fig. 4). Numerous BL–BLIs are under development, primarily based on one of two mentioned scaffolds: DBOs, of which avibactam (which inhibits class A, class C and some OXA-48-like carbapenemases) was the first to enter clinical use; and boronic acid inhibitors, which act via competitive inhibition by forming a transition state intermediate in the enzyme active site¹⁸⁸. Aztreonam–avibactam targets Gram-negative infections, particularly those caused by organisms producing metallo-β-lactamases. Aztreonam is stable to hydrolysis by metallo-β-lactamases, selectively targets PBP3 of Enterobacterales and, in combination with avibactam, protects aztreonam from other serine β-lactamases¹⁸⁹. The combination was studied in the REVISIT trial, showing non-inferiority to meropenem¹⁹⁰. Although avibactam also possesses weak activity against PBP2 from some Enterobacterales species, newer DBOs such as zidebactam include expanded activity against PBP2 from P. aeruginosa and A. baumannii¹⁹¹. One of the promising aspects of boronic acid inhibitors (taniborbactam and xeruborbactam) is the broad spectrum, including inhibition of several metallo-β-lactamases^192,193.

Table 2 |.

New drugs in development for ESKAPE pathogens

Drug	Mechanism of action	Target organisms	Development stage	Notes	Trial identifier
Traditional antibiotics
β-Lactam or β-lactamase inhibitors
Cefepime + enmetazobactam	PBP inhibition; β-lactamase inhibition (Ambler class A ESBLs)	ESBL-producing Enterobacterales	Phase III	Enmetazobactam is a methylated derivative of tazobactam that forms a zwitterion and improves penetration to periplasmic space; carbapenem-sparing therapy for complicated UTI	NCT03687255 (ref. 214)
Cefepime + taniborbactam	PBP inhibition; β-lactamase inhibition (Ambler class A, B (except IMP), C and D (OXA-48-like only) enzymes)	CRE and CR-PA	Phase III	Broad-range boronic acid-based β-lactamase inhibitor; covalent inactivation of serine β-lactamases, transition state inhibitor of VIM and NDM metallo-β-lactamases	NCT03840148 (ref. 215)
Cefepime + zidebactam	PBP inhibition; β-lactamase inhibition (Ambler class A and C enzymes)	CRE, CR-PA, CRAB	Phase III	Zidebactam is a bicyclo-acyl hydrazide DBO derivative with expanded inhibition of PBP2 in Pseudomonas aeruginosa and Acinetobacter baumannii	NCT04979806 (ref. 216)
Ceftibuten + ARX-1796	PBP inhibition; β-lactamase inhibition (Ambler class A, C and D (OXA-48-like only) enzymes)	CRE, ESBL-producing Enterobacterales	Phase I	ARX-1796 is an orally bioavailable prodrug of the DBO β-lactamase inhibitor avibactam	NCT03931876 (ref. 217)
Ceftibuten + ledaborbactam (VNRX-7145)	PBP inhibition; β-lactamase inhibition (Ambler class A, C and D (OXA-48-like only) enzymes)	CRE, ESBL-producing Enterobacterales	Phase I	Ledaborbactam etzadroxil is the orally bioavailable prodrug of the boronic acid β-lactamase inhibitor ledaborbactam	NCT04243863 (ref. 218)
Cefpodoxime proxetil + ETX0282	PBP inhibition; β-lactamase inhibition (Ambler class A, C and D (OXA-48-like only) enzymes)	CRE, ESBL-producing Enterobacterales	Phase I	ETX0282 is an orally bioavailable prodrug of a novel DBO β-lactamase inhibitor, which also has intrinsic activity against Escherichia coli PBP2	NCT03491748 (ref. 219)
Imipenem–cilastatin + funobactam (XNW4107)	PBP inhibition; β-lactamase inhibition (Ambler class A and C enzymes, some inhibition of class D (OXA-24/40))	CRE, CRAB	Phase I	Funobactam is a DBO β-lactamase inhibitor without antibacterial activity	NCT05204368 (ref. 220)
Meropenem + nacubactam	PBP inhibition; β-lactamase inhibition (Ambler class A and C enzymes)	CRE	Phase I	Nacubactam is a DBO β-lactamase inhibitor, which also has intrinsic activity against PBP2 of Enterobacterales	NCT03182504 (ref. 221)
QPX2014 + xeruborbactam (QPX7728)	PBP inhibition; β-lactamase inhibition (Ambler class A, B, C and D (including OXA-23) enzymes)	CRE, CR-PA, CRAB	Phase I	QPX2014 is an undisclosed β-lactam; xeruborbactam (QPX7728) is a broad-range boronic acid β-lactamase inhibitor with potential for intravenous and oral delivery	NCT04380207 (ref. 222) NCT05072444 (ref. 223)
Protein synthesis inhibitors
Apramycin (EBL-1003)	Binds 16S rRNA and leads to codon misreading and inhibition of tRNA translocation	CRAB, CRE	Phase I	Apramycin is not impacted by most traditional aminoglycoside resistance determinants, including ribosomal methyltransferases; resistance via N-acetyltransferase activity of AAC(3)-IV is rare	NCT05590728 (ref. 224)
Contezolid	Binds 50S ribosomal subunit and blocks peptidyltransferase site	MRSA, VRE	Approved (China, 2021) Phase III	Undergoing development in the United States for ABSSSI; oxazolidinone designed to have less myelosuppression and mono-amine oxidase inhibition than linezolid	NCT03747497 (ref. 225) NCT05369052 (ref. 226)
KBP-7072	Binds 30S ribosomal subunit, blocking aminoacyl-tRNA binding site	CRE, CRAB, MRSA, VRE	Phase I	Aminomethylcycline not impaired by traditional tetracycline efflux and ribosomal protection determinants; intravenous and oral formulations, with planned trials for ABSSSI, CAP and complicated intra-abdominal infection	NCT04532957 (ref. 227) NCT05507463 (ref. 228)
Nucleic acid synthesis inhibitors
BV100	Inhibits DNA-dependent RNA polymerase	CRAB	Phase II	Intravenous formulation of rifabutin; in A. baumannii, active uptake of rifabutin occurs via the TonB-dependent receptor FhuE	NCT05685615 (ref. 229)
Gepotidacin	Binds to DNA gyrase and topoisomerase IV and leads to high rate of single-strand DNA breaks	ESBL-producing E. coli, MRSA	Phase III	Triazaacenaphthylene topoisomerase II inhibitor with unique mode of inhibition compared with fluoroquinolones; completed phase II trials for ABSSSI, and phase III trials for uncomplicated UTI and urogenital gonorrhoea	NCT04020341 (ref. 230) NCT05630833 (ref. 231)
Membrane-active antimicrobials
MRX-8	Binds lipopolysaccharide and disrupts the Gram-negative cell envelope	CRE, CR-PA, CRAB	Phase I	Polymyxin derivative designed to have less toxicity as compared with polymyxin B	NCT04649541 (ref. 232)
PLG0206 (WLBU2)	Cationic antimicrobial peptide that disrupts bacterial cell membrane	MRSA	Phase I	Synthetic antimicrobial peptide with anti-biofilm activity; undergoing investigation as adjunct to standard of care in periprosthetic joint infections	NCT05137314 (ref. 233)
QPX9003	Binds lipopolysaccharide and disrupts the Gram-negative cell envelope	CRE, CR-PA, CRAB	Phase I	Polymyxin derivative with improved potency, less surfactant binding and lower nephrotoxicity in a primate model	NCT04808414 (ref. 234)
SPR206	Binds lipopolysaccharide and disrupts the Gram-negative cell envelope	CRE, CR-PA, CRAB	Phase I	Polymyxin derivative with lower minimum inhibitory concentrations and designed to have decreased toxicity as compared with polymyxin B	NCT04865393 (ref. 235)
Fatty acid synthesis inhibitors
Afabicin	Inhibits the bacterial enoyl-acyl carrier protein reductase and disrupts fatty acid synthesis	MRSA	Phase II	Narrow-spectrum FabI inhibitor active against Staphylococcus aureus; under clinical development for ABSSSI and bone and joint infections	NCT03723551 (ref. 236)
Cell division inhibitors
TXA709	Alters FtsZ polymerization and disrupts cell septum synthesis	MRSA	Phase I	Difluorobenzamide prodrug that is orally bioavailable and retains activity against vancomycin, daptomycin and oxazolidonone resistant isolates due to its unique mechanism of action	None
Non-traditional antibiotics and anti-infective compounds
Phage and phage-derived therapies
Exebacase	Recombinant staphylococcal phage endolysin	MRSA	Phase III	Phase III study was halted in interim analysis for futility; pending evaluation for periprosthetic joint infection	NCT04160468 (ref. 237)
AP-PA02	Bacteriophage active against P. aeruginosa	CR-PA	Phase II	Completed phase II study in chronic P. aeruginosa infection in patients with cystic fibrosis	NCT04596319 (ref. 238)
YPT-01	Bacteriophage active against P. aeruginosa	CR-PA	Phase II	Phase II study in chronic P. aeruginosa infection in patients with cystic fibrosis	NCT04684641 (ref. 239)
BX004-A	Bacteriophage active against P. aeruginosa	CR-PA	Phase II	Completed phase II study in chronic P. aeruginosa infection in patients with cystic fibrosis	NCT05010577 (ref. 240)
Antivirulence
CAL02	Liposomal antitoxin agent	CRE, CRAB, CR-PA, MRSA	Phase II	Liposomes serve as a sink for bacterial toxins; improves survival in animal models of sepsis when given with traditional antimicrobials	NCT05776004 (ref. 241)
Cefepime + fluorothyazinon	PBP inhibition; inhibition of the bacterial type III secretion system	CR-PA	Phase II	Fluorothyazinon inhibits the type III secretion system in Gram-negative bacteria, including P. aeruginosa, and is associated with increased bacterial clearance in animal models of infection	NCT03638830 (ref. 242)
Suvratoxumab (MEDI4893)	Human IgG1 monoclonal antibody against the pore-forming staphylococcal α-toxin	MRSA	Phase III	Effective in decreasing mortality in mouse models of pneumonia and sepsis; however, a phase II trial of patients colonized with S. aureus in intensive care units did not show significant difference from placebo on progression to pneumonia	NCT05331885 (ref. 243)
Tosatoxumab (AR-301)	Human IgG1 monoclonal antibody against the pore-forming staphylococcal α-toxin	MRSA	Phase III	Phase II study showed shorter duration of mechanical ventilation in the AR-301 group but no differences in confirmed microbiologic eradication	NCT03816956 (ref. 244)
TRL1068	Monoclonal antibody against a conserved region of DNABII proteins which disrupts biofilm formation	MRSA	Phase I	Currently undergoing phase I study as an adjuvant in periprosthetic joint infections	NCT04763759 (ref. 245)
ALS-4	Inhibits α-acetolactate synthase and impairs staphyloxanthin production	MRSA	Phase I	Staphyloxanthin plays a role in protecting S. aureus from oxidative stress and killing by neutrophils and antimicrobials	NCT05274802 (ref. 246)
GSK3882347	A mannoside that inhibits the E. coli adhesin FimH	E. coli	Phase I	Inhibits uroepithelial binding of E. coli, preventing recurrent UTIs	NCT05138822 (ref. 247)
Reltecimod	CD28 T cell receptor mimetic	Various	Phase III	Inhibition of T cell stimulation by superantigens of Gram-positive organisms and endotoxin of Gram-negative organisms	NCT02469857 (ref. 248)

ABSSSI, acute bacterial skin and skin structure infection; CAP, community-acquired pneumonia; CRAB, carbapenem-resistant Acinetobacter baumannii; CRE, carbapenem-resistant Enterobacterales; CR-PA, carbapenem-resistant Pseudomonas aeruginosa; DBO, diazabicyclooctane; ESBL, extended-spectrum β-lactamase; ESKAPE pathogens, Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp. MRSA, methicillin-resistant Staphylococcus aureus; PBP, penicillin-binding protein; UTI, urinary tract infection; VRE, vancomycin-resistant enterococci.

Fig. 4 |. Novel treatment approaches for ESKAPE pathogens.

Therapeutic approaches for ESKAPE bacteria (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa and Enterobacter spp.) include traditional antibiotics, novel small molecules and non-traditional approaches to augmenting infectivity and virulence. Combination therapy aims to utilize a mechanistic understanding of bacterial defence against antibiotics, such as the see-saw effect between daptomycin and β-lactam antibiotics in Staphylococcus aureus and enterococci, to kill multidrug-resistant pathogens or delay the emergence of resistance. New small molecules, including β-lactam–β-lactamase inhibitor (BL-BLI) combinations, compounds targeting bacterial fatty acid synthesis or the FtsZ protein critical for cell septum synthesis, are currently in clinical development. Bacteriophages, used alone or in combination with traditional antibiotics, offer the promise of a therapy that can evolve alongside resistant microorganisms. Therapies targeting bacterial virulence factors aim to mitigate the ability of bacteria to establish infection or cause severe disease. Immunotherapy such as passively administered monoclonal antibodies and vaccines can leverage the host immune system to reduce colonization, decrease virulence or alter the bacterial response to traditional antimicrobial compounds.

Additional existing classes of antibiotics are being modified to improve activity against common resistance mechanisms and decrease toxicity, including the aminoglycoside apramycin and several polymyxin B derivatives. Gepotidacin is an oral triazaacenaphthylene antibiotic that targets topoisomerase IV and the B subunit of DNA gyrase, but binds its target in a fashion different from fluoroquinolones¹⁹⁴. Gepotidacin is currently being studied for the treatment of urogenital gonorrhoea and for uncomplicated UTIs including those caused by multidrug-resistant Enterobacterales¹⁹⁵. The novel mechanism of inhibition indicates that gepotidacin will be active against fluroquinolone-resistant isolates.

Antibiotics belonging to classes with novel mechanisms of action are particularly attractive, as these compounds are likely to be active against drug-resistant bacteria. Such drugs also face hurdles, including unexpected toxicity or the rapid emergence of resistance. Murepavadin, an antimicrobial peptide with activity against CR-PA, was abandoned in phase III trials after concerns with renal toxicity. Several LpxC inhibitors targeting lipid A biosynthesis were withdrawn due to cardiotoxicity and trials of epetraborole, a leucyl-tRNA synthetase inhibitor, were stalled by concerns of resistance¹⁹⁶. TXA709, an inhibitor of the FtsZ protein critical to bacterial cell division, and afabicin, which blocks the FabI enzyme in the bacterial fatty acid synthesis pathway, are still in early-stage trials and may offer interesting activities against ESKAPE pathogens^197,198.

Several compounds are in development targeting specific virulence factors to block pathogenesis or decrease the virulence of ESKAPE organisms. Some examples that have entered clinical development include fluorothyazinone, an inhibitor of type III secretion systems that is being developed as a combination with cefepime against P. aeruginosa; GSK3882347, a mannoside that inhibits the E. coli adhesin FimH and prevents bacteria from binding to the urothelium of the bladder; and ALS-4, a staphyloxanthin inhibitor that alters the virulence of S. aureus, blocking infectivity^199–201. Reltecimod is a CD28 T cell receptor mimetic that inhibits T cell stimulation by an array of bacterial pathogens. In a randomized, double-blind, placebo-controlled trial, early administration of reltecimod in severe necrotizing soft tissue infection resulted in a significant improvement in the primary composite end point in the per-protocol population but not in the modified intention-to-treat population²⁰². Therefore, the role of this strategy in resistant infections is uncertain.

Phages, viruses that infect bacteria, have gained increased attention for their activity against ESKAPE pathogens and the potential for a treatment that can evolve alongside the target pathogen. Phages have been used since the early twentieth century, but the discovery of traditional antibiotics overshadowed phage development outside Eastern Europe²⁰³. Several phage products have entered clinical trials, primarily for the treatment of chronic infections due to P. aeruginosa in cystic fibrosis. Exebacase, a derivative of a staphylococcal phage lysin being developed as an adjunct to the standard of care for S. aureus bacteraemia, showed early promise in phase II studies, but the phase III trial was halted owing to futility at the interim analysis²⁰⁴. Challenges remain in the development of phages as an anti-infective therapy, including more rapid identification of phage susceptibility and time to purification, uncertainty in dosing, limited safety data, development of an anti-phage host immune response and regulatory challenges²⁰⁵.

Vaccines hold the promise of preventing infections altogether; however, despite successes with pathogens such as Streptococcus pneumoniae, there are limited vaccine candidates in the pipeline for ESKAPE pathogens²⁰⁶. Multiple vaccine candidates for S. aureus have entered clinical trials, although none to date has proved successful at decreasing rates of infection or nasal carriage²⁰⁷. A phase I/II study of an adjuvanted O polysaccharide antigen vaccine for four common K. pneumoniae serotypes recently completed enrolment. However, results have not been reported²⁰⁸. Vaccines for ESKAPE pathogens continue to face various challenges including identifying immunogenic targets in species that have adapted to colonizing and evading host immune responses, limited windows for administration and impaired immune responses in critically ill patients, and the low incidence of infection among colonized individuals in the general population, making determinations of efficacy problematic²⁰⁹. Monoclonal antibodies are also currently being developed for some ESKAPE pathogens. For example, tosatoxumab (AR-301) is a human monoclonal IgG1 antibody that targets S. aureus ɑ-toxin and is being studied for ventilator-associated pneumonia. Table 2 provides information of some other anti-ESKAPE pathogen monoclonal antibodies and indications being pursued.

Conclusions and perspectives

The bacterial species that make up the ESKAPE pathogens have adapted to the modern health-care environment around the globe. Taxonomic changes and the emergence of new threats, such as resistant isolates of E. coli or Stenotrophomonas maltophilia infections in immunocompromised hosts, suggest that the ESKAPE acronym will need to evolve as well. Recent estimates of the clinical impact of antimicrobial resistance suggest that in 2019 nearly 5 million deaths were associated with drug-resistant infections²¹⁰. Most importantly, this burden is not distributed equally, with a disproportionate impact on low-resource settings paralleling the massive health impacts of malaria, HIV, tuberculosis and the COVID-19 pandemic in the developing world.

Antimicrobial consumption for human use has steadily increased across the world, especially in low-income and middle-income countries, although these regions still lag high-income nations in total consumption²¹¹. This is dwarfed by antimicrobial consumption in animals, which accounts for 80% of antimicrobial use in the United States, and more than 63,000 tons globally, most for non-therapeutic use (growth promotion)²¹². Although a significant problem, it is not an insurmountable one. The use of avoparcin in Europe was associated with the rise of VRE in feed animals in the 1990s and potential spillovers to the human population. In 1995 the use of avoparcin as a growth promoter was banned, and recovery of VRE from animals dropped from near 80% to less than 10% within 5 years²¹³. Thus, mobilizing stakeholders across the ‘One Health’ continuum is critical to create innovative solutions to steward antimicrobial use.

Novel therapies are urgently needed, but so are the resources to bring these agents to the developing world where the burden of resistance is rising fastest. Infrastructure to track the spread of emerging high-risk genetic lineages and to deploy measures to prevent their spread in the hospital and community is rare outside developed nations, and infection prevention and control efforts remain critical aspects of the fight against resistance. Thus, it is imperative that strong international partnerships between high-income and low-income settings support the surveillance of ESKAPE pathogens, provide for effective infection control programmes and allow for the successful characterization, treatment and stewardship of novel antimicrobial agents. As we near a century of antimicrobial use, we must learn the lessons of the past to safeguard the therapies of tomorrow, with the understanding that fighting bacterial evolution is a Sisyphean task.